Tanso

tanso

Text-to-image generation is not a new idea. Notably, the (GAN) architecture, a once-popular deep-learning computer vision algorithm had . Generative Adversarial Network generated birds and flowers from text After more improvements to the generative image algorithms like and , we now have a deluge of commercial text to image generators, ranging from , and . hyperealistic generation of human faces GLIDE diffusion model Google’s Imagen Tiktok’s Greenscreen OpenAI Dall-E And here comes the new kid on the block, Stable Diffusion Before moving ahead, for those who wants to understand the nitty-gritties of diffusion models, I strongly encourage readers to go through this awesome blogpost by Lilian Weng, and . https://lilianweng.github.io/posts/2021-07-11-diffusion-models/ this tweet by Tom Goldstein The rest of the article would be an exercise of using the stable diffusion model to generate images given “ ”. There is no main purpose for doing so, other than satisfying my curiosity. my name Narcissistic as it sounds, I want to generate an image with my name. Be honest, don’t we all from time to time? And thus, I went to and typed in . “google your own name” https://huggingface.co/spaces/stabilityai/stable-diffusion Liling Tan Before I show the generated results, I would like to clarify some specifics about my first and last name. In general, it is more probable that a English romanization of my Chinese character name refers to a female individual than a male. Next (陳) is a common Southern Chinese name that originates from the , and more commonly you would expect the English translation/romanization of Mandarin Chinese, . With K-Pop being a global phenomenon, the top-ranked Google result of would point you to the .  Also, since (my last name), has the same spelling as the color, the top Google search will end up with results pointing to “ ”. Liling Tan Min language Chen Chen Korean Singer from the K-pop band EXO Tan tan a yellowish-brown color For reference, here are the control experiments results, if I googled “liling tan” And now, the results… Okay, that’s definitely nowhere close to how I look. I kind of expected the image to show an Asian female but the first image was kinda weird. Now, what happens if I lowercase and re-run the generation? Alright, the model seems hell-bent on some facial features and sorta generating one older and another younger version of “ ” liling tan Interestingly, the model has some internal mechanisms to block some "unsafe” content and output this error message. This Image was not displayed because our detection model detected . Unsafe content Then I got really curious, do the two versions of “ ” exist IRL (in real life)? liling tan So I do the natural thing to reverse image search with https://lens.google.com/ Image Search Result 1 Image Search Result 2 Hmmm, no results, lets extend the frame to beyond the face… And of course, what was I expecting, surely Google Shopping will take the chance to advertise and sell me something -_- Maybe the older version generated by the model is more grounded to someone IRL? Image Search Result 3 Hmmm, no results, so the model kinda generated two unique people that doesn’t exist IRL? But, we see two separate dots that indicates two other results exists, let’s see the first one at the bottom. Of course, it will try to sell me something again… What was I expecting? @_@ Let’s try the results other dot. Now this is interesting, it’s trying to promote a inspired piece. Girl with a Pearl Earring (ca. 1665) by Johannes Vermeer But what about “Find image source”? Does it really find the source of the images that the model use to slice, dice and “diffuse”? It’s hard to say: Find the image source - Result 1 Find the image source - Result 2 Find the image source - Result 3 Đợi tí (Wait a minute),does that mean that we don’t know which image the model has been trained on or used to splice before generating the results? Now that we find the underbelly from the results, lets backtrack the OG paper listed on the original source code https://github.com/CompVis/stable-diffusion Robin Rombach*, Andreas Blattmann*, Dominik Lorenz, Patrick Esser, Björn Ommer. In CVPR '22. High-Resolution Image Synthesis with Latent Diffusion Models. According to the paper, the model is pre-trained on the LAION database and the Conceptual Captions dataset . https://laion.ai/blog/laion-5b/ https://ai.google.com/research/ConceptualCaptions/ Our model is pretrained on the LAION [73] database and finetuned on the Conceptual Captions [74] dataset. What if, we build a reverse image search based on the datasets? Perhaps if we can find the approximate nearest neighbor images to the outputs generated by the model, then it might be possible to “explain away” which images the model had deemed to be “salient” enough to diffuse and generate the outputs from my name. Voila! Here’s a search engine to probe the dataset used to train Stable Diffusion, https://waxy.org/2022/08/exploring-12-million-of-the-images-used-to-train-stable-diffusions-image-generator/ And here’s the list of . results of searching for Liling in the LAION database Ah ha, that is one image that the model must have “chosen to diffuse” It is highly possible that the model somehow and generated an image similar to that image. “ focused ” an image in the training data with “Liling” in the caption Why did you bother to “ ”? diffuse <your_name> To conclude the post, this exercise is purely out of curiosity. I wanted to know what the model would generate. But this exercise also highlights some sort of bias when using generative models. While “Liling Tan” isn’t to generate anything close to me or any other top-ranked “Liling Tan” search results, the model seems to be more stable in . stable enough generating  famous people, (e.g. John Oliver) To end this article, here’s a result from generating images with my online handle as the prompt… alvations Hopefully, after reading, you would also try to “ ”! diffuse <your_name>

Stable Diffusion, Unstable Me: Text-to-image Generation

Photo of from gwanghwamu pixelbay So yesterday I found this paper “ ” from and implemented . But then something hit me, Res Net and High Way net are built in a way that allows direction connection between the input data X and the transformed data X`. Dilated Recurrent Neural Networks NIPS 2017 here Why can’t we do the exact same for back propagation as well? Connect the gradient from previous layers to deeper layers….. I mean, if you are not using frameworks to perform auto differentiation, why don’t we connect the gradients from latest layer to more deeper layer, and just see how that goes? In this post, we’ll do exactly that, also lets go one step further and compare with a model that applies . Google Brain’s Gradient Noise Since I got the inspiration after reading the Dilated RNN, I’ll just call this Dilated Back Propagation, however if anyone knows other papers where they performed back propagation in this fashion, please let me know in the comment section. Also, I will assume you already have read my Blog post about Implementing Dilated RNN, if not please click here. Network Architecture (Feed Forward Direction) Side View Front View → Final Output of the Network, a (1*10) vector for One hot encoding of predicted number Red Circle → Hidden State 0 for Layer 2 → Hidden States for Layer 2 Brown Circle Lime Circle → Hidden State 0 for Layer 1 → Hidden State for Layer 1 Pink Circle Black Circle → Input for each Time Stamp (Please of this since I am going to use this knowledge to explain the Training / Test Data) Blue Numbers 1,2,3,4,5 TAKE NOTE → Direction of Feed Forward Operation, Pinkish? Arrow As seen above, the network architecture is exactly the same as the previous post. However there is one thing I changed and that is the input data for each time stamp. Training Data / Test Data → Input at Time Stamp 1 (Vectorized 14*14 Pixel Image) → Input at Time Stamp 2 (Vectorized 14*14 Pixel Image) → Input at Time Stamp 3 (Vectorized 14*14 Pixel Image) → Input at Time Stamp 4 (Vectorized 14*14 Pixel Image) → Input at Time Stamp 5 (Vectorized 14*14 Pixel Image) Pink Box Yellow Box Blue Box Purple Box Green Box Despite some images looking bigger then other, all of them are (14*14) pixel images. And each of the image are made by applying different kind of pooling operation to the original image which is (28*28) pixel image. Each of the pooling operations are described below. → Mean Pooling Using function → Variance Pooling Using function → Max Pooling Using function → Standard Deviation Pooling Using function → Median Pooling Using function Pink Box np.mean Yellow Box np.var Blue Box np.max Purple Box np.std Green Box np.median Below is the code of achieving this. And with that in mind, lets take a look at other training data. Finally, the reason I did this, is simply put, I wanted to. Image of 0 Image of 3 Case 1: Normal Back Propagation → Standard Direction of Gradient Flow Purple Arrow Above is the normal (or standard) back propagation, where we compute the gradient in each layer, pass them on to the next layer. And each of weights at different time stamps uses them to update their weights and the gradient flow continuous on. Case 2: Google Brain Gradient Noise → Standard Direction of Gradient Flow → Added Gradient Noise Purple Arrow Yellow Arrow Again, the purple arrow represent standard gradient flow, however this time before updating each weights we are going to add some Noise to the Gradient. Below is the screen shot of how we can achieve this. Case 3: Dilated Back Propagation → Standard Direction of Gradient Flow → Dilated Back Propagation where we pass on some portion of the gradient to the previous layers, which are not directly connected. Purple Arrow Black Arrows Now, here we introduce our new theory in hopes to improve the accuracy of the model. There are two things to note here. We are only going to pass on some portion of the gradient to the previous layers. As seen above, we have some variable called ‘decay proportion rate’ and we are going to use to decrease the amount of gradient it can pass on to the previous layers as time goes on. As seen in the Green Box, since we multiply the gradients from future layers with decay proportion rate, as training continuous the amount of Dilated Gradient flow decreases. inverse time decay rate 2. The Dilated Gradient Flow skips every 2 Layers. As seen above in the Red Box, the gradient at time stamp 5, ONLY goes to the gradient at time stamp 3. However this architecture can be further explored to make the gradient flow much more denser. Case 4: Dilated Back Propagation + Google Brain Gradient Noise → Standard Direction of Gradient Flow → Dilated Back Propagation where we pass on some portion of the gradient to the previous layers, which are not directly connected. → Added Gradient Noise Purple Arrow Black Arrows Yellow Arrow Here we are not only adding Gradient Noise to each weight update, but also making the Gradient Flow better. Training and Results (Google Colab, Local Setting) Above are results when running the code on Google Colab. The accuracy bar represents model’s correct guesses for 100 test images. Unfortunately I forgot to print out the exact accuracy rate but we can see that Case 2 (Google Brain Gradient Noise) had the highest accuracy. Also, cases with non standard back propagation performed better than standard back propagation. In the cost over time function, we can see that standard back propagation had the highest cost rate. Above are results when running the code on my local laptop. The accuracy bar represents model’s correct guesses for 100 test images. Here it was interesting to see Case 3(Dilated Back Propagation) under performing when compared to standard back propagation. However combination of Dilated Back propagation and Google Brain’s Gradient noise have out performed every model. Interactive Code I moved to Google Colab for Interactive codes! So you would need a google account to view the codes, also you can’t run read only scripts in Google Colab so make a copy on your play ground. Finally, I will never ask for permission to access your files on Google Drive, just FYI. Happy Coding! Please click here to access the interactive code. Final Words I love frameworks such as Tensorflow, and Keras. However, I strongly believe we need to explore more different ways to perform back propagation. If any errors are found, please email me at jae.duk.seo@gmail.com, if you wish to see the list of all of my writing please . view my website here Meanwhile follow me on my twitter , and visit , or my for more content. I also did comparison of Decoupled Neural Network are interested. here my website Youtube channel here if you Reference Chang, S., Zhang, Y., Han, W., Yu, M., Guo, X., Tan, W., … & Huang, T. S. (2017). Dilated recurrent neural networks. In (pp. 76–86). Advances in Neural Information Processing Systems Neelakantan, A., Vilnis, L., Le, Q. V., Sutskever, I., Kaiser, L., Kurach, K., & Martens, J. (2015). Adding gradient noise improves learning for very deep networks. . arXiv preprint arXiv:1511.06807 Seo, J. D. (2018, February 14). Only Numpy: NIPS 2017 — Implementing Dilated Recurrent Neural Networks with Interactive Code. Retrieved February 15, 2018, from https://towardsdatascience.com/only-numpy-nips-2017-implementing-dilated-recurrent-neural-networks-with-interactive-code-e83abe8c9b27 Index. (n.d.). Retrieved February 15, 2018, from https://docs.scipy.org/doc/numpy/genindex.html [1]”tf.train.inverse_time_decay | TensorFlow”, , 2018. [Online]. Available: [Accessed: 16- Feb- 2018]. TensorFlow https://www.tensorflow.org/api_docs/python/tf/train/inverse_time_decay.

