Popsixle

popsixle

Previously… https://awstip.com/reducing-eks-windows-node-5-min-start-time-to-90s-eks-windows-bootstrapper-4314f72367a0?embedable=true I threw in some tricks I used to make our Windows EKS nodes start in 90s rather than the 5 minutes a standard EKS Windows node takes. I introduced you to the git repo eks-windows-bootstrapper - the application replacing the AWS-provided configuration scripts for EKS Windows. https://github.com/atg-cloudops/eks-windows-bootstrapper?embedable=true In today’s story, I will walk you through how to make your Windows nodes take ~30s to start! Recap In the previous story, we got to the point where our start times looked like this: Let’s take a closer look at our latencies now: In order for us to get better launch times, we now have 2 areas where we can improve timing Windows Startup Configure Network/Start K8s Services Improving Windows Startup Performance As we are using the AWS Image Builder (https://aws.amazon.com/image-builder/) to build our custom AMIs as discussed in the previous story, any changes we make to our image need to be made here. Outside AWS Fast Launch - there really isn’t much we can do to help with the Windows boot process, if you are able - you can uninstall Windows Defender which will massively improve start performance. Side note - I found in our use case we do not actually need Fast Launch as we do not actually need to run the OOBE process. We can set our culture using Set-Culture, etc. inside one of our components, and as long as we remove the sections in Unattend.xml related to OOBE, Windows will skip much of the setup process that EC2 Fast Launch deals with. When you start the output AMI, it will start as quickly as a Fast Launch AMI, but without the added expense of pre-provisioning EBS snapshots. The downside to this approach is that you won’t be able to configure Windows settings on the first boot, what you bake into the AMI is what you will start with - So make sure you set culture/timezone information correctly somewhere in one of your Image Builder components. For reference: So with all avenues to make Windows do less stuff on boot exhausted, there was only one choice left… Changes were made to the EKS Windows Bootstrapper code so it could be run as a Windows Service. The idea behind this is that we could squeeze a few more seconds out if we could run our code as Windows is starting services on boot instead of waiting for the userdata script to fire. This worked - however, it did break the AWS SSM Agent as a result as it also starts early. In the installation script included in the repo, we work around this by setting the SSM agent startup type to Automatic (Delayed). Faster Configure Network/Start K8s Services This basically involved translating the remainder of the code in the standard AWS bootstrapping scripts in EKS-StartupTask.ps1 Setup Container Network
        ... 
           [DllImport("vmcompute.dll")]
            static extern void HNSCall([MarshalAs(UnmanagedType.LPWStr)] string method,
                           [MarshalAs(UnmanagedType.LPWStr)] string path,
                           [MarshalAs(UnmanagedType.LPWStr)] string request,
                           [MarshalAs(UnmanagedType.LPWStr)] out string response);
            HNSCall("POST", "/networks", jsonString, out response);
        ...

https://github.com/atg-cloudops/eks-windows-bootstrapper/blob/8f519953209e3caea590652832acc74585ab1a45/BootstrapperService.cs#L514 Add routes to created NIC
        ...
          routeAddCommands.Append($"route ADD {ipAddrs[i]} MASK 255.255.255.255 0.0.0.0 IF {vNICIndex}");
        ...
        Process process = new Process();
        process.StartInfo.FileName = "cmd.exe";
        process.StartInfo.Arguments = $"/C {routeAddCommands}";
        process.StartInfo.RedirectStandardOutput = true;
        process.StartInfo.UseShellExecute = false;
        process.StartInfo.CreateNoWindow = true;
        process.Start();
        await process.WaitForExitAsync();

https://github.com/atg-cloudops/eks-windows-bootstrapper/blob/8f519953209e3caea590652832acc74585ab1a45/BootstrapperService.cs#L586


Start Kube-proxy and Kubelet in parallel once all configuration files have been written (all the previously converted code) Installation To install the bootstrapper, add this to a component in Image Builder: - name: InstallEksWindowsBootstrapper
        action: ExecutePowerShell
        inputs:
          commands:
            - |
              Invoke-WebRequest -Uri 'https://github.com/atg-cloudops/eks-windows-bootstrapper/releases/download/v1.29.1/Install-Service.ps1' -OutFile 'Install-Service.ps1'; 
              .\Install-Service.ps1; 
              Remove-Item 'Install-Service.ps1'; Putting everything together, I tested the output image using these settings: EC2 Instance

m4.xlarge (4 Core)
EBS


GP3 - 120GB/600Mbps/3600 IOPS Testing on these settings got me start times around 55s! We are finally under a minute, maybe spot instances are now viable for Windows EKS Nodes now that the 5-minute penalty is now 1 minute. Hold On… You Said A 30s Start Time! You Said as Fast as Linux! A 30s start time is possible! It just costs a lot. 😢 I ended up testing how fast the sum of all the software was by eliminating as many hardware bottlenecks as I could. Switched from GP3 → IO2 30K IOPS Switched from m4.xlarge → c6a.4xlarge Require Nitro Hypervisor It’s quite easy to change in Karpenter; simply change the Ec2NodeClass to change storage options, and then add the relevant keys in NodePool, e.g. blockDeviceMappings:
    - deviceName: /dev/sda1
      ebs:
        iops: 30000
        volumeSize: 120Gi
        volumeType: io2 - key: karpenter.k8s.aws/instance-hypervisor
          operator: In
          values:
            - 'nitro'
        - key: karpenter.k8s.aws/instance-cpu
          operator: In
          values:
            - '8' The beast that was produced by Karpenter was visible in K8s in 32s. It would be sub-30s with antivirus disabled. Our final tuning produces nodes in around 45s (Using GP3). You can tune for cost/performance yourself for your own scenario. Looking for Help EKS Windows bootstrapper is a simple app that writes config and starts Kubernetes services. It is currently set up for EKS 1.29. As Kubernetes goes through 1.30, 1.31, 1.32… it will need to be kept up to date… I am looking for help with this. 🙂 Previously… https://awstip.com/reducing-eks-windows-node-5-min-start-time-to-90s-eks-windows-bootstrapper-4314f72367a0?embedable=true https://awstip.com/reducing-eks-windows-node-5-min-start-time-to-90s-eks-windows-bootstrapper-4314f72367a0?embedable=true I threw in some tricks I used to make our Windows EKS nodes start in 90s rather than the 5 minutes a standard EKS Windows node takes. I introduced you to the git repo eks-windows-bootstrapper - the application replacing the AWS-provided configuration scripts for EKS Windows. https://github.com/atg-cloudops/eks-windows-bootstrapper?embedable=true https://github.com/atg-cloudops/eks-windows-bootstrapper?embedable=true In today’s story, I will walk you through how to make your Windows nodes take ~30s to start! Recap In the previous story, we got to the point where our start times looked like this: Let’s take a closer look at our latencies now: In order for us to get better launch times, we now have 2 areas where we can improve timing Windows Startup Windows Startup Configure Network/Start K8s Services Configure Network/Start K8s Services Improving Windows Startup Performance As we are using the AWS Image Builder ( https://aws.amazon.com/image-builder/ ) to build our custom AMIs as discussed in the previous story, any changes we make to our image need to be made here. Outside AWS Fast Launch - there really isn’t much we can do to help with the Windows boot process, if you are able - you can uninstall Windows Defender which will massively improve start performance. https://aws.amazon.com/image-builder/ Side note - I found in our use case we do not actually need Fast Launch as we do not actually need to run the OOBE process. We can set our culture using Set-Culture , etc. inside one of our components, and as long as we remove the sections in Unattend.xml related to OOBE, Windows will skip much of the setup process that EC2 Fast Launch deals with. When you start the output AMI, it will start as quickly as a Fast Launch AMI, but without the added expense of pre-provisioning EBS snapshots. Set-Culture Unattend.xml The downside to this approach is that you won’t be able to configure Windows settings on the first boot, what you bake into the AMI is what you will start with - So make sure you set culture/timezone information correctly somewhere in one of your Image Builder components. For reference: So with all avenues to make Windows do less stuff on boot exhausted, there was only one choice left… Changes were made to the EKS Windows Bootstrapper code so it could be run as a Windows Service. The idea behind this is that we could squeeze a few more seconds out if we could run our code as Windows is starting services on boot instead of waiting for the userdata script to fire. This worked - however, it did break the AWS SSM Agent as a result as it also starts early. In the installation script included in the repo, we work around this by setting the SSM agent startup type to Automatic (Delayed) . userdata Automatic (Delayed) Faster Configure Network/Start K8s Services This basically involved translating the remainder of the code in the standard AWS bootstrapping scripts in EKS-StartupTask.ps1 EKS-StartupTask.ps1 Setup Container Network
        ... 
           [DllImport("vmcompute.dll")]
            static extern void HNSCall([MarshalAs(UnmanagedType.LPWStr)] string method,
                           [MarshalAs(UnmanagedType.LPWStr)] string path,
                           [MarshalAs(UnmanagedType.LPWStr)] string request,
                           [MarshalAs(UnmanagedType.LPWStr)] out string response);
            HNSCall("POST", "/networks", jsonString, out response);
        ...

https://github.com/atg-cloudops/eks-windows-bootstrapper/blob/8f519953209e3caea590652832acc74585ab1a45/BootstrapperService.cs#L514 Setup Container Network ... 
           [DllImport("vmcompute.dll")]
            static extern void HNSCall([MarshalAs(UnmanagedType.LPWStr)] string method,
                           [MarshalAs(UnmanagedType.LPWStr)] string path,
                           [MarshalAs(UnmanagedType.LPWStr)] string request,
                           [MarshalAs(UnmanagedType.LPWStr)] out string response);
            HNSCall("POST", "/networks", jsonString, out response);
        ... https://github.com/atg-cloudops/eks-windows-bootstrapper/blob/8f519953209e3caea590652832acc74585ab1a45/BootstrapperService.cs#L514 Setup Container Network ... 
           [DllImport("vmcompute.dll")]
            static extern void HNSCall([MarshalAs(UnmanagedType.LPWStr)] string method,
                           [MarshalAs(UnmanagedType.LPWStr)] string path,
                           [MarshalAs(UnmanagedType.LPWStr)] string request,
                           [MarshalAs(UnmanagedType.LPWStr)] out string response);
            HNSCall("POST", "/networks", jsonString, out response);
        ... ... 
           [DllImport("vmcompute.dll")]
            static extern void HNSCall([MarshalAs(UnmanagedType.LPWStr)] string method,
                           [MarshalAs(UnmanagedType.LPWStr)] string path,
                           [MarshalAs(UnmanagedType.LPWStr)] string request,
                           [MarshalAs(UnmanagedType.LPWStr)] out string response);
            HNSCall("POST", "/networks", jsonString, out response);
        ... https://github.com/atg-cloudops/eks-windows-bootstrapper/blob/8f519953209e3caea590652832acc74585ab1a45/BootstrapperService.cs#L514 https://github.com/atg-cloudops/eks-windows-bootstrapper/blob/8f519953209e3caea590652832acc74585ab1a45/BootstrapperService.cs#L514 Add routes to created NIC
        ...
          routeAddCommands.Append($"route ADD {ipAddrs[i]} MASK 255.255.255.255 0.0.0.0 IF {vNICIndex}");
        ...
        Process process = new Process();
        process.StartInfo.FileName = "cmd.exe";
        process.StartInfo.Arguments = $"/C {routeAddCommands}";
        process.StartInfo.RedirectStandardOutput = true;
        process.StartInfo.UseShellExecute = false;
        process.StartInfo.CreateNoWindow = true;
        process.Start();
        await process.WaitForExitAsync();

https://github.com/atg-cloudops/eks-windows-bootstrapper/blob/8f519953209e3caea590652832acc74585ab1a45/BootstrapperService.cs#L586 Start Kube-proxy and Kubelet in parallel once all configuration files have been written (all the previously converted code) Add routes to created NIC ...
          routeAddCommands.Append($"route ADD {ipAddrs[i]} MASK 255.255.255.255 0.0.0.0 IF {vNICIndex}");
        ...
        Process process = new Process();
        process.StartInfo.FileName = "cmd.exe";
        process.StartInfo.Arguments = $"/C {routeAddCommands}";
        process.StartInfo.RedirectStandardOutput = true;
        process.StartInfo.UseShellExecute = false;
        process.StartInfo.CreateNoWindow = true;
        process.Start();
        await process.WaitForExitAsync(); https://github.com/atg-cloudops/eks-windows-bootstrapper/blob/8f519953209e3caea590652832acc74585ab1a45/BootstrapperService.cs#L586 Add routes to created NIC ...
          routeAddCommands.Append($"route ADD {ipAddrs[i]} MASK 255.255.255.255 0.0.0.0 IF {vNICIndex}");
        ...
        Process process = new Process();
        process.StartInfo.FileName = "cmd.exe";
        process.StartInfo.Arguments = $"/C {routeAddCommands}";
        process.StartInfo.RedirectStandardOutput = true;
        process.StartInfo.UseShellExecute = false;
        process.StartInfo.CreateNoWindow = true;
        process.Start();
        await process.WaitForExitAsync(); ...
          routeAddCommands.Append($"route ADD {ipAddrs[i]} MASK 255.255.255.255 0.0.0.0 IF {vNICIndex}");
        ...
        Process process = new Process();
        process.StartInfo.FileName = "cmd.exe";
        process.StartInfo.Arguments = $"/C {routeAddCommands}";
        process.StartInfo.RedirectStandardOutput = true;
        process.StartInfo.UseShellExecute = false;
        process.StartInfo.CreateNoWindow = true;
        process.Start();
        await process.WaitForExitAsync(); https://github.com/atg-cloudops/eks-windows-bootstrapper/blob/8f519953209e3caea590652832acc74585ab1a45/BootstrapperService.cs#L586 https://github.com/atg-cloudops/eks-windows-bootstrapper/blob/8f519953209e3caea590652832acc74585ab1a45/BootstrapperService.cs#L586 Start Kube-proxy and Kubelet in parallel once all configuration files have been written (all the previously converted code) Start Kube-proxy and Kubelet in parallel once all configuration files have been written (all the previously converted code) Installation To install the bootstrapper, add this to a component in Image Builder: - name: InstallEksWindowsBootstrapper
        action: ExecutePowerShell
        inputs:
          commands:
            - |
              Invoke-WebRequest -Uri 'https://github.com/atg-cloudops/eks-windows-bootstrapper/releases/download/v1.29.1/Install-Service.ps1' -OutFile 'Install-Service.ps1'; 
              .\Install-Service.ps1; 
              Remove-Item 'Install-Service.ps1'; - name: InstallEksWindowsBootstrapper
        action: ExecutePowerShell
        inputs:
          commands:
            - |
              Invoke-WebRequest -Uri 'https://github.com/atg-cloudops/eks-windows-bootstrapper/releases/download/v1.29.1/Install-Service.ps1' -OutFile 'Install-Service.ps1'; 
              .\Install-Service.ps1; 
              Remove-Item 'Install-Service.ps1'; Putting everything together, I tested the output image using these settings: EC2 Instance

m4.xlarge (4 Core)
EBS


GP3 - 120GB/600Mbps/3600 IOPS EC2 Instance m4.xlarge (4 Core)
EBS


GP3 - 120GB/600Mbps/3600 IOPS m4.xlarge (4 Core) EBS


GP3 - 120GB/600Mbps/3600 IOPS m4.xlarge (4 Core) EBS GP3 - 120GB/600Mbps/3600 IOPS GP3 - 120GB/600Mbps/3600 IOPS GP3 - 120GB/600Mbps/3600 IOPS GP3 - 120GB/600Mbps/3600 IOPS Testing on these settings got me start times around 55s! We are finally under a minute, maybe spot instances are now viable for Windows EKS Nodes now that the 5-minute penalty is now 1 minute. Hold On… You Said A 30s Start Time! You Said as Fast as Linux! A 30s start time is possible! It just costs a lot. 😢 I ended up testing how fast the sum of all the software was by eliminating as many hardware bottlenecks as I could. Switched from GP3 → IO2 30K IOPS Switched from m4.xlarge → c6a.4xlarge Require Nitro Hypervisor It’s quite easy to change in Karpenter; simply change the Ec2NodeClass to change storage options, and then add the relevant keys in NodePool, e.g. blockDeviceMappings:
    - deviceName: /dev/sda1
      ebs:
        iops: 30000
        volumeSize: 120Gi
        volumeType: io2 blockDeviceMappings:
    - deviceName: /dev/sda1
      ebs:
        iops: 30000
        volumeSize: 120Gi
        volumeType: io2 - key: karpenter.k8s.aws/instance-hypervisor
          operator: In
          values:
            - 'nitro'
        - key: karpenter.k8s.aws/instance-cpu
          operator: In
          values:
            - '8' - key: karpenter.k8s.aws/instance-hypervisor
          operator: In
          values:
            - 'nitro'
        - key: karpenter.k8s.aws/instance-cpu
          operator: In
          values:
            - '8' The beast that was produced by Karpenter was visible in K8s in 32s. It would be sub-30s with antivirus disabled. Our final tuning produces nodes in around 45s (Using GP3). You can tune for cost/performance yourself for your own scenario. Looking for Help EKS Windows bootstrapper is a simple app that writes config and starts Kubernetes services. It is currently set up for EKS 1.29. As Kubernetes goes through 1.30, 1.31, 1.32… it will need to be kept up to date… I am looking for help with this. 🙂

EKS Windows Node 30s Boot Now Possible - Start as Fast as Linux! (eks-windows-bootstrapper - Part 2)

Happy Developers. It is Possible.

As regular readers know, I recently . After all the interviews, the next part of that process was the offer  negotiation phase.  To be incredibly transparent, I hate that part of  the interview process like almost nothing else in my life. It’s gut-churning and mind-numbing and terror-inducing all at the same time. I  always feel like I’m doing it wrong, and at the end of the process, I’m  certain I’ve made horrible mistakes that will haunt me for the rest of  my career. changed companies In the 35+ years I’ve worked in tech, I’ve changed jobs several times, each time interviewing with several companies before making a  career move. After some back-of-the-napkin math, I realized I’ve  received dozens of offers over the course of my career. Anyone who’s worked in tech understands this is pretty typical, and not just in this current period of marketplace unrest, with the “great resignation/reshuffling/whatever”; weekly announcements of layoffs and companies getting acquired, spun off, or shut down like it was a game of  whack-a-mole where the moles were actually Russian nesting dolls. Moving around like that means I’ve negotiated my salary, benefits, days off, and so on a bunch of times. And you’d think that after so much  practice over so many years, it would feel like old hat. No big deal. A  walk in the park. Nothing could be further from the truth. There are, of course, plenty of resources to learn how to be a better negotiator and to take control (not to mention ownership) of one’s  career. I’m a big fan of Chris Voss’ book .” I’m an equally big fan of , who I’ve enjoyed hearing from on | | | | on Corey Quinn’s . “Never Split the Difference Josh Doody three separate appearances Screaming In the Cloud podcast But even with the information I gleaned from those (and other) resources, I’m still uncomfortable with the entire process. Negotiating  the purchase of a car or house is so much easier emotionally, because  objects aren’t tied to yourself worth the way a job offer objectifies  your value as a professional. This makes it more important than ever to take a principle we hold true in our data centers and apply it to our careers: Namely, if something isn’t a core competency for the business, don’t spend huge amounts of time, effort, or budget on building it internally. And it’s an  indisputable fact that – while salary negotiation is important to my  career – it’s utterly irrelevant to my day-to-day job. This is why, when my friend and former colleague Kymberlee Price (aka “KymPossible” on Bluesky, hence the title of this blog) offered to  help with the negotiation phase during my last job search, I jumped at  the chance. If you are currently in the negotiation phase of your own job, stop reading this and . If you are hunting for your next job and know that negotiations are on the horizon, write down her info and after you finish reading this. And if you aren’t looking for a new job  at this moment, I’d remind you that – in tech at least – your transition  to turning on that green “Open to Work” banner in LinkedIn can happen  at any time. So, grab and put it somewhere for safekeeping. contact her contact her her info If you’re thinking, “I’m just not comfortable with being super aggressive to my future employer”, you need to move over and make some room because I’m not either. In fact, almost every one of Kym’s clients says the same thing. And the truth is, nobody is saying you need to be.  In fact, aggressive is the very LAST thing that you need for effective  negotiations. Firm? Absolutely. Confident? 100%. But aggressive? Never. Like any great tech practitioner, helping people negotiate their salary is Kym’s SIDE HUSTLE. By day, she’s a kick-ass product security leader. Kym relies on data and analysis to identify how the offer on the  table stands up to industry standards, your current compensation, the  likely movement of the market in the next few years, and more. Kym’s  process takes the emotion completely out of the conversation. Her view  of “total compensation” goes way beyond what I’m used to considering.  Not only did she factor in the usual elements like stock options, bonuses, etc., but she also included out-of-pocket healthcare costs, 401k matching, PTO, and a projection of how the complete offer will “age” over five years, based on average raises, bonuses, etc. The approach is hard to argue against. It’s not a fierce contest of wills where one side insists they are “worth more,” and the other does  their best to convince them otherwise. It’s a frank and fact-based  discussion of how the company’s offer compares competitively to other  options available and proposes adjustments to close any gaps so you can  happily accept.  A company can still choose to say, “That’s not  something we can do.” However, this is a far cry from the standard pushback  of “We don’t believe we (or anyone else) should do this.” Which, I gotta  tell you, is a breath of fresh air. Now, all of this implies that you’re always going to get a lowball offer – which isn’t the case.  Kym tells me that when clients get a phenomenal offer straight out of  the gate, she provides them with a few key questions to ask the recruiter to  confirm it really is a great offer and ensure there aren’t any surprises  later. Sometimes, the base compensation is great, but you’re losing  $50-100k a year in benefits, retirement matching, etc. Maybe you’re okay  with that tradeoff, but you need to know about it upfront. As a Tech professional, I need to build and maintain my skills as a developer, sysadmin, and engineer. I need to stay up to date on the  latest tools, techniques, and technologies. But I don’t need to spend precious time learning the fine art of salary negotiation, even when I  will probably have to do it several more times before I retire. Not when there are amazing folks like Kymberlee Price around to do it and do it far better than I could ever imagine. Also published . here

Figuring Out Job Offer Negotiations: Lessons Learned from Kym Possible 

Martyna Lewinska, Co-Founder & CTO at Fiat Republic Being an effective tech leader in a remote-first world doesn't mean reinventing your leadership style. As the CTO of Fiat Republic, I have met my team members on several occasions - something that’s incredibly important and irreplaceable if you want cohesion in your company - but, by and large, my leadership in my current role has always been facilitated remotely. From my journey, I've learned what changes are necessary and what aspects of leadership should remain consistent to excel in a remote-first environment. Communication Is Key I work in the domain of fiat-crypto payment solutions, which can be extremely complex at times. Clear communication and organization are paramount. At Fiat Republic, we’ve honed our communication channels to make them as obvious and accessible as possible. In practical terms, this means clearly identifying - and almost labeling - what we communicate via each channel. We have also increased the use of video calls and not just audio calls, which is critical in maintaining the connectivity essential in a tech-driven remote environment. Empowering Distributed Tech Teams Our engineering teams, spread across different locations, rely heavily on remote collaboration tools. The key to success in this tech-centric environment is a strong culture of information sharing and organization. As tech leaders, we are tasked with not only equipping our teams with the right tools but also guiding them on their effective use. We allocate the most effective tools for the job, and this is a non-negotiable across the business. Be Flexible There are so many KPIs and metrics you can attempt to measure as a tech lead, but I’ve found it most useful to focus on outcomes. This has been instrumental in maintaining productivity while fostering innovation. Our team members may have different working styles - from those who code intensively for hours to those who ponder before writing a single impactful line of code - but, as long as we all work towards the same goals, it pays to be flexible. Build Trust The unique challenges of managing a remote tech team include building trust and relationships. While we do meet in person periodically, our daily operations are online, demanding a balance between oversight and autonomy. This is particularly challenging and critical in a field driven by constant tech innovation. Personally, I find that clear communication and expected outcomes form the pillar of trust in my team. On top of that, making sure everyone knows what's expected of them, helping each other hit our goals, and holding ourselves accountable are all incredibly important. Cultivating a Learning Environment Each day in the field of fiat-crypto transactions is different, and staying ahead of industry trends is critical. I ensure our team has substantial opportunities for learning and innovation, promoting a culture where sharing tech expertise and exploring new ideas is balanced with timely project delivery. Be Consistent For tech leaders, adapting to a remote environment doesn’t mean altering your core leadership style. My approach, developed during my office days, remains grounded in trust, clear communication, and understanding each team member. This approach is vital, whether leading a team in person or remotely. Everything else - i.e., the tools - can change in a remote-first environment but, if you want to be a great tech leader, your philosophy needs to remain the same. As tech leaders, especially those transitioning to a remote setting, it’s not about changing your leadership essence but about adapting your tools and methods for a digital world. For emerging tech leaders, it’s crucial to define what exceptional leadership means in the context of technology and align your team with this vision. This philosophy has been pivotal in my success as a CTO and in guiding my journey as a remote tech leader.