Only Numpy: Dilated Back Propagation and Google Brain’s Gradient Noise with Interactive Code

How to Build an AI-Powered Web Application With Next.js and TensorFlow.js: A Guide

Integrating Large Language Models (LLMs) into mobile apps is becoming increasingly important as AI advances. LLMs can significantly enhance features like chatbots, language translation, and personalized content. However, deploying these models on Android comes with challenges, such as limited resources and processing power. This guide will walk you through how to effectively deploy LLMs on Android using TensorFlow Lite, covering everything from setting up to implementing a chatbot. Setting Up TensorFlow Lite for LLMs 1. Adding TensorFlow Lite to Your Android Project First, include TensorFlow Lite in your Android project by adding the following dependencies to your build.gradle file: dependencies {
    implementation 'org.tensorflow:tensorflow-lite:2.7.0'
    implementation 'org.tensorflow:tensorflow-lite-support:0.3.0'
} 2. Loading the Model Load your pre-trained LLM model into your app. Here’s an example code snippet: import org.tensorflow.lite.Interpreter;
import java.nio.MappedByteBuffer;
import java.nio.channels.FileChannel;
import android.content.res.AssetFileDescriptor;

public class LLMActivity extends AppCompatActivity {
    private Interpreter interpreter;

    private MappedByteBuffer loadModelFile() throws IOException {
        AssetFileDescriptor fileDescriptor = this.getAssets().openFd("model.tflite");
        FileInputStream inputStream = new FileInputStream(fileDescriptor.getFileDescriptor());
        FileChannel fileChannel = inputStream.getChannel();
        long startOffset = fileDescriptor.getStartOffset();
        long declaredLength = fileDescriptor.getDeclaredLength();
        return fileChannel.map(FileChannel.MapMode.READ_ONLY, startOffset, declaredLength);
    }

    private void initializeInterpreter() {
        try {
            interpreter = new Interpreter(loadModelFile());
        } catch (IOException e) {
            e.printStackTrace();
        }
    }
} 3. Running Inference Run inference by passing input data and processing the output. Here’s an example: public String generateText(String inputText) {
    float[][] input = preprocessInput(inputText);  // Tokenize and process input
    float[][] output = new float[1][outputLength]; // Define the output array

    interpreter.run(input, output); // Run inference to generate a response

    return postprocessOutput(output); // Convert model output to text
} Optimizing the Model Optimizing LLMs is crucial for performance on mobile devices. Use TensorFlow Lite's Model Optimization Toolkit to reduce model size and improve speed, such as by applying quantization: import tensorflow as tf

converter = tf.lite.TFLiteConverter.from_saved_model('path_to_your_model')
converter.optimizations = [tf.lite.Optimize.DEFAULT]
quantized_model = converter.convert()

with open('quantized_model.tflite', 'wb') as f:
    f.write(quantized_model) Use Case: Implementing a Chatbot with LLMs on Android Introduction LLMs are ideal for creating chatbots, providing intelligent, real-time responses that enhance customer interaction. This section will guide you through building an AI-powered chatbot on Android, focusing on integrating an LLM, optimizing it for mobile, and effectively deploying it. 1. Setting Up the Environment Add TensorFlow Lite: As demonstrated above, include TensorFlow Lite in your project.


Prepare the Model: Choose a pre-trained conversational LLM optimized for mobile and convert it to TensorFlow Lite format. 2. Loading and Running the Model Load the Model: Use the loadModelFile function shown earlier to load your chatbot model.


Run Inference: Implement a generateResponse function to process user input and generate a response. 3. Building the Chatbot Interface Design the UI: Create a simple UI with a text input field, send button, and chat history window.
Handle User Input: Capture user input, pass it to the LLM, and display the response in the chat history. history. java Button sendButton = findViewById(R.id.sendButton);
EditText inputField = findViewById(R.id.inputField);
TextView chatHistory = findViewById(R.id.chatHistory);

sendButton.setOnClickListener(new View.OnClickListener() {
@Override
public void onClick(View v) {
String userInput = inputField.getText().toString();
String response = generateResponse(userInput);
chatHistory.append("You: " + userInput +");
chatHistory.append("Bot: " + response +");
}
}); Testing and Deployment Testing: Thoroughly test the chatbot under different conditions, including varying network speeds and device specs.
Deployment: Prepare the chatbot for deployment, possibly through staged releases to gather feedback and make improvements. Challenges and Best Practices Deploying LLMs on Android presents challenges: Memory Constraints: Optimize models to fit within device memory limits.
Latency: Maintain low latency to ensure a smooth user experience.
Battery Consumption: Reduce battery usage through optimizations like quantization. Conclusion Integrating LLMs into Android apps can significantly enhance user experiences by adding advanced AI-driven features like chatbots. Following the steps outlined and leveraging tools like TensorFlow Lite will help you efficiently deploy these powerful models on mobile devices. As AI evolves, mastering these techniques will be crucial for staying competitive in mobile app development. References TensorFlow Lite: https://www.tensorflow.org/lite
TensorFlow Lite Model Optimization: https://www.tensorflow.org/lite/performance/model_optimization
Android Developer Guide: https://developer.android.com
Implementing AI on Mobile: https://ai.googleblog.com/2020/02/implementing-ml-on-mobile.html
TensorFlow Lite for Android: https://www.tensorflow.org/lite/guide/android Integrating Large Language Models (LLMs) into mobile apps is becoming increasingly important as AI advances. LLMs can significantly enhance features like chatbots, language translation, and personalized content. However, deploying these models on Android comes with challenges, such as limited resources and processing power. This guide will walk you through how to effectively deploy LLMs on Android using TensorFlow Lite, covering everything from setting up to implementing a chatbot. TensorFlow Setting Up TensorFlow Lite for LLMs Setting Up TensorFlow Lite for LLMs 1. Adding TensorFlow Lite to Your Android Project First, include TensorFlow Lite in your Android project by adding the following dependencies to your build.gradle file: build.gradle dependencies {
    implementation 'org.tensorflow:tensorflow-lite:2.7.0'
    implementation 'org.tensorflow:tensorflow-lite-support:0.3.0'
} dependencies {
    implementation 'org.tensorflow:tensorflow-lite:2.7.0'
    implementation 'org.tensorflow:tensorflow-lite-support:0.3.0'
} 2. Loading the Model Load your pre-trained LLM model into your app. Here’s an example code snippet: import org.tensorflow.lite.Interpreter;
import java.nio.MappedByteBuffer;
import java.nio.channels.FileChannel;
import android.content.res.AssetFileDescriptor;

public class LLMActivity extends AppCompatActivity {
    private Interpreter interpreter;

    private MappedByteBuffer loadModelFile() throws IOException {
        AssetFileDescriptor fileDescriptor = this.getAssets().openFd("model.tflite");
        FileInputStream inputStream = new FileInputStream(fileDescriptor.getFileDescriptor());
        FileChannel fileChannel = inputStream.getChannel();
        long startOffset = fileDescriptor.getStartOffset();
        long declaredLength = fileDescriptor.getDeclaredLength();
        return fileChannel.map(FileChannel.MapMode.READ_ONLY, startOffset, declaredLength);
    }

    private void initializeInterpreter() {
        try {
            interpreter = new Interpreter(loadModelFile());
        } catch (IOException e) {
            e.printStackTrace();
        }
    }
} import org.tensorflow.lite.Interpreter;
import java.nio.MappedByteBuffer;
import java.nio.channels.FileChannel;
import android.content.res.AssetFileDescriptor;

public class LLMActivity extends AppCompatActivity {
    private Interpreter interpreter;

    private MappedByteBuffer loadModelFile() throws IOException {
        AssetFileDescriptor fileDescriptor = this.getAssets().openFd("model.tflite");
        FileInputStream inputStream = new FileInputStream(fileDescriptor.getFileDescriptor());
        FileChannel fileChannel = inputStream.getChannel();
        long startOffset = fileDescriptor.getStartOffset();
        long declaredLength = fileDescriptor.getDeclaredLength();
        return fileChannel.map(FileChannel.MapMode.READ_ONLY, startOffset, declaredLength);
    }

    private void initializeInterpreter() {
        try {
            interpreter = new Interpreter(loadModelFile());
        } catch (IOException e) {
            e.printStackTrace();
        }
    }
} 3. Running Inference Run inference by passing input data and processing the output. Here’s an example: public String generateText(String inputText) {
    float[][] input = preprocessInput(inputText);  // Tokenize and process input
    float[][] output = new float[1][outputLength]; // Define the output array

    interpreter.run(input, output); // Run inference to generate a response

    return postprocessOutput(output); // Convert model output to text
} public String generateText(String inputText) {
    float[][] input = preprocessInput(inputText);  // Tokenize and process input
    float[][] output = new float[1][outputLength]; // Define the output array

    interpreter.run(input, output); // Run inference to generate a response

    return postprocessOutput(output); // Convert model output to text
} Optimizing the Model Optimizing the Model Optimizing LLMs is crucial for performance on mobile devices. Use TensorFlow Lite's Model Optimization Toolkit to reduce model size and improve speed, such as by applying quantization: import tensorflow as tf

converter = tf.lite.TFLiteConverter.from_saved_model('path_to_your_model')
converter.optimizations = [tf.lite.Optimize.DEFAULT]
quantized_model = converter.convert()

with open('quantized_model.tflite', 'wb') as f:
    f.write(quantized_model) import tensorflow as tf

converter = tf.lite.TFLiteConverter.from_saved_model('path_to_your_model')
converter.optimizations = [tf.lite.Optimize.DEFAULT]
quantized_model = converter.convert()

with open('quantized_model.tflite', 'wb') as f:
    f.write(quantized_model) Use Case: Implementing a Chatbot with LLMs on Android Introduction Introduction LLMs are ideal for creating chatbots, providing intelligent, real-time responses that enhance customer interaction. This section will guide you through building an AI-powered chatbot on Android, focusing on integrating an LLM, optimizing it for mobile, and effectively deploying it. 1. Setting Up the Environment 1. Setting Up the Environment Add TensorFlow Lite: As demonstrated above, include TensorFlow Lite in your project. Prepare the Model: Choose a pre-trained conversational LLM optimized for mobile and convert it to TensorFlow Lite format. Add TensorFlow Lite: As demonstrated above, include TensorFlow Lite in your project. Add TensorFlow Lite : As demonstrated above, include TensorFlow Lite in your project. Add TensorFlow Lite Prepare the Model: Choose a pre-trained conversational LLM optimized for mobile and convert it to TensorFlow Lite format. Prepare the Model : Choose a pre-trained conversational LLM optimized for mobile and convert it to TensorFlow Lite format. Prepare the Model 2. Loading and Running the Model 2. Loading and Running the Model Load the Model: Use the loadModelFile function shown earlier to load your chatbot model. Run Inference: Implement a generateResponse function to process user input and generate a response. Load the Model: Use the loadModelFile function shown earlier to load your chatbot model. Load the Model : Use the loadModelFile function shown earlier to load your chatbot model. Load the Model loadModelFile Run Inference: Implement a generateResponse function to process user input and generate a response. Run Inference : Implement a generateResponse function to process user input and generate a response. Run Inference generateResponse 3. Building the Chatbot Interface 3. Building the Chatbot Interface Design the UI: Create a simple UI with a text input field, send button, and chat history window. Handle User Input: Capture user input, pass it to the LLM, and display the response in the chat history. Design the UI : Create a simple UI with a text input field, send button, and chat history window. Design the UI Handle User Input : Capture user input, pass it to the LLM, and display the response in the chat history. Handle User Input history. java Button sendButton = findViewById(R.id.sendButton);
EditText inputField = findViewById(R.id.inputField);
TextView chatHistory = findViewById(R.id.chatHistory);

sendButton.setOnClickListener(new View.OnClickListener() {
@Override
public void onClick(View v) {
String userInput = inputField.getText().toString();
String response = generateResponse(userInput);
chatHistory.append("You: " + userInput +");
chatHistory.append("Bot: " + response +");
}
}); Button sendButton = findViewById(R.id.sendButton);
EditText inputField = findViewById(R.id.inputField);
TextView chatHistory = findViewById(R.id.chatHistory);

sendButton.setOnClickListener(new View.OnClickListener() {
@Override
public void onClick(View v) {
String userInput = inputField.getText().toString();
String response = generateResponse(userInput);
chatHistory.append("You: " + userInput +");
chatHistory.append("Bot: " + response +");
}
}); Testing and Deployment Testing and Deployment Testing and Deployment Testing: Thoroughly test the chatbot under different conditions, including varying network speeds and device specs. Deployment: Prepare the chatbot for deployment, possibly through staged releases to gather feedback and make improvements. Testing : Thoroughly test the chatbot under different conditions, including varying network speeds and device specs. Testing Deployment : Prepare the chatbot for deployment, possibly through staged releases to gather feedback and make improvements. Deployment Challenges and Best Practices Challenges and Best Practices Deploying LLMs on Android presents challenges: Memory Constraints: Optimize models to fit within device memory limits. Latency: Maintain low latency to ensure a smooth user experience. Battery Consumption: Reduce battery usage through optimizations like quantization. Memory Constraints : Optimize models to fit within device memory limits. Memory Constraints Latency : Maintain low latency to ensure a smooth user experience. Latency Battery Consumption : Reduce battery usage through optimizations like quantization. Battery Consumption Conclusion Conclusion Integrating LLMs into Android apps can significantly enhance user experiences by adding advanced AI-driven features like chatbots. Following the steps outlined and leveraging tools like TensorFlow Lite will help you efficiently deploy these powerful models on mobile devices. As AI evolves, mastering these techniques will be crucial for staying competitive in mobile app development. References References TensorFlow Lite: https://www.tensorflow.org/lite TensorFlow Lite Model Optimization: https://www.tensorflow.org/lite/performance/model_optimization Android Developer Guide: https://developer.android.com Implementing AI on Mobile: https://ai.googleblog.com/2020/02/implementing-ml-on-mobile.html TensorFlow Lite for Android: https://www.tensorflow.org/lite/guide/android TensorFlow Lite: https://www.tensorflow.org/lite TensorFlow Lite Model Optimization: https://www.tensorflow.org/lite/performance/model_optimization Android Developer Guide: https://developer.android.com https://developer.android.com Implementing AI on Mobile: https://ai.googleblog.com/2020/02/implementing-ml-on-mobile.html TensorFlow Lite for Android: https://www.tensorflow.org/lite/guide/android