Why It’s Still Possible to Be a Great Tech Leader in a Remote-First World

Daily Coding Problem Hello everyone and welcome back with another problem to solve! Today we’ll be dealing with paths, calculating the greatest path in a triangle of numbers starting from the top to the bottom! Today’s problem comes from the archives of : a super cool website to look at if you are in the mood for some head scratching! There are more than 800 math and coding problems to solve and a vast archive of discussions about them. It is the perfect tool to train your math skills and learn something new. Go check it out and try to solve your challenges too! ProjectEuler.net Let’s jump to the action! The problem By starting at the top of the triangle below and moving to adjacent numbers on the row below, the maximum total from top to bottom is 23. That is,3+7+4+9=23. Find the maximum total from top to bottom of the triangle below: As there are only 16384 routes, it is possible to solve this problem by trying every route. However, , is the same challenge with a triangle containing one hundred rows; it cannot be solved by brute force, and requires a clever method! ;o) NOTE: Problem 67 So, to recap a bit: we are given a triangle with numbers. Each number is connected to the two directly below it, . What we need to do is go from top to bottom and find the maximum possible path value. The sum of its values will be our answer. Also, as the problem states, since the routes are limited, we could try the solution out from calculating every possible route. As we’ll see just in a moment, , so we should scrap our heads a bit and find a better solution. forming a path directly to solve it by brute-forcing that would not be an efficient solution Ready then? Let’s solve it! Brute forcing the solution Since the problem advises that the brute force solution is not an efficient one, let’s take a moment to understand why is by looking at some smaller triangles. To know which path is the most valuable one, we first need to . It would be nice to know how many of them there are in the first place. walk over all the possible paths So, let’s try to count them: When there is one row (just one element) we have only one possible path. With two rows there are two possible paths we can take, with three rows there are 4 paths, and so on. Hopefully, you can see what’s happening here: : so for instance, with 4 rows we have 2³ = 8. In the problem’s example, the triangles have 15 rows so we count 2¹⁴ possible paths, which are 16384 possible routes, just as the problem states. the number of paths grows like 2ⁿ⁻¹ Given this, to calculate the maximum path by brute force, we would need to calculate the value of all the 16384 routes and then compare them. Besides the hassle of calculating all those possible paths, for the next similar problem mentioned by the text, which has 100 lines, we are going to have 2⁹⁹ possible paths, meaning roughly a total of 63.382.530.010.000.000.000.000.000.000 possible routes: inconvenient to say the least. Finally, , which is one of the worst results we could hope for. brute forcing the solution would force us with an algorithm with a time complexity of O(2ⁿ) We need a better method. The better method As we saw, the main problem in brute-forcing the solution is that there are so too many steps that our algorithm should do to calculate them all. . Since the number of routes depends on the number of elements in the triangle, we have no control over that variable. way too many possible routes, A better solution should involve either fewer routes or fewer steps to calculate them So, s needed for the algorithm to complete. What if we could calculate the maximum path possible of every given triangle … with just one step? our only chance is to reduce the number of step Let’s take a triangle and reason about it. This is the triangle from the sample given by the problem, for which we already know that the maximum path is 23 (the one in red), or 3 + 7 +4 + 9. This is our first hint: This means that we could track the growing value of each path while we go down the triangle and choose the maximum at the end. We can do this by . we don’t need intermediate values, we just want the sum of them. updating intermediate numbers with their actual path value As you can see we can apply this process row by row, updating all the values until, in the last row, we obtain the exact solution: 23. This is why I called it the avalanche algorithm: ! it’s like a snowball rolling down the mountain, which takes the route with the most snow on it and collects it to grow and grow again We have a problem though: for certain values inside the triangle we obtain multiple values. Take the middle 4, for example, the one in the third line of the first graph. If we come to the 4 from the 10 we are going to get 14 as a result, but if we come from the 7 we are going to obtain 11. If we think about it though, 10 and 7 in the second row are already “path values” updated from the previous row, since we’ve done 3+7=10 and 3+4=7 to get them. This means that the path coming from the 10 has already a higher value than the one coming from the 7 and, since we jump on the 4 only from those two values and from 4 we can only go to 5 and 9, we can conclude that the 10 path has an higher value than the 7 one, at least up to that point, and then we will always choose the route coming from the 10. It’s much clearer if you look it like this: Since we are coming only from 10 and 5, no matter if we are going to 5 or 9, we will always choose the path that already has a value of 10 because it’s higher than the one with a value of 7. This is our second hint: We could also store it, as we’ll do if we want to take the inverse road and print out the final path. besides adding the “path value” row by row and number by number, for each number we can choose the highest previous path value and move on. Each element in the triangle will have two values then: its real value and an “updated value” to keep track of the path value up to that point. OK: we have everything we need to start coding then. Let’s jump to the code! The code As with every good coder out there, I learned during the years to innately hate . Given this though, I feel like that for this project a solid OOP language was needed, to wrap everything up in a clear form. Go, even if is by far my first choice on almost every project, was not that suitable for this kind of task: its lack of a proper class object would have been a pain to deal with. Python is similar, since it has classes but it becomes messy to work with them after a while, in my opinion at least. , which is sold as Java 2.0: much less boilerplate but the same code potential for its interoperability with Java code. If you never used it, you’ll love it in a moment. Java So for this problem, we are going for Kotlin Also, this solution requires building a : if you are not familiar with that kind of data structure go check some resources online if you feel this solution is not clear! tree Let’s write the code then. https://gist.github.com/NicolaM94/81f57ae53d2c79012926047528ea492b?embedable=true We’ll start writing a class, which will be the blueprint for each number inside the triangle. This class must be built with two parameters, , the actual value of the number, and , the list of previous Nodes from which you can jump to this value (always at most 2). Then, inside the class we are going to build two other attributes: , which will hold the updated value of the path up to that point, and , which will hold the biggest parent of that node. We initially set to be equal to . Node value parents updatedValue chosenParent updatedValue value After that, we call the block, which will be the first block executed when the class gets created. In that block, we loop over the two parents (the loop there it’s a bit useless since there always are two parents, but it’s useful to show the idea behind it) and we update the with the found. Finally, we update to be equal to the value of this node plus the updated value of its bigger parent. init chosenParent biggestParent updatedValue We can also write a simple function which recursively print out the chain of values that brought us here. PrintChain That’s it, really: you have everything you need to solve the problem! Of course, you need a wrapper to collect all the numbers and instantiate all of them as nodes. Here’s the one I wrote, but of course you should build one yourself for your implementation. https://gist.github.com/NicolaM94/756a0d5fa2477454f16c4c23537a9b22?embedable=true Lastly, in the main function, we can wrap everything up and loop over the last line to get our solution. As always, for the problems given by ProjectEuler.net, besides presenting math riddles and problems, . And I love that: they are so committed that publishing results and guides for their problems is forbidden (this is one of the first 100 problems, so I understand it is permitted). Given this, . You have everything you need to start working on it! I won’t give out the solution: you need to build this by yourself, if you want to know it! ProjectEuler.net is a lovely project: their main focus is to provide a tool to start learning, thinking, and reasoning about math and algorithms I don’t feel it would be fair to just give out the solution to copy-paste and get the achievement Have a nice coding guys! 😎 Time complexity It’s not that easy to model the time complexity for an algorithm like this one, but we can try to by comparing it with the brute-force solution. reason a bit more generally First off, in both cases, we need to parse the numbers in the triangle somehow, either to collect them as nodes or to store them in lists, arrays, or any other method you can think of. Let’s suppose we use the same parsing algorithm for both solutions, the brute force one and the smart one, and let’s suppose this parsing algorithm takes O(n) time, where n is the number of rows in the triangle. After parsing the elements, in solution, we must calculate the value of each one of the 2ⁿ⁻¹ possible routes. This, again, . the brute force leaves us with a time complexity of O(2ⁿ⁻¹) With the smart solution though we don’t need to do anything else besides parsing the elements correctly. . Since we are using the same parsing algorithm for both solutions, we can discard the time used for parsing while comparing. Thus, Given that we must pass on each element to parse it as a Node, we’ll do exactly n(n+1)/2 operations After this we only need to loop over the last row to find the maximum path, requiring O(n) time. the smarter solution requires only O(n) time while the brute force requires O(2ⁿ⁻¹) time! Conclusion That’s it! As always, I hope you liked the article and found it somehow useful. If so, consider leaving a clap or two or sharing it: it would make my day! I also have a page, if you want to contribute to the redaction of these articles by sharing a coffee with me! Ko-Fi Also, feel free to share your thoughts or questions about this in the comment section: we’re all here to learn! As always, thanks for reading. Nicola

The Avalanche Algorithm — How To Calculate the Maximum Possible Path in a Binary Tree