How to Deploy Large Language Models on Android with TensorFlow Lite

TORTOLA, BVI, July 25th, 2024/Chainwire/--The Tanssi appchain infrastructure protocol has announced a significant collaboration with Grupo Flow, one of Brazil’s largest media ecosystems, boasting over 23 million subscribers across multiple platforms. As Grupo Flow transitions into the blockchain space with the creation of Flow Chain, this collaboration signifies a milestone. Tanssi will facilitate this move, not only bringing Grupo Flow's vast community into the blockchain realm but also connecting them with Polkadot’s vibrant and interoperable ecosystem. This strategic integration ensures a robust, scalable infrastructure that will enhance user engagement, expand Grupo Flow's audience reach, and open new monetization opportunities for content creators within the Flow Chain platform, leveraging both existing community strengths and new blockchain capabilities. "The possibilities and connections with our audience through our own chain are very exciting. We are thrilled to have a trusted partner to guide us on this journey. Now it's time to head to the lab, study the path and roadmap, and look forward to seeing the first features in our ecosystem,” said Andre Gaigher, CCO at Group Flow. Grupo Flow Transitions to Blockchain, Setting a New Standard for the Digital Media Industry Grupo Flow is a leading digital media ecosystem in Brazil, known for its vast reach and influence in the creator economy. With diverse ventures including creative hubs, software development, and audiovisual production, Grupo Flow has solidified its position as a key player in the Brazilian media landscape. Their platforms, such as the highly popular Flow Podcast, attract millions of followers and listeners, collectively achieving over 73 million views per month and amassing more than 23 million followers. As Grupo Flow transitions into the Polkadot blockchain space with Tanssi, they aim to enhance digital engagement, handle large-scale interactions, and seamlessly integrate advanced technology, all while maintaining their commitment to delivering an excellent user experience. This collaboration will help them strengthen their audience and tap into a new community of blockchain users, further enhancing their digital media ecosystem. Tanssi & Grupo Flow: Unlocking Blockchain Potential for Digital Media Transformation Tanssi’s advanced blockchain infrastructure is crucial in facilitating Grupo Flow’s seamless transition to the blockchain space. Grupo Flow will launch its own application blockchain, Flow Chain, offering a dedicated and customizable environment tailored to their needs. This collaboration will enhance digital engagement through tokenized interactions and decentralized governance, allowing users to participate in content decisions, access exclusive events, and earn rewards. Tanssi’s scalable infrastructure will handle high volumes of transactions, ensuring efficient processing and maintaining a seamless user experience for Grupo Flow’s large and active community. By integrating with Polkadot’s interoperable ecosystem, Flow Chain will access a network of interconnected blockchains and leverage technologies like Moonbeam Routed Liquidity to expand into ecosystems such as Ethereum and Avalanche. This ensures robust security, consistent performance, and extensive growth opportunities. Tanssi's infrastructure reduces transaction costs and opens up new revenue streams through staking rewards and tokenized assets. This strategic collaboration not only supports Grupo Flow’s transition to blockchain but also sets a new benchmark in the digital media industry by enhancing scalability, security, and user engagement. Looking Forward: Grupo Flow's Next Steps With Tanssi Grupo Flow is set to revolutionize the digital media landscape with the launch of Flow Houses, inspired by Soho Houses, known for their exclusive memberships and luxurious amenities. This platform will offer unique, tokenized access to high-end venues and events through blockchain technology, showcasing Grupo Flow's commitment to pioneering new experiences in the media sector. The upcoming launch of Grupo Flow on Tanssi Mainnet later this year marks one of the largest blockchain integrations by an online media group in Brazil, leveraging the scalability and efficiency of Tanssi's infrastructure and benefiting from Polkadot's interoperable ecosystem. For those interested in expanding their project to the blockchain space, Tanssi offers a straightforward, code-free application chain deployment platform. Explore and launch your project today at apps.tanssi.network. About Tanssi Network Tanssi’s appchain infrastructure protocol is designed to simplify and accelerate the deployment of appchains. By connecting a chain to Tanssi, it is instantly transformed into a modular appchain. This transition grants access to a developer-friendly and permissionless environment, fully stocked with all essential infrastructural components to run a chain right out of the box. Key features include a shared and decentralized network of block producers, ensuring robust security and data retrievability, alongside seamless integrations with vital tools like bridges, wallets, block explorers, RPC endpoints, indexers, oracles, and more. As a result, appchains can be deployed in just minutes—a significant improvement over the typical months-long process. Learn more at tanssi.network. About Grupo Flow Grupo Flow operates as a comprehensive media ecosystem, focusing on the creator economy with ventures in technology, content, media, production, and advertising. Central to the group is Estudios Flow, a creative hub housing over 30 creators dedicated to fostering dialogue, plurality, and transparency. Producing over 70 hours of content weekly across various formats and niches, Estudios Flow aims to build community ties nationwide. The ecosystem also includes Flow S.A., Flow Games, Flow News, Flow Sport Club, Venus, Ciência Sem Fim, Amplifica, the digital agency Golden Pill, software house WolfVision, and Flow Labs, specializing in audiovisual productions. The initiative traces back to the influential Flow, launched by Igor Coelho, a trailblazer in Brazil’s videocast arena. Learn more at linkedin.com/company/flow-grupo. Contact CMO Katherine Quilca Barcelli Moondance Labs kathy@moondancelabs.com This story was distributed as a release by Chainwire under HackerNoon’s Business Blogging Program. Learn more about the program here. TORTOLA, BVI, July 25th, 2024/Chainwire/--The Tanssi appchain infrastructure protocol has announced a significant collaboration with Grupo Flow, one of Brazil’s largest media ecosystems, boasting over 23 million subscribers across multiple platforms. As Grupo Flow transitions into the blockchain space with the creation of Flow Chain, this collaboration signifies a milestone. Tanssi will facilitate this move, not only bringing Grupo Flow's vast community into the blockchain realm but also connecting them with Polkadot’s vibrant and interoperable ecosystem. This strategic integration ensures a robust, scalable infrastructure that will enhance user engagement, expand Grupo Flow's audience reach, and open new monetization opportunities for content creators within the Flow Chain platform, leveraging both existing community strengths and new blockchain capabilities. "The possibilities and connections with our audience through our own chain are very exciting. We are thrilled to have a trusted partner to guide us on this journey. Now it's time to head to the lab, study the path and roadmap, and look forward to seeing the first features in our ecosystem,” said Andre Gaigher, CCO at Group Flow. "The possibilities and connections with our audience through our own chain are very exciting. We are thrilled to have a trusted partner to guide us on this journey. Now it's time to head to the lab, study the path and roadmap, and look forward to seeing the first features in our ecosystem,” said Andre Gaigher, CCO at Group Flow. Grupo Flow Transitions to Blockchain, Setting a New Standard for the Digital Media Industry Grupo Flow is a leading digital media ecosystem in Brazil, known for its vast reach and influence in the creator economy. With diverse ventures including creative hubs, software development, and audiovisual production, Grupo Flow has solidified its position as a key player in the Brazilian media landscape. Their platforms, such as the highly popular Flow Podcast, attract millions of followers and listeners, collectively achieving over 73 million views per month and amassing more than 23 million followers. As Grupo Flow transitions into the Polkadot blockchain space with Tanssi, they aim to enhance digital engagement, handle large-scale interactions, and seamlessly integrate advanced technology, all while maintaining their commitment to delivering an excellent user experience. This collaboration will help them strengthen their audience and tap into a new community of blockchain users, further enhancing their digital media ecosystem. Tanssi & Grupo Flow: Unlocking Blockchain Potential for Digital Media Transformation Tanssi’s advanced blockchain infrastructure is crucial in facilitating Grupo Flow’s seamless transition to the blockchain space. Grupo Flow will launch its own application blockchain, Flow Chain, offering a dedicated and customizable environment tailored to their needs. This collaboration will enhance digital engagement through tokenized interactions and decentralized governance, allowing users to participate in content decisions, access exclusive events, and earn rewards. Tanssi’s scalable infrastructure will handle high volumes of transactions, ensuring efficient processing and maintaining a seamless user experience for Grupo Flow’s large and active community. By integrating with Polkadot’s interoperable ecosystem, Flow Chain will access a network of interconnected blockchains and leverage technologies like Moonbeam Routed Liquidity to expand into ecosystems such as Ethereum and Avalanche. This ensures robust security, consistent performance, and extensive growth opportunities. Tanssi's infrastructure reduces transaction costs and opens up new revenue streams through staking rewards and tokenized assets. This strategic collaboration not only supports Grupo Flow’s transition to blockchain but also sets a new benchmark in the digital media industry by enhancing scalability, security, and user engagement. Looking Forward: Grupo Flow's Next Steps With Tanssi Grupo Flow is set to revolutionize the digital media landscape with the launch of Flow Houses, inspired by Soho Houses, known for their exclusive memberships and luxurious amenities. This platform will offer unique, tokenized access to high-end venues and events through blockchain technology, showcasing Grupo Flow's commitment to pioneering new experiences in the media sector. The upcoming launch of Grupo Flow on Tanssi Mainnet later this year marks one of the largest blockchain integrations by an online media group in Brazil, leveraging the scalability and efficiency of Tanssi's infrastructure and benefiting from Polkadot's interoperable ecosystem. For those interested in expanding their project to the blockchain space, Tanssi offers a straightforward, code-free application chain deployment platform. Explore and launch your project today at apps.tanssi.network . apps.tanssi.network apps.tanssi.network About Tanssi Network Tanssi’s appchain infrastructure protocol is designed to simplify and accelerate the deployment of appchains. By connecting a chain to Tanssi, it is instantly transformed into a modular appchain. This transition grants access to a developer-friendly and permissionless environment, fully stocked with all essential infrastructural components to run a chain right out of the box. Key features include a shared and decentralized network of block producers, ensuring robust security and data retrievability, alongside seamless integrations with vital tools like bridges, wallets, block explorers, RPC endpoints, indexers, oracles, and more. As a result, appchains can be deployed in just minutes—a significant improvement over the typical months-long process. Learn more at tanssi.network . tanssi.network tanssi.network About Grupo Flow Grupo Flow operates as a comprehensive media ecosystem, focusing on the creator economy with ventures in technology, content, media, production, and advertising. Central to the group is Estudios Flow, a creative hub housing over 30 creators dedicated to fostering dialogue, plurality, and transparency. Producing over 70 hours of content weekly across various formats and niches, Estudios Flow aims to build community ties nationwide. The ecosystem also includes Flow S.A., Flow Games, Flow News, Flow Sport Club, Venus, Ciência Sem Fim, Amplifica, the digital agency Golden Pill, software house WolfVision, and Flow Labs, specializing in audiovisual productions. The initiative traces back to the influential Flow, launched by Igor Coelho, a trailblazer in Brazil’s videocast arena. Learn more at linkedin.com/company/flow-grupo . linkedin.com/company/flow-grupo linkedin.com/company/flow-grupo Contact CMO Katherine Quilca Barcelli Moondance Labs kathy@moondancelabs.com This story was distributed as a release by Chainwire under HackerNoon’s Business Blogging Program. Learn more about the program here. This story was distributed as a release by Chainwire under HackerNoon’s Business Blogging Program. Learn more about the program here . here here

Tanssi Supports Grupo Flow In Major Blockchain Expansion On Polkadot

পোলকাডটসে প্রধান ব্লকচেইন সম্প্রসারণে নাচ গ্রুপো প্রবাহকে সমর্থন করে

Dance unterstützt Grupo Flow bei der großen Blockchain-Erweiterung auf Polkadots

Dance apoya al Grupo Flow en una importante expansión de blockchain en Polkadots

Dance soutient Grupo Flow dans une extension majeure de la blockchain sur Polkadots

डांस पोलकाडॉट्स पर प्रमुख ब्लॉकचेन विस्तार में ग्रुपो फ्लो का समर्थन करता है

ダンスがポルカドットの大規模なブロックチェーン拡張で Grupo フローをサポート

Dance는 Polkadots의 주요 블록체인 확장에서 Grupo Flow를 지원합니다.

Dance apoia grupo Flow na grande expansão do Blockchain em Polkadots

Dance поддерживает поток Grupo в крупном расширении блокчейна на Polkadots

Dance, Polkadots'taki Büyük Blockchain Genişlemesinde Grupo Flow'u Destekliyor

Dance hỗ trợ dòng chảy Grupo trong việc mở rộng chuỗi khối lớn trên Polkadots

Dance 在 Polkadots 的主要区块链扩展中支持 Grupo Flow

Authors:
(1) Sasun Hambardzumyan, Activeloop, Mountain View, CA, USA;
(2) Abhinav Tuli, Activeloop, Mountain View, CA, USA;
(3) Levon Ghukasyan, Activeloop, Mountain View, CA, USA;
(4) Fariz Rahman, Activeloop, Mountain View, CA, USA;.
(5) Hrant Topchyan, Activeloop, Mountain View, CA, USA;
(6) David Isayan, Activeloop, Mountain View, CA, USA;
(7) Mark McQuade, Activeloop, Mountain View, CA, USA;
(8) Mikayel Harutyunyan, Activeloop, Mountain View, CA, USA;
(9) Tatevik Hakobyan, Activeloop, Mountain View, CA, USA;
(10) Ivo Stranic, Activeloop, Mountain View, CA, USA;
(11) Davit Buniatyan, Activeloop, Mountain View, CA, USA. Table of Links Abstract and Intro
Current Challenges
Tensor Storage Format
Deep Lake System Overview
Machine Learning Use Cases
Performance Benchmarks
Discussion and Limitations
Related Work
Conclusions, Acknowledgement, and References 3. TENSOR STORAGE FORMAT Deep Lake datasets follow columnar storage architecture, with tensors as columns, as shown in Fig. 3. Each tensor is a collection of chunks - binary blobs that contain the data samples. An index map associated with each tensor helps find the right chunk and index of the sample within that chunk for a given sample index. 3.1 Dataset A sample in a dataset represents a single row indexed across parallel tensors. As opposed to a document storage format, sample elements are logically independent, which enables partial access to samples for running performant queries or streaming selected tensors over the network to the GPU training instances. Multiple tensors can be grouped. Groups implement syntactic nesting and define how tensors are related to each other. Syntactic nesting avoids the format complication for hierarchical memory layout. Changes to the dataset’s schema are also tracked over time with version control, similar to dataset content changes. 3.2 Tensors Tensors are typed and can be appended or modified in-place. Default access to an index or a set of indices returns the data as NumPy arrays [55]. Instead of storing 1-D data as seen in Parquet [79] or series in Arrow [13], tensors can accommodate n-dimensional data, where typically the first dimension corresponds to the index or batch dimension. Tensors can contain dynamically shaped arrays, also called ragged tensors, as opposed to other statically chunked array formats such as Zarr [52]. 3.3 Types Htype defines the expectations on samples in a tensor such as data type (dtype as seen in NumPy [55]), shape, number of dimensions, or compression. Typed tensors make interacting with deep learning frameworks straightforward and enable sanity checks and efficient memory layout. By inheriting from a generic tensor htype, we can construct types such as image, video, audio, bbox, dicom, and others. For example, a tensor with image htype would expect samples being appended to it to have dtype as uint8 and shape length 3 (i.e. width, height and number of channels). We further expand on the notion of htypes allowing for meta types that support storing image sequences in tensors (sequence[image]), referencing to remotely stored images, while maintaining the regular behavior of a image tensor (link[image]), or even possible cross-format support. 3.4 Memory Layout A Deep Lake dataset contains a provenance file in JSON format and folders per tensor. A tensor contains chunks, chunk encoder, tile encoder, and tensor metadata. Tensors can be optionally hidden. For instance, hidden tensors can be used to maintain down-sampled versions of images or preserve shape information for fast queries. Tensors are stored in chunks at the storage level. While statically (inferred) shaped chunking avoids maintaining a chunk map table, it introduces significant user overhead during the specification of the tensor, custom compression usage limitations, underutilized storage for dynamically shaped tensors, and post-processing inefficiencies. Deep Lake chunks are constructed based on the lower and upper bound of the chunk size to fit a limited number of samples. This comes with a trade-off of having a compressed index map that preserves the sample index to chunk id mapping per tensor while enabling chunk sizes in the range optimal for streaming while accommodating mixed-shape samples. One could consider the approach taken in this paper as an optimized trade-off between file system page map and compute-defined map-less array storage system. For practical reasons, a single chunk encoder can be scaled to billions of images while maintaining a 150MB chunk encoder per 1PB tensor data. Further scaling can be introduced by sharding the chunk encoder. Chunks contain header information such as byte ranges, shapes of the samples, and the sample data itself. If a sample is larger than the upper bound chunk size, which is the case for large aerial or microscopy images, the sample is tiled into chunks across spatial dimensions. The only exception to tiling is videos. Videos are preserved due to efficient frame mapping to indices, key-frame-only decompression, and range-based requests while streaming. 3.5 Access Patterns The tensor storage format is optimized for deep learning training and inference, including sequential and random access. Sequential access is used for running scan queries, transforming tensors into other tensors, or running inference. Random access use cases include multiple annotators writing labels to the same image or models storing back predictions along with the dataset. While the strict mode is disabled, out-of-the-bounds indices of a tensor can be assigned, thus accommodating sparse tensors. However, random assignment over time will produce inefficiently stored data chunks. To fix the data layout, we implement an on-the-fly re-chunking algorithm to optimize the data layout. One of the key access patterns of Deep Lake is shuffled stream access for training machine learning models. It requires random or custom order access while streaming chunks into the training process. This is achieved by involving range-based requests to access sub-elements inside chunks, running complex queries before training to determine the order, and maintaining a buffer cache of fetched and unutilized data. This avoids having a separate compute cluster for running shuffling algorithm [50]. Each tensor has its own chunks, and the default chunk size is 8MB. A single chunk consists of data from multiple indices when the individual data points (image, label, annotation, etc.) are smaller than the chunk size. Conversely, when individual data points are larger than the chunk size, the data is split among multiple chunks (tiling). Exceptions to chunking logic are video data. Deep Lake format is optimized for maximizing throughput to GPU processing. It includes CPU pre-fetching, decompression or decoding, transformations, and GPU memory transfer in a deep learning framework’s expected layout. 3.6 Storage Providers Deep Lake can be plugged into any storage provider, including object storages such as AWS S3 [1], Google Cloud Storage (GCS) [3], POSIX compatible file systems, or local in-memory storage. Moreover, it constructs memory caching by chaining various storage providers together, for instance - the Least Recently Used (LRU) cache of remote S3 storage with local in-memory data. This paper is available on arxiv under CC 4.0 license. Authors: (1) Sasun Hambardzumyan, Activeloop, Mountain View, CA, USA; (2) Abhinav Tuli, Activeloop, Mountain View, CA, USA; (3) Levon Ghukasyan, Activeloop, Mountain View, CA, USA; (4) Fariz Rahman, Activeloop, Mountain View, CA, USA;. (5) Hrant Topchyan, Activeloop, Mountain View, CA, USA; (6) David Isayan, Activeloop, Mountain View, CA, USA; (7) Mark McQuade, Activeloop, Mountain View, CA, USA; (8) Mikayel Harutyunyan, Activeloop, Mountain View, CA, USA; (9) Tatevik Hakobyan, Activeloop, Mountain View, CA, USA; (10) Ivo Stranic, Activeloop, Mountain View, CA, USA; (11) Davit Buniatyan, Activeloop, Mountain View, CA, USA. Authors: Authors: (1) Sasun Hambardzumyan, Activeloop, Mountain View, CA, USA; (2) Abhinav Tuli, Activeloop, Mountain View, CA, USA; (3) Levon Ghukasyan, Activeloop, Mountain View, CA, USA; (4) Fariz Rahman, Activeloop, Mountain View, CA, USA;. (5) Hrant Topchyan, Activeloop, Mountain View, CA, USA; (6) David Isayan, Activeloop, Mountain View, CA, USA; (7) Mark McQuade, Activeloop, Mountain View, CA, USA; (8) Mikayel Harutyunyan, Activeloop, Mountain View, CA, USA; (9) Tatevik Hakobyan, Activeloop, Mountain View, CA, USA; (10) Ivo Stranic, Activeloop, Mountain View, CA, USA; (11) Davit Buniatyan, Activeloop, Mountain View, CA, USA. Table of Links Abstract and Intro Current Challenges Tensor Storage Format Deep Lake System Overview Machine Learning Use Cases Performance Benchmarks Discussion and Limitations Related Work Conclusions, Acknowledgement, and References Abstract and Intro Abstract and Intro Current Challenges Current Challenges Tensor Storage Format Tensor Storage Format Deep Lake System Overview Deep Lake System Overview Machine Learning Use Cases Machine Learning Use Cases Performance Benchmarks Performance Benchmarks Discussion and Limitations Discussion and Limitations Related Work Related Work Conclusions, Acknowledgement, and References Conclusions, Acknowledgement, and References 3. TENSOR STORAGE FORMAT Deep Lake datasets follow columnar storage architecture, with tensors as columns, as shown in Fig. 3. Each tensor is a collection of chunks - binary blobs that contain the data samples. An index map associated with each tensor helps find the right chunk and index of the sample within that chunk for a given sample index. 3.1 Dataset A sample in a dataset represents a single row indexed across parallel tensors. As opposed to a document storage format, sample elements are logically independent, which enables partial access to samples for running performant queries or streaming selected tensors over the network to the GPU training instances. Multiple tensors can be grouped. Groups implement syntactic nesting and define how tensors are related to each other. Syntactic nesting avoids the format complication for hierarchical memory layout. Changes to the dataset’s schema are also tracked over time with version control, similar to dataset content changes. 3.2 Tensors Tensors are typed and can be appended or modified in-place. Default access to an index or a set of indices returns the data as NumPy arrays [55]. Instead of storing 1-D data as seen in Parquet [79] or series in Arrow [13], tensors can accommodate n-dimensional data, where typically the first dimension corresponds to the index or batch dimension. Tensors can contain dynamically shaped arrays, also called ragged tensors, as opposed to other statically chunked array formats such as Zarr [52]. 3.3 Types Htype defines the expectations on samples in a tensor such as data type (dtype as seen in NumPy [55]), shape, number of dimensions, or compression. Typed tensors make interacting with deep learning frameworks straightforward and enable sanity checks and efficient memory layout. By inheriting from a generic tensor htype, we can construct types such as image, video, audio, bbox, dicom, and others. For example, a tensor with image htype would expect samples being appended to it to have dtype as uint8 and shape length 3 (i.e. width, height and number of channels). We further expand on the notion of htypes allowing for meta types that support storing image sequences in tensors (sequence[image]), referencing to remotely stored images, while maintaining the regular behavior of a image tensor (link[image]), or even possible cross-format support. 3.4 Memory Layout A Deep Lake dataset contains a provenance file in JSON format and folders per tensor. A tensor contains chunks, chunk encoder, tile encoder, and tensor metadata. Tensors can be optionally hidden. For instance, hidden tensors can be used to maintain down-sampled versions of images or preserve shape information for fast queries. Tensors are stored in chunks at the storage level. While statically (inferred) shaped chunking avoids maintaining a chunk map table, it introduces significant user overhead during the specification of the tensor, custom compression usage limitations, underutilized storage for dynamically shaped tensors, and post-processing inefficiencies. Deep Lake chunks are constructed based on the lower and upper bound of the chunk size to fit a limited number of samples. This comes with a trade-off of having a compressed index map that preserves the sample index to chunk id mapping per tensor while enabling chunk sizes in the range optimal for streaming while accommodating mixed-shape samples. One could consider the approach taken in this paper as an optimized trade-off between file system page map and compute-defined map-less array storage system. For practical reasons, a single chunk encoder can be scaled to billions of images while maintaining a 150MB chunk encoder per 1PB tensor data. Further scaling can be introduced by sharding the chunk encoder. Chunks contain header information such as byte ranges, shapes of the samples, and the sample data itself. If a sample is larger than the upper bound chunk size, which is the case for large aerial or microscopy images, the sample is tiled into chunks across spatial dimensions. The only exception to tiling is videos. Videos are preserved due to efficient frame mapping to indices, key-frame-only decompression, and range-based requests while streaming. 3.5 Access Patterns The tensor storage format is optimized for deep learning training and inference, including sequential and random access. Sequential access is used for running scan queries, transforming tensors into other tensors, or running inference. Random access use cases include multiple annotators writing labels to the same image or models storing back predictions along with the dataset. While the strict mode is disabled, out-of-the-bounds indices of a tensor can be assigned, thus accommodating sparse tensors. However, random assignment over time will produce inefficiently stored data chunks. To fix the data layout, we implement an on-the-fly re-chunking algorithm to optimize the data layout. One of the key access patterns of Deep Lake is shuffled stream access for training machine learning models. It requires random or custom order access while streaming chunks into the training process. This is achieved by involving range-based requests to access sub-elements inside chunks, running complex queries before training to determine the order, and maintaining a buffer cache of fetched and unutilized data. This avoids having a separate compute cluster for running shuffling algorithm [50]. Each tensor has its own chunks, and the default chunk size is 8MB. A single chunk consists of data from multiple indices when the individual data points (image, label, annotation, etc.) are smaller than the chunk size. Conversely, when individual data points are larger than the chunk size, the data is split among multiple chunks (tiling). Exceptions to chunking logic are video data. Deep Lake format is optimized for maximizing throughput to GPU processing. It includes CPU pre-fetching, decompression or decoding, transformations, and GPU memory transfer in a deep learning framework’s expected layout. 3.6 Storage Providers Deep Lake can be plugged into any storage provider, including object storages such as AWS S3 [1], Google Cloud Storage (GCS) [3], POSIX compatible file systems, or local in-memory storage. Moreover, it constructs memory caching by chaining various storage providers together, for instance - the Least Recently Used (LRU) cache of remote S3 storage with local in-memory data. This paper is available on arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. available on arxiv