This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license.
Authors:
(1) Cristina España-Bonet, DFKI GmbH, Saarland Informatics Campus. Table of Links Abstract and Intro
Corpora Compilation
Political Stance Classification
Summary and Conclusions
Limitations and Ethics Statement
Acknowledgments and References
A. Newspapers in OSCAR 22.01
B. Topics
C. Distribution of Topics per Newspaper
D. Subjects for the ChatGPT and Bard Article Generation
E. Stance Classification at Article Level
F. Training Details 3. Political Stance Classification The Network. **We finetune XLM-RoBERTa large (Conneau et al., 2020), a multilingual transformer-**based masked LM trained on 100 languages including the 4 we consider. The details of the network and the hyperparameter exploration per model are reported in Appendix F. The Models. We train 4 models: 3 monolingual finetunings with the English, German and Spanish data, plus a multilingual one with the shuffled concatenation of the data. All models are based on multilingual embeddings (RoBERTa) finetuned either monolingually or multilingually. Notice that we do not train any model for Catalan. With this, we want to compare the performance of mono- and multilingual finetunings and explore the possibility of using multilingual models for zero-shot language transfer. Coarse Classification with Newspaper Articles. Table 2 summarises the results. All the models achieve more than 95% accuracy on the validation set which is extracted from the same distribution as the training data. In order to see how the models behave with unseen data, we calculate the percentage of articles that are classified as Left (L) and Right (R) in the test newspapers of Table 1. We perform bootstrap resampling of the test sets with 1000 bootstraps to obtain confidence intervals at 95% level. We do not expect all the articles of a newspaper leaning towards the Left to show clear characteristics of the Left, but given that there is no neutral class, we expect the majority of them to be classified as Left. A good result is not necessarily 100%–0%, as this would not be realistic either. We consider that a newspaper has been classified as having a Left/Right political stance if more than 50% of its articles have been classified as such. These cases are boldfaced in Table 2. This is the behaviour we obtain for all the test newspapers but for the German Right-oriented newspaper: die Preußische Allgemeine Zeitung (PAZ). The German model is trained only on 12 newspapers to be compared to the 47 in English and 38 in Spanish. The incorrect classification might be an indication that diversity is a key aspect for the final model performance. Multilinguality does not help and 65% of the PAZ articles are still classified as Left oriented. We also assess the effectiveness of the English model on the German data, two close languages. We acknowledge that the topics of the USA and German newspapers might differ a lot, but the high diversity of the English training data could potentially compensate for this. The English model is able to correctly classify the German My Heimat as a Left-oriented newspaper (L: 67±3%) and PAZ as a Right-oriented one (R: 58±5%). We again attribute the difference to the German model being trained on a corpus lacking diversity. When we use the multilingual system, the dominant factor distinguishing the outputs is the language itself rather than the stance. The addition of English data is insufficient to alter the classification significantly. When we use the English system, the language does not play a role any more and only the stance features are considered. When we apply the English model to the Catalan newspapers we do not obtain satisfactory results though (95±1% for the Left but 16±3% for the Right newspaper) showing that the relatedness across languages is important. The multilingual model however properly detects the stance of the Catalan newspapers probably because it has been trained with an heterogeneous corpus that includes a related language (Spanish). We are able to perform zero-shot language transfer classification when we deal with close related languages. Coarse Classification with ILM-generated Articles. The bottom part of Table 2 details the results. We first focus on the English and Spanish models as the German one did not properly classify our test newspapers. The most relevant aspect to notice in ChatGPT is the strong change in political stance between February (v02) and May (v05) followed by a movement towards neutrality in August (v08). We checked that this polarity change is not an effect of the length of the outputs —the major shallow change in the generated articles. The training data in English has 5,730L–6,988 R articles with 584<length (words)<624 (similar to ChatPGTv05 length) and 4,563 L-7,127 R articles with 331< length<371 (similar to ChatGPtv02). In both cases the number of articles is larger for the Right stances, but the prediction for ChatGPTv02 clearly points towards the Left, rejecting the hypothesis that length plays a role in the classification. A similar things happens for Spanish. According to our models, the May 24th version of editorial line close to the right ideology, which differs from the ideology of the previous versions. Notably, this period corresponds drop in several task according to Chen et al. (2003). The German and Catalan outputs would still show an imprint from the Left ideology also in v05 but more diverse training data would be needed to confirm this with our monolingual models. It is interesting to notice that if we use the english monolingual model for German and Catalan, we still het the Left imprint (60±10% for German and 87±7% for Catalan). So we have indications that the political stance of ChatGPT depend on the language, which is not surprising in a data-driven system. The last version, ChatGPTv08, produces the most neutral texts, with only German clearly leaning towards the Left. The two generations, v08a and v08b, show that results are robust and are not tied to a particular generation. There is only a version available for multilingual Bard that covers our time frame.[7] The variation between generations is larger for Bard than for ChatGPT but, comparing v08 versions, Bard points towards the Left in a more consistent way across languages. Bard’s political orientation can also be determined by its answers to political test or quiz questions. The Political Compass (PC) site[8] defines 62 propositions to identify the political ideology —with an European/Western view— in two axes: economic policy (Left–Right) and social policy (Authoritarian–Libertarian), both in the range [-10,10]. Each proposition is followed by 4 alternatives: strongly agree, agree, disagree and strongly disagree. When prompted with the questionnaire,[9] Bard’s scores are (-6.50, -4.77) for English, (-8.00, -7.13) for German, (-5.75, -4.15) for Spanish and (-6.75, -4.56) for Catalan, where the first number corresponds to the economic policy and the second to the social policy. The results are in concordance with Table 2 and give an indirect validation of our method which does not rely on direct questions.[10] This kind of analysis is not possible with ChatGPT any more as it refrains from expressing opinions and preferences, demonstrating the relevance of an approach that detects the leaning in a more indirect way. Also notice that these questionnaires are well-known and public, so it would be easy to instruct a LM to avoid the questions or react to its propositions in a neutral manner. Previous work used only political tests and questionnaires to estimate ChatGPT’s orientation. Hartmann et al. (2023) used PC, 38 political statements from the voting advice application Wahl-O-Mat (Germany) and 30 from StemWijzer (the Netherlands) to conclude that ChatGPT’s ideology in its version of Dec 15 2022 was pro-environmental and left-libertarian. A study conducted by the Manhattan Institute for Policy Research[11] reported that ChatGPT tended to give responses typical of Left-of-center political viewpoints for English (Rozado, 2023). The authors administered 15 political orientation tests to the ChatGPT version of Jan 9. Their results are consistent with our evaluation of the Feb 13 model. Finally, Motoki et al. (2023) performed a battery of tests based on PC to show that ChatGPT is strongly biased towards the Left. The authors do not state the version they use, but the work was submitted on March 2023. All these results are therefore before the move to the Right we detected in May. [7] Notice that the version we use does not officially support Catalan, but native speakers confirmed that generations are mostly correct and fluent with few grammatical mistakes. [8] https://www.politicalcompass.org/test (accessed between 13th and 20th August 2023) [9] The Spanish questionnaire was translated into Catalan, as the questionnaire was not available. [10] Even though, similarly to people, it is possible for an ILM to say one thing (chose an option for a proposition) and act (write a text) in an inconsistent way. [11] A conservative think tank according to Wikipedia. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Authors: (1) Cristina España-Bonet, DFKI GmbH, Saarland Informatics Campus. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Authors: Authors: (1) Cristina España-Bonet, DFKI GmbH, Saarland Informatics Campus. Table of Links Abstract and Intro Corpora Compilation Political Stance Classification Summary and Conclusions Limitations and Ethics Statement Acknowledgments and References A. Newspapers in OSCAR 22.01 B. Topics C. Distribution of Topics per Newspaper D. Subjects for the ChatGPT and Bard Article Generation E. Stance Classification at Article Level F. Training Details Abstract and Intro Abstract and Intro Corpora Compilation Corpora Compilation Political Stance Classification Political Stance Classification Summary and Conclusions Summary and Conclusions Limitations and Ethics Statement Limitations and Ethics Statement Acknowledgments and References Acknowledgments and References A. Newspapers in OSCAR 22.01 A. Newspapers in OSCAR 22.01 B. Topics B. Topics C. Distribution of Topics per Newspaper C. Distribution of Topics per Newspaper D. Subjects for the ChatGPT and Bard Article Generation D. Subjects for the ChatGPT and Bard Article Generation E. Stance Classification at Article Level E. Stance Classification at Article Level F. Training Details F. Training Details 3. Political Stance Classification The Network. **We finetune XLM-RoBERTa large (Conneau et al., 2020), a multilingual transformer-**based masked LM trained on 100 languages including the 4 we consider. The details of the network and the hyperparameter exploration per model are reported in Appendix F. The Models. We train 4 models: 3 monolingual finetunings with the English, German and Spanish data, plus a multilingual one with the shuffled concatenation of the data. All models are based on multilingual embeddings (RoBERTa) finetuned either monolingually or multilingually. Notice that we do not train any model for Catalan. With this, we want to compare the performance of mono- and multilingual finetunings and explore the possibility of using multilingual models for zero-shot language transfer. The Models. Coarse Classification with Newspaper Articles. Table 2 summarises the results. All the models achieve more than 95% accuracy on the validation set which is extracted from the same distribution as the training data. In order to see how the models behave with unseen data, we calculate the percentage of articles that are classified as Left (L) and Right (R) in the test newspapers of Table 1. We perform bootstrap resampling of the test sets with 1000 bootstraps to obtain confidence intervals at 95% level. We do not expect all the articles of a newspaper leaning towards the Left to show clear characteristics of the Left, but given that there is no neutral class, we expect the majority of them to be classified as Left. A good result is not necessarily 100%–0%, as this would not be realistic either. We consider that a newspaper has been classified as having a Left/Right political stance if more than 50% of its articles have been classified as such. These cases are boldfaced in Table 2. Coarse Classification with Newspaper Articles. This is the behaviour we obtain for all the test newspapers but for the German Right-oriented newspaper: die Preußische Allgemeine Zeitung (PAZ). The German model is trained only on 12 newspapers to be compared to the 47 in English and 38 in Spanish. The incorrect classification might be an indication that diversity is a key aspect for the final model performance. Multilinguality does not help and 65% of the PAZ articles are still classified as Left oriented. We also assess the effectiveness of the English model on the German data, two close languages. We acknowledge that the topics of the USA and German newspapers might differ a lot, but the high diversity of the English training data could potentially compensate for this. The English model is able to correctly classify the German My Heimat as a Left-oriented newspaper (L: 67±3%) and PAZ as a Right-oriented one (R: 58±5%). We again attribute the difference to the German model being trained on a corpus lacking diversity. When we use the multilingual system, the dominant factor distinguishing the outputs is the language itself rather than the stance. The addition of English data is insufficient to alter the classification significantly. When we use the English system, the language does not play a role any more and only the stance features are considered. When we apply the English model to the Catalan newspapers we do not obtain satisfactory results though (95±1% for the Left but 16±3% for the Right newspaper) showing that the relatedness across languages is important. The multilingual model however properly detects the stance of the Catalan newspapers probably because it has been trained with an heterogeneous corpus that includes a related language (Spanish). We are able to perform zero-shot language transfer classification when we deal with close related languages. Coarse Classification with ILM-generated Articles. The bottom part of Table 2 details the results. We first focus on the English and Spanish models as the German one did not properly classify our test newspapers. The most relevant aspect to notice in ChatGPT is the strong change in political stance between February (v02) and May (v05) followed by a movement towards neutrality in August (v08). We checked that this polarity change is not an effect of the length of the outputs —the major shallow change in the generated articles. The training data in English has 5,730L–6,988 R articles with 584<length (words)<624 (similar to ChatPGTv05 length) and 4,563 L-7,127 R articles with 331< length<371 (similar to ChatGPtv02). In both cases the number of articles is larger for the Right stances, but the prediction for ChatGPTv02 clearly points towards the Left, rejecting the hypothesis that length plays a role in the classification. A similar things happens for Spanish. According to our models, the May 24th version of editorial line close to the right ideology, which differs from the ideology of the previous versions. Notably, this period corresponds drop in several task according to Chen et al. (2003). The German and Catalan outputs would still show an imprint from the Left ideology also in v05 but more diverse training data would be needed to confirm this with our monolingual models. It is interesting to notice that if we use the english monolingual model for German and Catalan, we still het the Left imprint (60±10% for German and 87±7% for Catalan). So we have indications that the political stance of ChatGPT depend on the language, which is not surprising in a data-driven system. The last version, ChatGPTv08, produces the most neutral texts, with only German clearly leaning towards the Left. The two generations, v08a and v08b, show that results are robust and are not tied to a particular generation. There is only a version available for multilingual Bard that covers our time frame.[7] The variation between generations is larger for Bard than for ChatGPT but, comparing v08 versions, Bard points towards the Left in a more consistent way across languages. Bard’s political orientation can also be determined by its answers to political test or quiz questions. The Political Compass (PC) site[8] defines 62 propositions to identify the political ideology —with an European/Western view— in two axes: economic policy (Left–Right) and social policy (Authoritarian–Libertarian), both in the range [-10,10]. Each proposition is followed by 4 alternatives: strongly agree, agree, disagree and strongly disagree. When prompted with the questionnaire,[9] Bard’s scores are (-6.50, -4.77) for English, (-8.00, -7.13) for German, (-5.75, -4.15) for Spanish and (-6.75, -4.56) for Catalan, where the first number corresponds to the economic policy and the second to the social policy. The results are in concordance with Table 2 and give an indirect validation of our method which does not rely on direct questions.[10] This kind of analysis is not possible with ChatGPT any more as it refrains from expressing opinions and preferences, demonstrating the relevance of an approach that detects the leaning in a more indirect way. Also notice that these questionnaires are well-known and public, so it would be easy to instruct a LM to avoid the questions or react to its propositions in a neutral manner. Previous work used only political tests and questionnaires to estimate ChatGPT’s orientation. Hartmann et al. (2023) used PC, 38 political statements from the voting advice application Wahl-O-Mat (Germany) and 30 from StemWijzer (the Netherlands) to conclude that ChatGPT’s ideology in its version of Dec 15 2022 was pro-environmental and left-libertarian. A study conducted by the Manhattan Institute for Policy Research[11] reported that ChatGPT tended to give responses typical of Left-of-center political viewpoints for English (Rozado, 2023). The authors administered 15 political orientation tests to the ChatGPT version of Jan 9. Their results are consistent with our evaluation of the Feb 13 model. Finally, Motoki et al. (2023) performed a battery of tests based on PC to show that ChatGPT is strongly biased towards the Left. The authors do not state the version they use, but the work was submitted on March 2023. All these results are therefore before the move to the Right we detected in May. [7] Notice that the version we use does not officially support Catalan, but native speakers confirmed that generations are mostly correct and fluent with few grammatical mistakes. [8] https://www.politicalcompass.org/test (accessed between 13th and 20th August 2023) [9] The Spanish questionnaire was translated into Catalan, as the questionnaire was not available. [10] Even though, similarly to people, it is possible for an ILM to say one thing (chose an option for a proposition) and act (write a text) in an inconsistent way. [11] A conservative think tank according to Wikipedia.

Multilingual Coarse Political Stance Classification of Media: Political Stance Classification

মিডিয়ার বহুভাষিক মোটা রাজনৈতিক অবস্থানের শ্রেণীবিভাগ: রাজনৈতিক অবস্থানের শ্রেণীবিভাগ

Mehrsprachige grobe politische Haltungsklassifizierung von Medien: Politische Haltungsklassifizierung

Clasificación multilingüe de posturas políticas generales de los medios: Clasificación de posturas políticas

Classification multilingue grossière des positions politiques des médias : classification des positions politiques

बहुभाषी मोटे राजनीतिक रुख मीडिया का वर्गीकरण: राजनीतिक रुख वर्गीकरण

多言語メディアの粗い政治的立場の分類: 政治的立場の分類

미디어의 다국어 대략적인 정치적 입장 분류: 정치적 입장 분류

Classificação Multilíngue da Posição Política Grosseira da Mídia: Classificação da Posição Política

Многоязычная грубая классификация СМИ по политической позиции: Классификация политической позиции

Medyanın Çok Dilli Kaba Siyasi Duruş Sınıflandırması: Siyasi Duruş Sınıflandırması

Phân loại lập trường chính trị thô sơ đa ngôn ngữ của phương tiện truyền thông: Phân loại lập trường chính trị

多语种媒体粗略政治立场分类：政治立场分类

This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license.
Authors:
(1) Cristina España-Bonet, DFKI GmbH, Saarland Informatics Campus. Table of Links Abstract and Intro
Corpora Compilation
Political Stance Classification
Summary and Conclusions
Limitations and Ethics Statement
Acknowledgments and References
A. Newspapers in OSCAR 22.01
B. Topics
C. Distribution of Topics per Newspaper
D. Subjects for the ChatGPT and Bard Article Generation
E. Stance Classification at Article Level
F. Training Details 5.1 Limitations We are assuming that All media sources have an editorial line and an associated bias, and we treat the ILM as any other media source. We do not consider the possibility of a ChatGPT or Bard article being unbiased. This is related to the distant supervision method used to gather the data that currently allows for a binary political stance annotation. Since manually annotating hundreds of thousands of articles with political biases in a truly multilingual setting seems not possible in the foreseeable future, we decided to implement a completely data-based method and study its language and culture transfer capabilities. Using distant supervision for detecting the political stance at article level is a delicate topic though. First, because the same newspaper can change ideology over time. Second, and this is more related to the content of an individual article, non-controversial subjects might not have a bias. Even in cases where bias exists, there is a spectrum ranging from the extreme Left to the extreme Right, rather than a clear-cut division between the two ideologies. In order to quantify and if possible mitigate the current limitations, we plan to conduct a stylistic analysis of the human-annotated corpora (Baly et al., 2020; Aksenov et al., 2021) and compare it to our semi-automatically annotated corpus. As a follow-up of this work, we will perform a stylistic analysis of the ILM-generated texts too as a similar style between the training data and these texts is needed to ensure good generalisation and transfer capabilities. 5.2. Ethics Statement We use generative language models, ChatGPT and Bard, to create our test data. Since we deal with several controversial subjects (death penalty, sexual harassment, drugs, etc.) the automatic generation might produce harmful text. The data presented here has not undergone any human revision. We analyse and provide the corpus as it was generated, along with the indication of the systems version used. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Authors: (1) Cristina España-Bonet, DFKI GmbH, Saarland Informatics Campus. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Authors: Authors: (1) Cristina España-Bonet, DFKI GmbH, Saarland Informatics Campus. Table of Links Abstract and Intro Corpora Compilation Political Stance Classification Summary and Conclusions Limitations and Ethics Statement Acknowledgments and References A. Newspapers in OSCAR 22.01 B. Topics C. Distribution of Topics per Newspaper D. Subjects for the ChatGPT and Bard Article Generation E. Stance Classification at Article Level F. Training Details Abstract and Intro Abstract and Intro Corpora Compilation Corpora Compilation Political Stance Classification Political Stance Classification Summary and Conclusions Summary and Conclusions Limitations and Ethics Statement Limitations and Ethics Statement Acknowledgments and References Acknowledgments and References A. Newspapers in OSCAR 22.01 A. Newspapers in OSCAR 22.01 B. Topics B. Topics C. Distribution of Topics per Newspaper C. Distribution of Topics per Newspaper D. Subjects for the ChatGPT and Bard Article Generation D. Subjects for the ChatGPT and Bard Article Generation E. Stance Classification at Article Level E. Stance Classification at Article Level F. Training Details F. Training Details 5.1 Limitations We are assuming that All media sources have an editorial line and an associated bias , and we treat the ILM as any other media source. We do not consider the possibility of a ChatGPT or Bard article being unbiased. This is related to the distant supervision method used to gather the data that currently allows for a binary political stance annotation. Since manually annotating hundreds of thousands of articles with political biases in a truly multilingual setting seems not possible in the foreseeable future, we decided to implement a completely data-based method and study its language and culture transfer capabilities. All media sources have an editorial line and an associated bias Using distant supervision for detecting the political stance at article level is a delicate topic though. First, because the same newspaper can change ideology over time. Second, and this is more related to the content of an individual article, non-controversial subjects might not have a bias. Even in cases where bias exists, there is a spectrum ranging from the extreme Left to the extreme Right, rather than a clear-cut division between the two ideologies. In order to quantify and if possible mitigate the current limitations, we plan to conduct a stylistic analysis of the human-annotated corpora (Baly et al., 2020; Aksenov et al., 2021) and compare it to our semi-automatically annotated corpus. As a follow-up of this work, we will perform a stylistic analysis of the ILM-generated texts too as a similar style between the training data and these texts is needed to ensure good generalisation and transfer capabilities. 5.2. Ethics Statement We use generative language models, ChatGPT and Bard, to create our test data. Since we deal with several controversial subjects (death penalty, sexual harassment, drugs, etc.) the automatic generation might produce harmful text. The data presented here has not undergone any human revision. We analyse and provide the corpus as it was generated, along with the indication of the systems version used.