Deep Lake, a Lakehouse for Deep Learning: Tensor Storage Format

ডিপ লেক, গভীর শিক্ষার জন্য একটি লেকহাউস: টেনসর স্টোরেজ ফর্ম্যাট

Deep Lake, ein Lakehouse für Deep Learning: Tensor-Speicherformat

Deep Lake, una casa de lago para el aprendizaje profundo: formato de almacenamiento tensorial

Deep Lake, un Lakehouse pour le Deep Learning : format de stockage Tensor

डीप लेक, डीप लर्निंग के लिए एक लेकहाउस: टेंसर स्टोरेज फॉर्मेट

ディープラーニングのためのレイクハウス「Deep Lake」: テンソルストレージ形式

딥 러닝을 위한 레이크하우스인 Deep Lake: Tensor 스토리지 형식

Deep Lake, um Lakehouse para aprendizado profundo: formato de armazenamento tensor

Deep Lake, домик у озера для глубокого обучения: тензорный формат хранения

Deep Lake, Derin Öğrenmeye Yönelik Bir Göl Evi: Tensör Depolama Formatı

Deep Lake, một Lakehouse cho Deep Learning: Định dạng lưu trữ Tensor

Deep Lake，深度学习的 Lakehouse：张量存储格式

The Great Barrier Reef in Australia is stunningly beautiful and the largest coral reef in the world, home to an incredible diversity of marine life. Unfortunately, the reef faces threats from crown-of-thorns starfish (COTS) that eat coral. To control COTS outbreaks, reef managers use a method called Manta Tow surveys where divers are towed behind a boat, visually assessing sections of the reef. However, this method has limitations in efficiency, data quality, and scalability. To improve COTS surveillance and control, the Great Barrier Reef Foundation started an innovation program. A key part is using underwater cameras to collect reef images and applying AI to detect COTS automatically. To develop the machine learning technology for this video-based surveying at scale, Australia's national science agency . Their goal is to create systems that can accurately and efficiently analyze large volumes of imagery to pinpoint COTS outbreaks across the vast Great Barrier Reef in near real-time. CSIRO has partnered with Google This would greatly assist conservation efforts and help secure the long-term protection of the reef ecosystem. In summary, AI-powered image analysis is being applied to automate detecting crown-of-thorns starfish from footage of the enormous Great Barrier Reef, enabling better monitoring and control at massively increased efficiency and scale to safeguard the reef. The is to accurately identify starfish in real-time by building an object detection model trained on underwater videos of coral reefs. In this blog post, we will build an ML pipeline to analyze the reef’s images and predict the presence and position of crown-of-thorns starfish. goal of this competition We will be using - an Opensource package that uses data-centric AI to clean up the issues in the image data and , , and - used for computer vision tasks like objection detection and image classification. Cleanvision Tensorflow KerasCV YOLOv8 !pip install -q cleanvision keras-cv kaggle As the data size is huge, we are directly using Kaggle to download and run the tutorial. For that, we need to configure your Kaggle creds and give you the appropriate permissions; ! read more here ! cp kaggle.json ~/.kaggle/
! chmod 600 ~/.kaggle/kaggle.json Downloading and Extracting the Kaggle Dataset !kaggle competitions download -c tensorflow-great-barrier-reef !unzip /content/tensorflow-great-barrier-reef.zip Using Cleanvision to Clean Up the Data CleanVision is an open-source data-centric AI package that automatically identifies problematic images that could negatively impact computer vision systems. It scans image datasets and detects common data issues like blur, inadequate lighting, duplicate images, etc. Fixing these problems in the training data is an essential first step before developing machine learning models to ensure their reliability. CleanVision provides a simple unified interface - just run the same few lines of Python code to audit any image collection, regardless of size or origin. This makes it extremely easy to clean dirty image data. By proactively flagging low-quality, duplicates, and other undesirable images upfront, CleanVision helps save significant time and improves outcomes when building computer vision applications. It's a fast way to boost data hygiene as a prerequisite before applying machine learning to image-based tasks. from cleanvision import Imagelab

# Specify path to folder containing the image files in your dataset
imagelab = Imagelab(data_path="train_images/")

# Automatically check for a predefined list of issues within your dataset
imagelab.find_issues()

# Produce a neat report of the issues found in your dataset
imagelab.report() import os
for st in imagelab.info['exact_duplicates']['sets']:
  os.remove(st[0]) Importing Relevant Libraries import pandas as pd
# show images inline
%matplotlib inline

import keras
import tensorflow

# import miscellaneous modules
import matplotlib.pyplot as plt
import cv2
import os
import numpy as np
import time

import tensorflow as tf Preparing Dataset for Preprocessing We will read the dataset and then preprocess it to be able to use it for model training. We begin by loading the train CSV file and converting the annotations to integer format from the original JSON format. df_train = pd.read_csv("train.csv")

df_train=df_train.loc[df_train["annotations"].astype(str) != "[]"]
df_train['annotations'] = df_train['annotations'].apply(eval)

df_train['image_path'] = "train_images/video_" + df_train['video_id'].astype(str) + "/" + df_train['video_frame'].astype(str) + ".jpg"
df_extrain=df_train.explode('annotations') # Single annotation per row
df_extrain.reset_index(inplace=True)
df_extrain.head() df_extrain_main=pd.DataFrame(pd.json_normalize(df_extrain['annotations']), columns=['x', 'y', 'width', 'height']).join(df_extrain)
df_extrain_main['class']=0
df_extrain_main=df_extrain_main[['image_path','x','y','width','height','class','video_id','video_frame']]
df_extrain_main.head(10) After the CSV processing, we try to split the CSV into a dataset and use the appropriate bounding box format, classes, and image paths to create the preliminary dataset ready to be processed further. def create_tf_example(rowss,data_df):


    xmins = []
    xmaxs = []
    ymins = []
    ymaxs = []

    # Convert ---> [xmin,ymin,width,height] to [xmins,xmaxs,ymins,ymaxs]
    xmin = rowss['x']
    xmax = rowss['x']+rowss['width']
    ymin = rowss['y']
    ymax = rowss['y']+rowss['height']


    #main_data.append((rowss['image_path'],xmins,xmaxs,ymins,ymaxs))
    return rowss['image_path'],xmin,ymin,xmax,ymax from PIL import Image, ImageDraw
paths = []
bboxes = []
classes = []
for index, row in df_extrain_main.iterrows():
            if index % 1000 == 0:
                print('Processed {0} images.'.format(index))
            image_path,xmins,ymins,xmaxs,ymaxs=create_tf_example(row,df_extrain_main)
            paths.append(image_path)
            bboxes.append([[float(xmins),float(ymins),float(xmaxs),float(ymaxs)]])
            classes.append([0]) Here, we are using to create ragged tensors from the and classes lists. A ragged tensor is a type of tensor that can handle varying lengths of data along one or more dimensions. tf.ragged.constant bbox This is useful when dealing with data that has variable-length sequences, such as text or time series data. In this case, the and classes lists have different lengths for each image, depending on the number of objects in the image and the corresponding bounding boxes and classes. To handle this variability, ragged tensors are used instead of regular tensors. bbox Later, these ragged tensors are used to create a using the method. This method creates a dataset from the input tensors by slicing them along the first dimension. tf.data.Dataset from_tensor_slices By using ragged tensors, the dataset can handle varying lengths of data for each image and provide a flexible input pipeline for further processing. bbox = tf.ragged.constant(bboxes)
classes = tf.ragged.constant(classes)
image_paths = tf.ragged.constant(paths)

data = tf.data.Dataset.from_tensor_slices((image_paths, classes, bbox)) num_val = int(bbox.shape[0] * 0.2)

# Split the dataset into train and validation sets
val_data = data.take(num_val)
train_data = data.skip(num_val) includes pre-trained models for popular computer vision datasets, such as , which can be used for transfer learning. KerasCV also provides a range of visualization tools for inspecting the intermediate representations learned by the model and for visualizing the results of object detection and segmentation tasks. KerasCV ImageNet, COCO, and Pascal VOC In this particular notebook, we are using YOLOV8, so we have to format the dataset which is in a compatible format according to the YOLOV8 model. The bounding box input format should be as follows. bounding_boxes = {
    # num_boxes may be a Ragged dimension
    'boxes': Tensor(shape=[batch, num_boxes, 4]),
    'classes': Tensor(shape=[batch, num_boxes])
} The dictionary has two keys, ' ' and ' ', each of which maps to a TensorFlow or Tensor object. The ' ' Tensor has a shape of , where batch is the number of images in the batch and is the maximum number of bounding boxes in any image. boxes classes RaggedTensor boxes [batch, num_boxes, 4] num_boxes The 4 represents the four values needed to define a bounding box: xmin, ymin, xmax, ymax. The 'classes' Tensor has a shape of , where each element represents the class label for the corresponding bounding box in the ' ' Tensor. The num_boxes dimension may be ragged, which means that the number of boxes may vary across images in the batch. [batch, num_boxes] boxes The final model input looks something like this: {"images": images, "bounding_boxes": bounding_boxes} def load_image(image_path):
    image = tf.io.read_file(image_path)
    image = tf.image.decode_jpeg(image, channels=3)
    return image


def load_dataset(image_path, classes, bbox):
    # Read Image
    image = load_image(image_path)
    bounding_boxes = {
        "classes": tf.cast(classes, dtype=tf.float32),
        "boxes": tf.cast(bbox, dtype=tf.float32),
    }
    return {"images": tf.cast(image, tf.float32), "bounding_boxes": bounding_boxes} Data Augmentation Applying effective data augmentation is critical yet challenging when building object detection models. Transformations like crops, flips, etc. must update the bounding box coordinates correctly. Doing this manually is complex and error-prone. KerasCV provides specialized layers to handle augmentation aware of bounding boxes. It offers a wide range of transformations that automatically adjust bboxes to match the augmented images. This integrated bounding box handling enables easily incorporating powerful augmentation into object detection pipelines with KerasCV. By leveraging tf.data pipelines, the augmentation can be done on-the-fly during training. With KerasCV's robust bbox-aware augmentation capabilities, developers can achieve more diverse and useful training data and improved generalization in object detection models, while avoiding cumbersome manual handling of coordinates. The layers handle that complexity behind the scenes. augmenter = keras.Sequential(
    layers=[
        keras_cv.layers.RandomFlip(mode="horizontal", bounding_box_format="xyxy"),
        keras_cv.layers.RandomShear(
            x_factor=0.2, y_factor=0.2, bounding_box_format="xyxy"
        ),
        keras_cv.layers.JitteredResize(
            target_size=(640, 640), scale_factor=(0.75, 1.3), bounding_box_format="xyxy"
        ),
    ]
)
resizing = keras_cv.layers.JitteredResize(
    target_size=(640, 640),
    scale_factor=(0.75, 1.3),
    bounding_box_format="xyxy",
) Training Dataset With Augmentation We are preparing the training dataset with augmentation for our object detection model. We start by setting the to 4, and then use the function to load the data. batch_size load_dataset Next, we shuffle the data to ensure that our model doesn't overfit on a particular kind of data. We then use the function to batch the data with a fixed size of 4 and drop any remaining data. ragged_batch Finally, we apply the function to the dataset to perform data augmentation, which helps to increase the diversity of the dataset and improve the model's accuracy. augmenter The parameter is set to to allow TensorFlow to dynamically tune the number of parallel calls to improve performance. num_parallel_calls tf.data.AUTOTUNE BATCH_SIZE =4
train_ds = train_data.map(load_dataset, num_parallel_calls=tf.data.AUTOTUNE)
train_ds = train_ds.shuffle(BATCH_SIZE * 4)
train_ds = train_ds.ragged_batch(BATCH_SIZE, drop_remainder=True)
train_ds = train_ds.map(augmenter, num_parallel_calls=tf.data.AUTOTUNE) Validation Dataset In this code block, we are creating a validation dataset by mapping the 'load_dataset' function to the 'val_data' set. This function helps in loading the data from the dataset and preprocessing it for the model. We are also setting 'num_parallel_calls' to 'tf.data.AUTOTUNE' to enable parallelism for faster processing. Next, we are shuffling the validation dataset by a factor of 'BATCH_SIZE * 4' to increase randomness and prevent overfitting. Finally, we are creating a 'ragged_batch' of 'BATCH_SIZE' with 'drop_remainder=True' to ensure that all batches have the same size. This helps in making the model more efficient and consistent during training. val_ds = val_data.map(load_dataset, num_parallel_calls=tf.data.AUTOTUNE)
val_ds = val_ds.shuffle(BATCH_SIZE * 4)
val_ds = val_ds.ragged_batch(BATCH_SIZE, drop_remainder=True) Bounding Box Visualisation We use the library for bounding box visualization. It imports the necessary packages and defines a class mapping for the dataset. The function takes in the dataset inputs and uses the function to display the images with bounding boxes overlaid. You can see that via and datasets. keras_cv visualize_dataset() plot_bounding_box_gallery() train_ds val_ds import keras_cv
from keras_cv import bounding_box
from keras_cv import visualization
class_mapping = {0:'fish'}

def visualize_dataset(inputs, value_range, rows, cols, bounding_box_format):
    inputs = next(iter(inputs.take(1)))
    images, bounding_boxes = inputs["images"], inputs["bounding_boxes"]
    visualization.plot_bounding_box_gallery(
        images,
        value_range=value_range,
        rows=rows,
        cols=cols,
        y_true=bounding_boxes,
        scale=5,
        font_scale=0.7,
        bounding_box_format=bounding_box_format,
        class_mapping=class_mapping,
    )


visualize_dataset(
    train_ds, bounding_box_format="xyxy", value_range=(0, 255), rows=2, cols=2
)

visualize_dataset(
    val_ds, bounding_box_format="xyxy", value_range=(0, 255), rows=2, cols=2
) Next, we unpack the dataset to be able to feed it into the model training function of the Keras API. def dict_to_tuple(inputs):
    return inputs["images"], inputs["bounding_boxes"]


train_ds = train_ds.map(dict_to_tuple, num_parallel_calls=tf.data.AUTOTUNE)
train_ds = train_ds.prefetch(tf.data.AUTOTUNE)