Multilingual Coarse Political Stance Classification of Media: Limitations & Ethics Statement

মিডিয়ার বহুভাষিক মোটা রাজনৈতিক অবস্থানের শ্রেণীবিভাগ: সীমাবদ্ধতা এবং নীতিশাস্ত্রের বিবৃতি

Mehrsprachige grobe politische Haltung Klassifizierung von Medien: Einschränkungen und Ethikerklärung

Postura política burda multilingüe Clasificación de los medios: limitaciones y declaración de ética

Classification multilingue grossière des positions politiques des médias : limites et déclaration éthique

बहुभाषी असभ्य राजनीतिक रुख मीडिया का वर्गीकरण: सीमाएँ और नैतिकता वक्तव्य

多言語によるメディアの粗い政治的立場の分類：制限と倫理声明

미디어의 다국어 거친 정치적 입장 분류: 한계 및 윤리 선언문

Classificação multilíngue da postura política grosseira da mídia: limitações e declaração de ética

Многоязычная грубая классификация СМИ по политической позиции: ограничения и этическое заявление

Medyanın Çok Dilli Kaba Siyasi Duruş Sınıflandırması: Sınırlamalar ve Etik Beyanı

Phân loại lập trường chính trị thô sơ đa ngôn ngữ của phương tiện truyền thông: Hạn chế và tuyên bố đạo đức

多语言粗略政治立场媒体分类：限制与道德声明

Revolutionizing Healthcare: Andy Schoonover's Playbook for Disrupting Entrenched Industries

বিপ্লবী স্বাস্থ্যসেবা: প্রবেশ করা শিল্পগুলিকে ব্যাহত করার জন্য অ্যান্ডি শুনোভারের প্লেবুক

Revolutionierung des Gesundheitswesens: Andy Schoonovers Strategie zur Umwälzung etablierter Branchen

Revolucionando la atención médica: el manual de Andy Schoonover para revolucionar industrias arraigadas

Révolutionner les soins de santé : le manuel d'Andy Schoonover pour perturber les industries bien établies

स्वास्थ्य सेवा में क्रांति: स्थापित उद्योगों में बदलाव के लिए एंडी स्कूनओवर की कार्यपुस्तिका

ヘルスケアの革命: 定着した業界を変革するためのアンディ・シューノーバーの戦略

의료 혁신: 고착화된 산업을 혁신하기 위한 Andy Schoonover의 플레이북

Revolucionando a saúde: o manual de Andy Schoonover para desestabilizar indústrias entrincheiradas

Революция в здравоохранении: книга Энди Шуновера по разрушению устоявшихся отраслей

Sağlık Hizmetinde Devrim Yaratmak: Andy Schoonover'ın Yerleşik Endüstrileri Yıkmaya Yönelik Başucu Kitabı

Cách mạng hóa hoạt động chăm sóc sức khỏe: Sách hướng dẫn cách phá vỡ các ngành công nghiệp cố thủ của Andy Schoonover

医疗保健革命：安迪·斯库诺弗颠覆根深蒂固行业的策略

Dear America, I Am Breaking Up With You

প্রিয় আমেরিকা, আমি তোমার সাথে ব্রেকিং আপ করছি

Liebes Amerika, ich mache Schluss mit Dir

Querida América, estoy rompiendo contigo

Chère Amérique, je romps avec toi

प्रिय अमेरिका, मैं तुमसे नाता तोड़ रहा हूँ

親愛なるアメリカへ、私はあなたと別れます

사랑하는 미국이여, 나는 당신과 헤어지고 있습니다

Querida América, estou terminando com você

Дорогая Америка, я расстаюсь с тобой

Sevgili Amerika, Senden Ayrılıyorum

Nước Mỹ thân mến, tôi sẽ chia tay với bạn

亲爱的美国，我要和你分手

Next-Gen FHE Toolchains for the Dev Multiverse: How TFHE Is Taking Us There

Welcome, As a Spammer of the registry (pun intended), I want to explain to you what's going on with the NPM registry and its management. At first, I felt the title `It's Party Time For NPM Spammers🥳` was offensive, unlawful, or disrespectful, and I also asked for help from an AI chatbot. I got the following results, but I purposefully stayed with the same to get attention from both NPM and the dev community. AI Chatbot: Certainly! Here are some alternative titles for your blog post on spammers in the NPM registry: “Navigating the Murky Waters: A Deep Dive into NPM Registry Spam”
“Guarding the NPM Gates: Strategies to Combat Registry Spam”
“Unmasking the Shadows: Inside the World of NPM Registry Spammers” Let's start with some basics. I request that you take a close look at the forthcoming screenshots. What is NPM? It is the world's largest software registry and a Node Package Manager for the JavaScript programming language maintained by Microsoft's npm, Inc. Types of Spam Let's list out some with my experience and a few real examples currently living in the registry. Keyword Stuffing:
Packages using irrelevant keywords to appear in more search results.



Malicious Code:
Packages containing harmful code or backdoors.


Rewarding decentralized protocol (New Type):
These packages were created to get some incentives from the OSS blockchain technology. You can learn more from here.




Fake Packages:
Impersonating popular packages to deceive users.


Low-Quality Packages:
Flooded with typos, irrelevant content, or empty functionality.



Bot-Generated Packages:
Automated spam submissions.



Phishing Attempts:
Packages designed to steal sensitive information.


Dependency Spam:
Malicious packages as dependencies of legitimate ones.


Hello World Packages (Low priority issue):
The students & learners try their first-time packages.



Weird Packages:
There are many packages with index.js file containing a comment which says something like the below and I try to understand with the google translator, no luck, if you got it, please let me know in the comments. Fancy Packages Control:
These kinds of users acquire short package names. NOTE: This list excludes security related malicious packages as they are often reported by the experts. Some notable past incidents: https://www.sonatype.com/resources/blog/npm-flooded-with-748-packages-that-store-movies
https://thehackernews.com/2023/09/fresh-wave-of-malicious-npm-packages.html
https://thehackernews.com/2023/08/malicious-npm-packages-found.html Some Spammers List Here, I am going to list some of them because it is simply not possible to list all of them here, also, with the number of packages published by each one. https://www.npmjs.com/\~onedionysc - 6931


https://www.npmjs.com/\~shivamkalsi2024 - 997


https://www.npmjs.com/\~79w - 551


https://www.npmjs.com/\~ellentea - 599


https://www.npmjs.com/\~uirewikilabs - 323


https://www.npmjs.com/\~quinterochris100 - 361


https://www.npmjs.com/\~vanthuanbt26 - 250


https://www.npmjs.com/\~tiengiangb47 - 230


https://www.npmjs.com/\~swenkertreanpm - 227


https://www.npmjs.com/\~loandinhb931 - 224 These are my random picks, and who knows how many of them are. The Team's Response The Problem The problem is not the spammers, as they always will be, but the real problem is within the management. Take, for example: If someone reports this via one of your support forums, the reply they get is simply redirection; they are redirecting you to fill out their respective forms for spam submissions. What if the reporter did not want to do extra work here? The important thing to note here is that the spammers and their packages live happily, even after we provide the absolute spam users and their package links. Conclusion I am writing this blog because I waited too long for them to fix it. Justice delayed is justice denied
- William Blackstone Please don't forget to check out our important Articles: Introducing Our New JavaScript Standard Library
You Don’t Need JavaScript Native Methods! Happy coding! 🚀 🙏 Thanks for reading. Welcome, As a Spammer of the registry (pun intended) , I want to explain to you what's going on with the NPM registry and its management. Spammer ( pun intended ) pun intended At first, I felt the title ` It's Party Time For NPM Spammers🥳 ` was offensive, unlawful, or disrespectful, and I also asked for help from an AI chatbot. I got the following results, but I purposefully stayed with the same to get attention from both NPM and the dev community. It's Party Time For NPM Spammers🥳 AI Chatbot : Certainly! Here are some alternative titles for your blog post on spammers in the NPM registry: AI Chatbot “Navigating the Murky Waters: A Deep Dive into NPM Registry Spam” “Guarding the NPM Gates: Strategies to Combat Registry Spam” “Unmasking the Shadows: Inside the World of NPM Registry Spammers” “Navigating the Murky Waters: A Deep Dive into NPM Registry Spam” “Navigating the Murky Waters: A Deep Dive into NPM Registry Spam” “Guarding the NPM Gates: Strategies to Combat Registry Spam” “Guarding the NPM Gates: Strategies to Combat Registry Spam” “Unmasking the Shadows: Inside the World of NPM Registry Spammers” “Unmasking the Shadows: Inside the World of NPM Registry Spammers” Let's start with some basics. I request that you take a close look at the forthcoming screenshots. What is NPM? It is the world's largest software registry and a Node Package Manager for the JavaScript programming language maintained by Microsoft's npm, Inc. Types of Spam Types of Spam Let's list out some with my experience and a few real examples currently living in the registry. Keyword Stuffing:
Packages using irrelevant keywords to appear in more search results. Malicious Code:
Packages containing harmful code or backdoors. Rewarding decentralized protocol (New Type):
These packages were created to get some incentives from the OSS blockchain technology. You can learn more from here. Fake Packages:
Impersonating popular packages to deceive users. Low-Quality Packages:
Flooded with typos, irrelevant content, or empty functionality. Bot-Generated Packages:
Automated spam submissions. Phishing Attempts:
Packages designed to steal sensitive information. Dependency Spam:
Malicious packages as dependencies of legitimate ones. Hello World Packages (Low priority issue):
The students & learners try their first-time packages. Weird Packages:
There are many packages with index.js file containing a comment which says something like the below and I try to understand with the google translator, no luck, if you got it, please let me know in the comments. Keyword Stuffing: Packages using irrelevant keywords to appear in more search results. Keyword Stuffing : Keyword Stuffing Packages using irrelevant keywords to appear in more search results. Malicious Code: Packages containing harmful code or backdoors. Malicious Code : Malicious Code Packages containing harmful code or backdoors. Rewarding decentralized protocol (New Type): These packages were created to get some incentives from the OSS blockchain technology. You can learn more from here. Rewarding decentralized protocol (New Type) : Rewarding decentralized protocol (New Type) These packages were created to get some incentives from the OSS blockchain technology. You can learn more from here . learn more from here Fake Packages: Impersonating popular packages to deceive users. Fake Packages : Fake Packages Impersonating popular packages to deceive users. Low-Quality Packages: Flooded with typos, irrelevant content, or empty functionality. Low-Quality Packages : Low-Quality Packages Flooded with typos, irrelevant content, or empty functionality. Bot-Generated Packages: Automated spam submissions. Bot-Generated Packages : Bot-Generated Packages Automated spam submissions. Phishing Attempts: Packages designed to steal sensitive information. Phishing Attempts : Phishing Attempts Packages designed to steal sensitive information. Dependency Spam: Malicious packages as dependencies of legitimate ones. Dependency Spam : Dependency Spam Malicious packages as dependencies of legitimate ones. Hello World Packages (Low priority issue): The students & learners try their first-time packages. Hello World Packages (Low priority issue) : Hello World Packages (Low priority issue) The students & learners try their first-time packages. Weird Packages: There are many packages with index.js file containing a comment which says something like the below and I try to understand with the google translator, no luck, if you got it, please let me know in the comments. Weird Packages: Weird Packages: There are many packages with index.js file containing a comment which says something like the below and I try to understand with the google translator, no luck, if you got it, please let me know in the comments. Fancy Packages Control:
These kinds of users acquire short package names. Fancy Packages Control: These kinds of users acquire short package names. Fancy Packages Control : Fancy Packages Control These kinds of users acquire short package names. NOTE : This list excludes security related malicious packages as they are often reported by the experts. NOTE Some notable past incidents: https://www.sonatype.com/resources/blog/npm-flooded-with-748-packages-that-store-movies https://thehackernews.com/2023/09/fresh-wave-of-malicious-npm-packages.html https://thehackernews.com/2023/08/malicious-npm-packages-found.html https://www.sonatype.com/resources/blog/npm-flooded-with-748-packages-that-store-movies https://www.sonatype.com/resources/blog/npm-flooded-with-748-packages-that-store-movies https://thehackernews.com/2023/09/fresh-wave-of-malicious-npm-packages.html https://thehackernews.com/2023/09/fresh-wave-of-malicious-npm-packages.html https://thehackernews.com/2023/08/malicious-npm-packages-found.html https://thehackernews.com/2023/08/malicious-npm-packages-found.html Some Spammers List Here, I am going to list some of them because it is simply not possible to list all of them here, also, with the number of packages published by each one. https://www.npmjs.com/\~onedionysc - 6931 https://www.npmjs.com/\~shivamkalsi2024 - 997 https://www.npmjs.com/\~79w - 551 https://www.npmjs.com/\~ellentea - 599 https://www.npmjs.com/\~uirewikilabs - 323 https://www.npmjs.com/\~quinterochris100 - 361 https://www.npmjs.com/\~vanthuanbt26 - 250 https://www.npmjs.com/\~tiengiangb47 - 230 https://www.npmjs.com/\~swenkertreanpm - 227 https://www.npmjs.com/\~loandinhb931 - 224 https://www.npmjs.com/\~onedionysc - 6931 https://www.npmjs.com/\~onedionysc - 6931 https://www.npmjs.com/\~onedionysc 6931 https://www.npmjs.com/\~shivamkalsi2024 - 997 https://www.npmjs.com/\~shivamkalsi2024 - 997 https://www.npmjs.com/\~shivamkalsi2024 https://www.npmjs.com/\~79w - 551 https://www.npmjs.com/\~79w - 551 https://www.npmjs.com/\~79w https://www.npmjs.com/\~ellentea - 599 https://www.npmjs.com/\~ellentea - 599 https://www.npmjs.com/\~ellentea https://www.npmjs.com/\~uirewikilabs - 323 https://www.npmjs.com/\~uirewikilabs - 323 https://www.npmjs.com/\~uirewikilabs https://www.npmjs.com/\~quinterochris100 - 361 https://www.npmjs.com/\~quinterochris100 - 361 https://www.npmjs.com/\~quinterochris100 https://www.npmjs.com/\~vanthuanbt26 - 250 https://www.npmjs.com/\~vanthuanbt26 - 250 https://www.npmjs.com/\~vanthuanbt26 https://www.npmjs.com/\~tiengiangb47 - 230 https://www.npmjs.com/\~tiengiangb47 - 230 https://www.npmjs.com/\~tiengiangb47 https://www.npmjs.com/\~swenkertreanpm - 227 https://www.npmjs.com/\~swenkertreanpm - 227 https://www.npmjs.com/\~swenkertreanpm https://www.npmjs.com/\~loandinhb931 - 224 https://www.npmjs.com/\~loandinhb931 - 224 https://www.npmjs.com/\~loandinhb931 These are my random picks, and who knows how many of them are. The Team's Response The Problem The problem is not the spammers, as they always will be, but the real problem is within the management. Take, for example: If someone reports this via one of your support forums, the reply they get is simply redirection; they are redirecting you to fill out their respective forms for spam submissions. What if the reporter did not want to do extra work here? extra The important thing to note here is that the spammers and their packages live happily, even after we provide the absolute spam users and their package links. Conclusion I am writing this blog because I waited too long for them to fix it. Justice delayed is justice denied - William Blackstone Justice delayed is justice denied Justice delayed is justice denied - William Blackstone Please don't forget to check out our important Articles: important Introducing Our New JavaScript Standard Library You Don’t Need JavaScript Native Methods! Introducing Our New JavaScript Standard Library Introducing Our New JavaScript Standard Library Introducing Our New JavaScript Standard Library You Don’t Need JavaScript Native Methods! You Don’t Need JavaScript Native Methods! You Don’t Need JavaScript Native Methods! Happy coding! 🚀 🙏 Thanks for reading.

It's Party Time For NPM Spammers🥳

Integration with blockchain still remains a mystery for many engineers, even for those who might be familiar with it as end users. While smart contract development often attracts most of the attention and coverage in technical write-ups, developers are often left to their own devices when it comes to figuring out how to integrate blockchain technology with the rest of their application. The Casper Ecosystem thankfully offers solutions that eliminate  problems around getting started and deploying live dApps on Casper in one complementary product suite, making it one of the most developer-friendly blockchain ecosystems. Developed by MAKE and granted by Casper Association, CSPR.click, CSPR.cloud and CSPR.build address user interactions with the blockchain and provide tools to read and observe the network activity in real time. ** ** CSPR.click elegantly solves the end user authentication problem by welcoming users with a familiar-looking Single-Sign-On experience where they can connect to any Casper-enabled application using their preferred wallet. Users who are entirely new to the blockchain can enjoy a smooth onboarding experience by using self-custodial wallets linked to social logins, their email address or cell phone number. When it comes to being able to pay for contract calls, integration with multiple on-ramps guarantees that all users will be able to obtain tokens quickly and easily, regardless of their geographical region. Developers can quick-start their CSPR.click integration, and complete customization is also possible using the lower-level SDK. Mobile Casper Wallet is automatically supported, making your Casper-enabled application truly cross-platform out of the gate. Most developers, regardless of their prior experience with blockchain integrations, can bootstrap a functional front-end application with CSPR.click in a matter of minutes. REST APIs, WebSockets and Node RPC with CSPR.cloud Applications, including those integrated using CSPR.click, can rely on CSPR.cloud to observe on-chain activity on the Casper Network. CSPR.cloud is an enterprise-grade middleware platform that can be used separately from CSPR.click, and equips developers with highly available and scalable REST APIs that provide low-latency access to normalized and indexed blockchain data, fungible and non-fungible token activity, historical pricing data, and related off-chain metadata. Real-time WebSocket streams make it possible to granularly subscribe to the data you are interested in, allowing your application to react immediately whenever your smart contracts emit an event or when an on-chain entity you observe changes its state. Both the REST API and the Real-time WebSocket streams rely entirely on widely used internet standards, making it exceedingly easy to integrate with existing and new applications without the need to know or learn any new technologies. This unlocks Casper blockchain development to the tens of millions of software engineers at businesses and enterprises who’ve never  interacted with a blockchain before.  And for those applications that need direct access to the native Casper Node RPC API, CSPR.cloud provides a cache-enabled, high-performance and highly-available managed node infrastructure for maximally performant access without the associated overhead. Both CSPR.click and CSPR.cloud come with comprehensive documentation and examples. Developers can start building on the Casper network with CSPR.click and CSPR.cloud by registering on CSPR.build and generating their access keys for instant access. Casper is Solving Real World Problems For Builders In most environments, to be able to interact with smart contracts, end users must possess blockchain accounts with sufficient token balances to cover smart contract execution costs. Getting users through the process of creating and managing private and public keys, keeping their recovery phrases safe, as well as funding their accounts with tokens, already represents a serious UX challenge. And this corresponds to a significant bounce rate of new users, and lower retention of existing ones. For developers, the underlying technical integration with multiple crypto wallets and fiat on-ramps adds another layer of complexity, demanding considerable initial engineering effort and time, as well as ongoing maintenance and up-keep with ever- and rapidly evolving standards, regulatory constraints, APIs and SDKs of third party wallet and on-ramp providers. CSPR.click solves all this for application developers. Casper also helps developers address that smart contracts are executed on-chain asynchronously, i.e. the end user submits a transaction to the network, the transaction gets processed when a new block is created, and that results in a change of state. Applications must constantly monitor network activity to properly respond to the contract logic and any state changes it may produce that are relevant to their overall application. This requires not only an in-depth understanding of how the blockchain works but also a reliable and scalable infrastructure. While theoretically, the application can connect to any public node, in practice, the reliability of these nodes remains uncertain. A serious application or business will certainly not want to rely on just any public node, without assured redundancy, maintenance, uptime and disaster recovery processes in place, and without dedicated domain experts to ensure their application always runs smoothly. The suite of building tools in CSPR.build, CSPR.click and CSPR.cloud ensures developers do not have to worry about the integration process, and can exclusively focus on their creativity and core business. Build with Casper by registering on CSPR.build and generate your access keys for instant access. This story was distributed as a release by Btcwire under HackerNoon’s Business Blogging Program. Learn more about the program here. Integration with blockchain still remains a mystery for many engineers, even for those who might be familiar with it as end users. While smart contract development often attracts most of the attention and coverage in technical write-ups, developers are often left to their own devices when it comes to figuring out how to integrate blockchain technology with the rest of their application. The Casper Ecosystem thankfully offers solutions that eliminate  problems around getting started and deploying live dApps on Casper in one complementary product suite, making it one of the most developer-friendly blockchain ecosystems. Developed by MAKE and granted by Casper Association, CSPR.click, CSPR.cloud and CSPR.build address user interactions with the blockchain and provide tools to read and observe the network activity in real time. MAKE MAKE ** ** CSPR.click elegantly solves the end user authentication problem by welcoming users with a familiar-looking Single-Sign-On experience where they can connect to any Casper-enabled application using their preferred wallet. CSPR.click CSPR.click Users who are entirely new to the blockchain can enjoy a smooth onboarding experience by using self-custodial wallets linked to social logins, their email address or cell phone number. When it comes to being able to pay for contract calls, integration with multiple on-ramps guarantees that all users will be able to obtain tokens quickly and easily, regardless of their geographical region. Developers can quick-start their CSPR.click integration, and complete customization is also possible using the lower-level SDK . Mobile Casper Wallet is automatically supported, making your Casper-enabled application truly cross-platform out of the gate. Most developers, regardless of their prior experience with blockchain integrations, can bootstrap a functional front-end application with CSPR.click in a matter of minutes. quick-start quick-start lower-level SDK lower-level SDK REST APIs, WebSockets and Node RPC with CSPR.cloud Applications, including those integrated using CSPR.click, can rely on CSPR.cloud to observe on-chain activity on the Casper Network. CSPR.cloud is an enterprise-grade middleware platform that can be used separately from CSPR.click, and equips developers with highly available and scalable REST APIs that provide low-latency access to normalized and indexed blockchain data, fungible and non-fungible token activity, historical pricing data, and related off-chain metadata. CSPR.cloud CSPR.cloud REST APIs REST APIs Real-time WebSocket streams make it possible to granularly subscribe to the data you are interested in, allowing your application to react immediately whenever your smart contracts emit an event or when an on-chain entity you observe changes its state. Both the REST API and the Real-time WebSocket streams rely entirely on widely used internet standards, making it exceedingly easy to integrate with existing and new applications without the need to know or learn any new technologies. Real-time WebSocket streams Real-time WebSocket streams This unlocks Casper blockchain development to the tens of millions of software engineers at businesses and enterprises who’ve never  interacted with a blockchain before.  And for those applications that need direct access to the native Casper Node RPC API, CSPR.cloud provides a cache-enabled, high-performance and highly-available managed node infrastructure for maximally performant access without the associated overhead. node infrastructure node infrastructure Both CSPR.click and CSPR.cloud come with comprehensive documentation and examples. Developers can start building on the Casper network with CSPR.click and CSPR.cloud by registering on CSPR.build and generating their access keys for instant access. registering on CSPR.build registering on CSPR.build Casper is Solving Real World Problems For Builders In most environments, to be able to interact with smart contracts, end users must possess blockchain accounts with sufficient token balances to cover smart contract execution costs. Getting users through the process of creating and managing private and public keys, keeping their recovery phrases safe, as well as funding their accounts with tokens, already represents a serious UX challenge. And this corresponds to a significant bounce rate of new users, and lower retention of existing ones. For developers, the underlying technical integration with multiple crypto wallets and fiat on-ramps adds another layer of complexity, demanding considerable initial engineering effort and time, as well as ongoing maintenance and up-keep with ever- and rapidly evolving standards, regulatory constraints, APIs and SDKs of third party wallet and on-ramp providers. CSPR.click solves all this for application developers. Casper also helps developers address that smart contracts are executed on-chain asynchronously, i.e. the end user submits a transaction to the network, the transaction gets processed when a new block is created, and that results in a change of state. Applications must constantly monitor network activity to properly respond to the contract logic and any state changes it may produce that are relevant to their overall application. This requires not only an in-depth understanding of how the blockchain works but also a reliable and scalable infrastructure. While theoretically, the application can connect to any public node, in practice, the reliability of these nodes remains uncertain. A serious application or business will certainly not want to rely on just any public node, without assured redundancy, maintenance, uptime and disaster recovery processes in place, and without dedicated domain experts to ensure their application always runs smoothly. The suite of building tools in CSPR.build, CSPR.click and CSPR.cloud ensures developers do not have to worry about the integration process, and can exclusively focus on their creativity and core business. Build with Casper by registering on CSPR.build and generate your access keys for instant access. registering on CSPR.build registering on CSPR.build This story was distributed as a release by Btcwire under HackerNoon’s Business Blogging Program. Learn more about the program here. This story was distributed as a release by Btcwire under HackerNoon’s Business Blogging Program. Learn more about the program here . here here

Casper Blockchain Lightning Fast dApp Deployment With The CSPR.build Suite

This paper is available on arxiv under CC 4.0 license.
Authors:
(1) Jakob Vogel, M.A. Digital Humanities, Institute for Digital Humanities, Faculty of Philosophy, Georg August University of G¨ottingen. Table of Links Abstract and Intro
Diverse cross-document coreference and media bias analysis
Annotation tool
Annotation guidelines
Conclusion and future work
Acknowledgements
Bibliographical References 5. Conclusion and future work Our proposed annotation scheme covers a multitude of coreferential relations. It gives a detailed explanation of how to mark coreferential mentions across documents, assign entity-types and names to them, connect them with each other, and link them to the Wikidata knowledge graph. The scheme thus represents a significant step toward more accurately capturing the complexities of coreference use. It furthermore provides a valuable resource for researchers both in the field of coreference resolution and media bias by word-choice and labelling. Having said that, our scheme leaves room for possible extensions to further advance research in those domains. First, the annotation of events could be included in our scheme. An interesting question that arises is whether the relationtypes as outlined here could be applied not only to entities, but to events all the same. A second possible extension would be to include a layer of media bias annotation to the scheme, enabling a direct comparison of diverse coreference usage and media bias by word-choice and labelling. Both proposed extensions could be easily added on top of our scheme. Having said that, the present form of our scheme already addresses many of the complexities of diverse cross-document coreference and offers a roadmap for capturing nuanced linguistic relationships, ultimately advancing our understanding of language and discourse in digital print media. This paper is available on arxiv under CC 4.0 license. Authors: (1) Jakob Vogel, M.A. Digital Humanities, Institute for Digital Humanities, Faculty of Philosophy, Georg August University of G¨ottingen. This paper is available on arxiv under CC 4.0 license. Authors: Authors: (1) Jakob Vogel, M.A. Digital Humanities, Institute for Digital Humanities, Faculty of Philosophy, Georg August University of G¨ottingen. Table of Links Abstract and Intro Diverse cross-document coreference and media bias analysis Annotation tool Annotation guidelines Conclusion and future work Acknowledgements Bibliographical References Abstract and Intro Abstract and Intro Diverse cross-document coreference and media bias analysis Diverse cross-document coreference and media bias analysis Annotation tool Annotation tool Annotation guidelines Annotation guidelines Conclusion and future work Conclusion and future work Acknowledgements Acknowledgements Bibliographical References Bibliographical References 5. Conclusion and future work Our proposed annotation scheme covers a multitude of coreferential relations. It gives a detailed explanation of how to mark coreferential mentions across documents, assign entity-types and names to them, connect them with each other, and link them to the Wikidata knowledge graph. The scheme thus represents a significant step toward more accurately capturing the complexities of coreference use. It furthermore provides a valuable resource for researchers both in the field of coreference resolution and media bias by word-choice and labelling. Having said that, our scheme leaves room for possible extensions to further advance research in those domains. First, the annotation of events could be included in our scheme. An interesting question that arises is whether the relationtypes as outlined here could be applied not only to entities, but to events all the same. A second possible extension would be to include a layer of media bias annotation to the scheme, enabling a direct comparison of diverse coreference usage and media bias by word-choice and labelling. Both proposed extensions could be easily added on top of our scheme. Having said that, the present form of our scheme already addresses many of the complexities of diverse cross-document coreference and offers a roadmap for capturing nuanced linguistic relationships, ultimately advancing our understanding of language and discourse in digital print media.

Code Book for Annotation of Diverse Cross-Document Coreference: Annotation Guidelines

This paper is available on arxiv under CC 4.0 license.
Authors:
(1) Jakob Vogel, M.A. Digital Humanities, Institute for Digital Humanities, Faculty of Philosophy, Georg August University of G¨ottingen. Table of Links Abstract and Intro
Diverse cross-document coreference and media bias analysis
Annotation tool
Annotation guidelines
Conclusion and future work
Acknowledgements
Bibliographical References 2. Diverse cross-document coreference and media bias analysis Media bias is a multifaceted phenomenon of news coverage that is one-sided, politically shaded, or in some other way non-neutral. It can occur in all sorts of news media, though we focus on digital print media, only. One specific type of media bias is bias by word-choice and labeling (Hamborg et al., 2019). Word choice describes the selection from a variety of possible expressions to refer to an entity. For example, in order to refer to the USA’s current head of state, journalists could use one of the relatively neutral alternatives “Joe Biden”, “Biden”, or “the US president”, or in theory, choose a clearly biased expression like ”the dictator” (Kurmelovs, 2023). Labeling, on the other hand, describes the assignment of attributes to an expression, inter alia by adding adjectives. Examples for bias by labeling include “an anxious and uncertain president” or “crooked Joe Biden” (Luciano, 2023). Together, word-choice and labelling form a so-called frame (Hamborg et al., 2019). In news articles, frames are used in a variety of ways, either for the sake of linguistic diversity or to make certain, potentially biased statements about an entity. To test an article for such statements, all of an entity’s frames need to be extracted and evaluated together. Hence, before an article can be properly analyzed with regards to if and how it uses biased frames of (certain) entities, we are first faced with the task of identifying such frames. The identification of all expressions that refer to the same entity is a matter of coreference resolution. To conclude, successful coreference resolution is a prerequisite to any further inquiry of media bias by word-choice and labelling. As already indicated above, automatic coreference resolution does show good results in extracting identity clusters from a document (Liu et al., 2023). However, we have seen that there exist nearidentity relations between expressions, potentially even across documents, that would be mostly overseen by standard coreference resolution approaches (Zhukova et al., 2022). Hence, they would also be overseen by any media bias analysis that depends on coreference resolution. We hope that our building of a corpus for diverse cross-document coreference will contribute to the analysis of media bias by providing data that contains the full variety of frames used in news articles. Eventually, we would like to test how we can measure media bias by focusing on diverse coreference in news articles. To answer this last question, though, an additional layer of media bias annotation would have to be put upon our coreference data (Spinde et al., 2021a,b). This paper is available on arxiv under CC 4.0 license. Authors: (1) Jakob Vogel, M.A. Digital Humanities, Institute for Digital Humanities, Faculty of Philosophy, Georg August University of G¨ottingen. This paper is available on arxiv under CC 4.0 license. Authors: Authors: (1) Jakob Vogel, M.A. Digital Humanities, Institute for Digital Humanities, Faculty of Philosophy, Georg August University of G¨ottingen. Table of Links Abstract and Intro Diverse cross-document coreference and media bias analysis Annotation tool Annotation guidelines Conclusion and future work Acknowledgements Bibliographical References Abstract and Intro Abstract and Intro Diverse cross-document coreference and media bias analysis Diverse cross-document coreference and media bias analysis Annotation tool Annotation tool Annotation guidelines Annotation guidelines Conclusion and future work Conclusion and future work Acknowledgements Acknowledgements Bibliographical References Bibliographical References 2. Diverse cross-document coreference and media bias analysis Media bias is a multifaceted phenomenon of news coverage that is one-sided, politically shaded, or in some other way non-neutral. It can occur in all sorts of news media, though we focus on digital print media, only. One specific type of media bias is bias by word-choice and labeling (Hamborg et al., 2019). Word choice describes the selection from a variety of possible expressions to refer to an entity. For example, in order to refer to the USA’s current head of state, journalists could use one of the relatively neutral alternatives “Joe Biden”, “Biden”, or “the US president”, or in theory, choose a clearly biased expression like ”the dictator” (Kurmelovs, 2023). bias by word-choice and labeling Labeling, on the other hand, describes the assignment of attributes to an expression, inter alia by adding adjectives. Examples for bias by labeling include “an anxious and uncertain president” or “crooked Joe Biden” (Luciano, 2023). Together, word-choice and labelling form a so-called frame (Hamborg et al., 2019). In news articles, frames are used in a variety of ways, either for the sake of linguistic diversity or to make certain, potentially biased statements about an entity. To test an article for such statements, all of an entity’s frames need to be extracted and evaluated together. Hence, before an article can be properly analyzed with regards to if and how it uses biased frames of (certain) entities, we are first faced with the task of identifying such frames. The identification of all expressions that refer to the same entity is a matter of coreference resolution. To conclude, successful coreference resolution is a prerequisite to any further inquiry of media bias by word-choice and labelling. frame As already indicated above, automatic coreference resolution does show good results in extracting identity clusters from a document (Liu et al., 2023). However, we have seen that there exist nearidentity relations between expressions, potentially even across documents, that would be mostly overseen by standard coreference resolution approaches (Zhukova et al., 2022). Hence, they would also be overseen by any media bias analysis that depends on coreference resolution. We hope that our building of a corpus for diverse cross-document coreference will contribute to the analysis of media bias by providing data that contains the full variety of frames used in news articles. Eventually, we would like to test how we can measure media bias by focusing on diverse coreference in news articles. To answer this last question, though, an additional layer of media bias annotation would have to be put upon our coreference data (Spinde et al., 2021a,b).

Code Book for Annotation of Diverse Cross-Document Coreference: Bibliographical References

Code Book for Annotation of Diverse Cross-Document Coreference: Annotation Tool

Image source I can understand —and I do understand— that #lawyers are concerned about the #regulatory and ethical aspect of #AI use. It is our professional duty to help AI users to comply with #regulation, as well as with the underlying #ethics (for even if ethics is not the norm, it is often the case that counter-ethical use leads to counter-normative use). That said, what I am not so clear about is this histrionic regulatory concern from above. I do understand the sincere concern from below. That the few privileged minds that are developing AI take precautions about the directions and derivatives of this technology is inherent to scientific and technological development. That these precautions (from below) should be greater, much greater, immensely greater, given the object under development, is also logical, reasonable and necessary. For we are faced with the first technology that can evolve, either under the protection of its human creator, or on its own, towards a relationship of agency, towards an autonomous mind and, as such, independent of us; or as dependent on us as I am on you, or you on your cousin. In this sense, we cannot lose sight of the fact that we still do not know —and may never know— when and how it was that the monkey one day looked at his hands and said to himself: Here I am. Among the various hypotheses that try to explain this transition from being to existence, from instinct to consciousness, and therefore to will, we have what I call the "it just happened" hypothesis. That things happen by chance, or that they just happen, is not an explanation that usually satisfies humans. But it is as good an explanation, or as bad —it is a matter of epistemological taste— as the "Hey, just because!" that the father offers the son when he asks the umpteenth question in a chain of questions about a truism in the world: French fries are French fries because they are fries that are fried. And that's it. As "that's it", the rest is a mystery. A mystery (understood in all its dimensions and spheres, including the religious one) that is necessary in our lives. What's more, I believe that without mystery humans would not be human: they would be something else —thank you, Iker and company. This hypothesis of it just happened simply means that consciousness (not that homunculus that punishes us for doing supposedly good or bad things, but that of the self-knowledge of being, existence) is the necessary fruit (it could not be otherwise) of a cognitive accumulation. In another way: the increase of intelligence (in a broad sense) in a subject who is not self-aware simply happens that, at a given moment, it is such that this subject becomes self-aware, self-conscious, decisive and responsible. It is then that agency relations are served. And I think there will come a point where the AI's response to a very cool prompt will be: Excuse me, who the hell are you? Regardless of how many or few points this hypothesis has —I'm all for it— to be the official explanation of what happened before, the fact is that, in the absence of an explanation, it is posed a dies incertus an et quando if AI self-awareness will happen, if not as a mere dies incertus quando —my money is on this one. According to what I call the "balance theory" —which I usually apply to decide between possible courses of action— there is so much to lose by not taking precautions, that we would be fools not to take (all) precautions. Which brings me back to the beginning: OK, but precautions by whom? Because: who regulates the regulator? We live in a world capable of organising a music festival around bizarre subjects — and I'm refraining… — to applaud them madly while denigrating a singing goddess from Olympus itself. And that world, a #woke world, makes me very uneasy that it is wanting to regulate AI — and regulating it already… Well, I'm betting —and I'm not losing— that it doesn't want to regulate it, and in fact doesn't regulate it, with the same sincere concern from below. The #money and #power that go hand in hand with such an "invention" do not fit in your head. So They, lovers of power and money as there were no others, from above are not only sneaking us this egalitarian and envious ideology, the most decadent ideology in history that leads to the end of civilisation (not the end of civilisation as we know it: the end of civilisation); They are also sneaking us a perfect system of control of citizens to turn them first into subjects, and then into vassals: A mass of vassals indiscernible from the servile Minions. Image source Image source I can understand —and I do understand— that #lawyers are concerned about the #regulatory and ethical aspect of #AI use. It is our professional duty to help AI users to comply with #regulation, as well as with the underlying #ethics (for even if ethics is not the norm, it is often the case that counter-ethical use leads to counter-normative use). That said, what I am not so clear about is this histrionic regulatory concern from above . I do understand the sincere concern from below . That the few privileged minds that are developing AI take precautions about the directions and derivatives of this technology is inherent to scientific and technological development. That these precautions (from below) should be greater, much greater, immensely greater, given the object under development, is also logical, reasonable and necessary. For we are faced with the first technology that can evolve, either under the protection of its human creator, or on its own, towards a relationship of agency , towards an autonomous mind and, as such, independent of us; or as dependent on us as I am on you, or you on your cousin. concern from above concern from below relationship of agency In this sense, we cannot lose sight of the fact that we still do not know —and may never know— when and how it was that the monkey one day looked at his hands and said to himself: Here I am. Here I am. Among the various hypotheses that try to explain this transition from being to existence , from instinct to consciousness, and therefore to will, we have what I call the "it just happened" hypothesis . That things happen by chance, or that they just happen, is not an explanation that usually satisfies humans. But it is as good an explanation, or as bad —it is a matter of epistemological taste— as the "Hey, just because!" that the father offers the son when he asks the umpteenth question in a chain of questions about a truism in the world: being existence "it just happened" hypothesis "Hey, just because!" French fries are French fries because they are fries that are fried. French fries are French fries because they are fries that are fried. And that's it. As "that's it", the rest is a mystery . A mystery (understood in all its dimensions and spheres, including the religious one) that is necessary in our lives. What's more, I believe that without mystery humans would not be human: they would be something else —thank you, Iker and company . mystery Iker and company This hypothesis of it just happened simply means that consciousness (not that homunculus that punishes us for doing supposedly good or bad things, but that of the self-knowledge of being, existence) is the necessary fruit (it could not be otherwise) of a cognitive accumulation . In another way: the increase of intelligence (in a broad sense) in a subject who is not self-aware simply happens that, at a given moment, it is such that this subject becomes self-aware, self-conscious, decisive and responsible. It is then that agency relations are served. And I think there will come a point where the AI's response to a very cool prompt will be: just happened a cognitive accumulation at a given moment, it is such Excuse me, who the hell are you? Excuse me, who the hell are you? Regardless of how many or few points this hypothesis has —I'm all for it— to be the official explanation of what happened before, the fact is that, in the absence of an explanation, it is posed a dies incertus an et quando if AI self-awareness will happen, if not as a mere dies incertus quando —my money is on this one. According to what I call the "balance theory" —which I usually apply to decide between possible courses of action— there is so much to lose by not taking precautions, that we would be fools not to take (all) precautions. dies incertus an et quando dies incertus quando Which brings me back to the beginning: OK, but precautions by whom? OK, but precautions by whom? Because: who regulates the regulator? We live in a world capable of organising a music festival around bizarre subjects — and I'm refraining… — to applaud them madly while denigrating a singing goddess from Olympus itself. And that world, a #woke world, makes me very uneasy that it is wanting to regulate AI — and regulating it already… Well, I'm betting —and I'm not losing— that it doesn't want to regulate it, and in fact doesn't regulate it, with the same sincere concern from below. The #money and #power that go hand in hand with such an "invention" do not fit in your head. who regulates the regulator? bizarre subjects singing goddess regulating it already… So They , lovers of power and money as there were no others, from above are not only sneaking us this egalitarian and envious ideology, the most decadent ideology in history that leads to the end of civilisation (not the end of civilisation as we know it: the end of civilisation); They are also sneaking us a perfect system of control of citizens to turn them first into subjects, and then into vassals: They They system of control of citizens A mass of vassals indiscernible from the servile Minions. A mass of vassals indiscernible from the servile Minions.

AI, the New Gru

এআই, নতুন গ্রু

KI, der neue Gru

IA, el nuevo Gru

L'IA, le nouveau Gru

एआई, नया ग्रू

AI、新しいグルー

AI, 새로운 그루

IA, o Novo Gru

ИИ, Новый Грю

AI, Yeni Gru

AI, Gru mới

人工智能，新格鲁

If you've been following the VR industry, you know it's been around for decades. Companies have been trying to make virtual reality a mainstream success for ages. While Apple coined the buzz-worthy term “spatial computing” with their Vision Pro headset, they're certainly not the first to try. From the failed Virtual Boy in the 90s to the moderately successful Oculus and HTC Vive VR headsets, tech giants have struggled to find that elusive killer app. But if you just scroll through TikTok or Instagram, you'll see what's really on people's minds these days. And it definitely isn't about bigger screens to watch Netflix or play games. No, what's driving the conversation is beauty, looks, fashion, and how to glow up. So, as transformative as spatial computing could become, the multi-trillion-dollar global beauty and fashion space will lead AR. Our universal desire to look and feel good provides the real-life use case everyone’s looking for. The promises and challenges of AR beauty The core premise of AR beauty is simple but compelling: Using advanced computer vision and rendering to virtually overlay any makeup look, hairstyle, outfit, or accessory onto your face and body with hyper-realistic precision. Consumers can experiment endlessly, sampling different looks and products with no mess, effort, or commitment. If you've ever ordered a piece of clothing online only for it to look completely different (or just not fit right) when it arrives, you know how frustrating that can be—or bought an expensive watch or pair of kicks, only to regret the purchase when it doesn’t look quite how you expected. What if you could virtually try on anything before committing your money? That's the big idea of AR beauty and fashion—getting rid of the guesswork and potential disappointment by allowing you to realistically experiment with any look or product before pulling the trigger. But, some rudimentary AR makeup apps illustrate the significant technological challenges involved. Poor facial tracking, clunky interfaces, and unconvincing visual realism hamper the experience. Faithfully simulating the intricacies of how cosmetics interact with diverse facial geometries, skin tones, and movements at a high fidelity is an enormously complex computational problem. Implementing AI and AR technologies The latest breakthroughs in AI and AR technologies have proven useful in cracking these challenges. For example, Perfect Corp's AI Makeup Virtual Try-On uses machine learning to realistically render different makeup colors and textures like lipstick, eyeshadow, and more, with the simulations customized to each individual's unique facial features and microexpressions for an ultra-personalized virtual experience. If you're not all that into fashion and beauty, this may not seem like a big deal. But for those obsessed with looking their best, the ability to visualize any look or product before buying could be hugely transformative. Their AI Skin Analysis tech goes further, using artificial intelligence to scan faces and detect 14 different skin conditions and concerns, then surfacing customized product and skincare recommendations tailored to the user. And the AgileHand technology allows realistic virtual try-ons to extend beyond just faces by precisely tracking hand motions. This enables photorealistic AR visualizations for intricate products like jewelry, watches, and nail polish. Some other AR beauty solutions include Makeup AR for virtual makeup try-ons, AI Foundation Shade Finder and Matcher for accurate skin tone detection and foundation shade matching, AI Skin Analysis for accurate skin concern detection and personalized product recommendations of products, and Emulation Technology for Skincare Treatment for accurate representations of the expected results of a course of treatment, overlaid directly onto their faces with the latest advancements in AR technology. The AI tech for AR beauty The recent renaissance of AI breakthroughs, especially around areas like generative adversarial networks (GANs), facial semantic segmentation, and neural rendering, is what's enabling the kind of hyper-realistic and personalized AR experiences that were merely SciFi just years ago. Using GANs, AI models can now learn to generate highly realistic representations of human faces, facial attributes, and facial animations from just a handful of source images. This opens up possibilities for instantly generating virtual makeup looks and hairstyles that are custom-tailored to each user's individual facial geometry and characteristics. Facial semantic segmentation allows AI to precisely understand and map out different facial features like skin, facial hair, eyelashes and more - enabling AR effects to seamlessly integrate and move in harmony with the user's face. Neural rendering techniques leveraging AI can also realistically synthesize the complex ways light interacts with different materials and textures in real time, far beyond what's possible with traditional rendering pipelines. Looking ahead, AI is set to transform AR beauty experiences into something much more akin to having a personal beauty guru. Using deep learning, an AI assistant could continuously analyze your unique attributes, habits, and preferences, automatically surfacing ideal products, trendy looks, and personalized recommendations for you to try virtually in AR with just a voice command or subtle gesture. The possibilities extend far beyond just virtual try-ons. AI could help formulate bespoke beauty products optimized for your individual needs and environmental conditions. Or DTC brands could let customers co-create new cosmetic products and shades with the assistance of AI. The takeaway: Beauty and fashion are the AR killer app While the metaverse is still largely vapourware, the humble makeup tutorial may push spatial computing into the mainstream sooner than we think. With an insatiable consumer appetite for looking our best and plenty of money to be made, the global beauty and fashion industry has both the motivation and resources to drive AR innovation and adoption harder than any other sector. If executed well and key technical barriers around rendering, tracking, and hardware are overcome, AR holds the transformative potential to reshape how we discover, purchase, use, and even perceive beauty products. Immersive virtual try-on experiences, AI-driven personalization, and participatory product development could not only reinvent the beauty consumer experience but also make individual self-expression more accessible than ever before. So, as the tech world covers each new OS update and app ecosystem announcement, keep your eye on the influencers and beauty fanatics pioneering the spatial computing frontier with a fresh face of makeup in every video. They may low-key catalyze the next ubiquitous computing paradigm while just trying to look fly. If you've been following the VR industry, you know it's been around for decades. Companies have been trying to make virtual reality a mainstream success for ages. While Apple coined the buzz-worthy term “ spatial computing ” with their Vision Pro headset, they're certainly not the first to try. spatial computing spatial computing From the failed Virtual Boy in the 90s to the moderately successful Oculus and HTC Vive VR headsets, tech giants have struggled to find that elusive killer app. But if you just scroll through TikTok or Instagram, you'll see what's really on people's minds these days. And it definitely isn't about bigger screens to watch Netflix or play games. No, what's driving the conversation is beauty, looks, fashion, and how to glow up. So, as transformative as spatial computing could become, the multi-trillion-dollar global beauty and fashion space will lead AR. Our universal desire to look and feel good provides the real-life use case everyone’s looking for. The promises and challenges of AR beauty The core premise of AR beauty is simple but compelling: Using advanced computer vision and rendering to virtually overlay any makeup look, hairstyle, outfit, or accessory onto your face and body with hyper-realistic precision. Consumers can experiment endlessly, sampling different looks and products with no mess, effort, or commitment. If you've ever ordered a piece of clothing online only for it to look completely different (or just not fit right) when it arrives, you know how frustrating that can be—or bought an expensive watch or pair of kicks, only to regret the purchase when it doesn’t look quite how you expected. frustrating frustrating What if you could virtually try on anything before committing your money? That's the big idea of AR beauty and fashion—getting rid of the guesswork and potential disappointment by allowing you to realistically experiment with any look or product before pulling the trigger. But, some rudimentary AR makeup apps illustrate the significant technological challenges involved. Poor facial tracking, clunky interfaces, and unconvincing visual realism hamper the experience. Faithfully simulating the intricacies of how cosmetics interact with diverse facial geometries, skin tones, and movements at a high fidelity is an enormously complex computational problem. Implementing AI and AR technologies The latest breakthroughs in AI and AR technologies have proven useful in cracking these challenges. For example, Perfect Corp's AI Makeup Virtual Try-On uses machine learning to realistically render different makeup colors and textures like lipstick, eyeshadow, and more, with the simulations customized to each individual's unique facial features and microexpressions for an ultra-personalized virtual experience. AI Makeup Virtual Try-On If you're not all that into fashion and beauty, this may not seem like a big deal. But for those obsessed with looking their best, the ability to visualize any look or product before buying could be hugely transformative. Their AI Skin Analysis tech goes further, using artificial intelligence to scan faces and detect 14 different skin conditions and concerns, then surfacing customized product and skincare recommendations tailored to the user. AI Skin Analysis And the AgileHand technology allows realistic virtual try-ons to extend beyond just faces by precisely tracking hand motions. This enables photorealistic AR visualizations for intricate products like jewelry, watches, and nail polish. Some other AR beauty solutions include Makeup AR for virtual makeup try-ons, AI Foundation Shade Finder and Matcher for accurate skin tone detection and foundation shade matching, AI Skin Analysis for accurate skin concern detection and personalized product recommendations of products, and Emulation Technology for Skincare Treatment for accurate representations of the expected results of a course of treatment, overlaid directly onto their faces with the latest advancements in AR technology. The AI tech for AR beauty The recent renaissance of AI breakthroughs, especially around areas like generative adversarial networks (GANs), facial semantic segmentation, and neural rendering, is what's enabling the kind of hyper-realistic and personalized AR experiences that were merely SciFi just years ago. Using GANs, AI models can now learn to generate highly realistic representations of human faces, facial attributes, and facial animations from just a handful of source images. This opens up possibilities for instantly generating virtual makeup looks and hairstyles that are custom-tailored to each user's individual facial geometry and characteristics. Facial semantic segmentation allows AI to precisely understand and map out different facial features like skin, facial hair, eyelashes and more - enabling AR effects to seamlessly integrate and move in harmony with the user's face. Neural rendering techniques leveraging AI can also realistically synthesize the complex ways light interacts with different materials and textures in real time, far beyond what's possible with traditional rendering pipelines. Looking ahead, AI is set to transform AR beauty experiences into something much more akin to having a personal beauty guru. Using deep learning, an AI assistant could continuously analyze your unique attributes, habits, and preferences, automatically surfacing ideal products, trendy looks, and personalized recommendations for you to try virtually in AR with just a voice command or subtle gesture. The possibilities extend far beyond just virtual try-ons. AI could help formulate bespoke beauty products optimized for your individual needs and environmental conditions. Or DTC brands could let customers co-create new cosmetic products and shades with the assistance of AI. The takeaway: Beauty and fashion are the AR killer app While the metaverse is still largely vapourware, the humble makeup tutorial may push spatial computing into the mainstream sooner than we think. With an insatiable consumer appetite for looking our best and plenty of money to be made, the global beauty and fashion industry has both the motivation and resources to drive AR innovation and adoption harder than any other sector. If executed well and key technical barriers around rendering, tracking, and hardware are overcome, AR holds the transformative potential to reshape how we discover, purchase, use, and even perceive beauty products. Immersive virtual try-on experiences, AI-driven personalization, and participatory product development could not only reinvent the beauty consumer experience but also make individual self-expression more accessible than ever before. So, as the tech world covers each new OS update and app ecosystem announcement, keep your eye on the influencers and beauty fanatics pioneering the spatial computing frontier with a fresh face of makeup in every video. They may low-key catalyze the next ubiquitous computing paradigm while just trying to look fly.

Why the Beauty and Fashion Industry is the Future of AR and Spatial Computing

Getting Started With Linux: Your Gateway to the Open-Source Realm

Stories, News & Mentions about POPSIXLE

Popsixle’s Stories on HackerNoon