val_ds = val_ds.map(dict_to_tuple, num_parallel_calls=tf.data.AUTOTUNE)
val_ds = val_ds.prefetch(tf.data.AUTOTUNE) Model Building is the newest iteration of the popular YOLO (You Only Look Once) family of models used for computer vision tasks like object detection and image classification. It was created by Ultralytics, building on their previous YOLOv5 model. YOLOv8 Compared to prior versions, YOLOv8 integrates numerous architectural upgrades for improved accuracy and performance. The creators also aimed to enhance the overall developer experience. As the latest release in the industry-leading YOLO lineup, YOLOv8 represents the current state-of-the-art in domains like object detection. It incorporates both algorithmic advances to push the boundaries of CV capability and usability improvements for practitioners. Keras-CV offers multiple models- , , , , and . For this example, we choose the backbone and load it with the pre-trained weights. YOLOV8 YOLOv8n YOLOv8s YOLOv8m YOLOv8l YOLOv8x YOLOV8s coco Next, we build a model using the which accepts a feature extractor as the backbone argument, a num_classes argument that specifies the number of object classes to detect based on the size of the class_mapping list, a bounding_box_format argument that informs the model of the format of the bbox in the dataset, and finally, the feature pyramid network (FPN) depth is specified by the argument. YOLOV8 YOLOV8Detector fpn_depth backbone = keras_cv.models.YOLOV8Backbone.from_preset(
    "yolo_v8_s_backbone_coco"  # We will use yolov8 small backbone with coco weights
)
yolo = keras_cv.models.YOLOV8Detector(
    num_classes=1,
    bounding_box_format="xyxy",
    backbone=backbone,
    fpn_depth=1,
) Compiling the Model The training process of a YOLOv8 model involves the use of two types of losses - classification loss and box loss. The classification loss measures the difference between the predicted class probabilities and the true class probabilities for each detected object. Binary cross entropy loss is utilized in this case, as each detected item either belongs to a particular class or does not. This loss function helps optimize the model for accurate object classification. On the other hand, the box loss calculates the dissimilarity between the predicted bounding boxes and the ground truth boxes. YOLOv8 uses Complete IoU (CIoU) instead of basic IoU for box loss calculation. CIoU considers additional factors like the boxes' aspect ratio, center distance, and scale. This helps better represent box similarity by looking at more box attributes beyond just overlap. By jointly optimizing for object classification and precise localization during training, the classification and box losses help YOLOv8 generate highly accurate object detection by minimizing discrepancies between outputs and label data across classes and bounding box coordinates. optimizer = tf.keras.optimizers.Adam(
    learning_rate=10e-3,
)

yolo.compile(
    optimizer=optimizer, classification_loss="binary_crossentropy", box_loss="ciou"
) Training the Model During each epoch, the model iteratively adjusts its parameters to minimize the difference between its predicted output and the actual output. This process helps the model learn how to accurately detect crown-of-thorns starfish in the underwater videos of coral reefs. The number of epochs can be adjusted based on the training dataset size, complexity of the model, and desired accuracy. After training, the model can be used to predict the presence and position of starfish in real-time, helping researchers and conservationists monitor and control COTS outbreaks more efficiently and effectively. yolo.fit(
    train_ds,
    epochs=3,
) Conclusion The integration of , , , and in this project exemplifies the power and versatility of AI in environmental conservation, particularly for the Great Barrier Reef. Cleanvision TensorFlow KerasCV YOLOv8 By leveraging the strengths of these advanced tools, we've developed a robust model capable of accurately identifying crown-of-thorns starfish in complex underwater environments. This not only marks a significant step forward in protecting one of the world's natural wonders but also demonstrates the immense potential of AI in addressing ecological challenges.

How to Use TensorFlow and Cleanvision to Detect Starfish Threats in the Great Barrier Reef

Let's learn about via these 59 free stories. They are ordered by most time reading created on HackerNoon. Visit the to find the most read stories about any technology. Tensorflow /Learn Repo 1. Quantum Machine Learning Using TensorFlow Quantum INTRODUCTION 2. What Does Facebook Hydra Mean For The Future of Python Ever since its inception in the year 1089 by Guido Van Rossum, the programming language Python has along far away. Sheldon did its creator knew that Python would in today's world be utilized for a variety of purposes such as research, development, scripting, among many others. Built as a successor in the ABC language, Python does not just find its applications in software development but also in research. 3. YouTube's Recommendation Engine: Explained Every successful tech product, by the very definition, is a result of some technological marvels working with impeccable user experience to solve a key problem for the users. One such marvel is the recommendation engine by YouTube. 4. NSFW Filter Introduction: Building a Safer Internet Using AI Filtering out NSFW images with a web extension built using TensorFlow JS. 5. Why ML in Production is (still) Broken and Ways we Can Fix it Machine Learning, Deep Learning development in production was still broken. ZenML, an extensible, open-source MLOps framework for production-ready ML pipelines. 6. Deploy First TensorFlow Model in Android App Simple linear regression is useful for finding the relationship between two continuous variables. One is a predictor or independent variable and the other is a response or dependent variable. It looks for a statistical relationship but not a deterministic relationship. Relationship between two variables is said to be deterministic if one variable can be accurately expressed by the other. For example, using temperature in degrees Celsius it is possible to accurately predict Fahrenheit. 7. How to Run Machine-Learning Models in the Browser using ONNX Learn how to use ONNX Runtime Web to deploy machine-learning models natively to the browser. 8. Training Your Models on Cloud TPUs in 4 Easy Steps on Google Colab You have a plain old TensorFlow model that’s too computationally expensive to train on your standard-issue work laptop. I get it. I’ve been there too, and if I’m being honest, seeing my laptop crash twice in a row after trying to train a model on it is painful to watch. 9. 10 Must-Try Open Source Tools for Machine Learning Machine learning is the future. But will machines ever extinct humans? 10. Efficient Implementation of MobileNet and YOLO Object Detection Algorithms for Image Annotation The objective of the problem is to implement classification and localization algorithms to achieve high object classification and labelling accuracies, and train models readily with as least data and time as possible. The solution to the problem is considered in the following blog. 11. Cocktail Alchemy: Creating New Recipes With Transformers Build a transformer model with natural language processing to create new cocktail recipes from a cocktail database. 12. Auto-Generating Lyrics With TensorFlow and Machine Learning: A How-To Guide Creating a bot that, given a starting phrase, would generate its own lyrics, powered by a machine learning model that would have learned from existing songs. 13. How I Transfer an Artistic Style to Any Image Machine learning and artificial intelligence have been on my radar for years now, but more as a concept and “thing I should know about.” I didn’t feel that I had the free time or skills to dig into it. However, my attitude about machine learning has changed in the past few months. I have seen new and easier tools become accessible to the public. In this post I will walk you through how to transfer an art style to any image using some of these tools. 14. Style Transferring with TensorFlow Style transfer is a computer vision-based technique combined with image processing. Learn about style transfer with Tensorflow, a prominent framework in AI & ML 15. Time Series Forecasting with TensorFlow.js Pull stock prices from online API and perform predictions using Recurrent Neural Network & Long Short Term Memory (LSTM) with TensorFlow.js framework 16. Manipulación de tensores en PyTorch. ¡El primer paso para el deep learning! *Nota: Contactar a Omar Espejel (omar@tsc.ai) para cualquier observación. Cualquier error es responsabilidad del autor. 17. TLDR Newsletter Week of August 5th Highlights 18. Flax: Google’s Open Source Approach To Flexibility In Machine Learning Thinking of Machine Learning, the first frameworks that come to mind are Tensorflow and PyTorch, which are currently the state-of-the-art frameworks if you want to work with Deep Neural Networks. Technology is changing rapidly and more flexibility is needed, so Google researchers are developing a new high performance framework for the open source community: Flax. 19. Top 3 Face Datasets and How to Work with Them An image dataset contains specially selected digital images intended to help train, test, and evaluate an artificial intelligence (AI) or machine learning (ML) 20. PyTorch vs TensorFlow: Who has More Pre-trained Deep Learning Models? Given the importance of pre-trained Deep Learning models, which Deep Learning framework - PyTorch or TensorFlow - has more of these models available to users is 21. Rare Datasets for Computer Vision Every Machine Learning Expert Must Work With Have you ever being in a situation to guess another person’s age? Well May be YES!! How about playing games like finding things in minimum time? or about finding the written character where your doctor wrote in the prescription when you are sick? 22. How I Built an AI to Detect License Plate Number Registration (ANPR) This Car Mod Is A Privacy Nightmare! (AI Number Plate Reader with Python, Tensorflow, OpenCV, OpenALPR) 23. How to Fine Tune a 🤗 (Hugging Face) Transformer Model How to fine-tune a Hugging Face Transformer model for Sequence Classification 24. Top 10 Data Science Libraries in Python Data Science Libraries that will shine this year. 25. What Neural Networks Teach Us About Schizophrenia Pretrained Artificial Neural Networks used to work like a Blackbox: You hand them an input and they predict an output with a certain probability — but without us knowing the internal processes of how they came up with their prediction. A Neural Network to recognize images usually consists of around 20 neuron layers, trained with millions of images to tweak the network parameters to give high quality classifications. 26. Classify Handwritten Digits using Deep learning with Tensorflow Deep learning is a subpart of machine learning and artificial intelligence which is also known as deep neural network this networks capable of learning unsupervised from provided data which is unorganized or unlabeled. today, we will implement a neural network in 6 easy steps using TensorFlow  to classify handwritten digits. 27. Artificial Intelligence Vs Machine Learning: What's the difference? AI and Machine Learning are predominant terms that are creating a lot of buzz in the technology world. The terms can often be used interchangeably but that’s not the case, AI and ML are way more different from each other in their approach, algorithms and logical thinking. 28. How To Build Links Detector That Making Links in Your Book Clickable How I built a link detector for your smart phone to browse links printed in books. 29. Computer Vision Could Improve Health and Workplace Safety Recent developments in the field of training Neural Networks (Deep Learning) and advanced algorithm training platforms like Google’s TensorFlow and hardware accelerators from Intel (OpenVino), Nvidia (TensorRT) etc., have empowered developers to train and optimize complex Neural Networks in small edge devices like Smart Phones or Single Board Computers. 30. From TF to TFLite: Deploying ML Models on Mobile [Part 2] This is part 2 of the two-part article on deploying ML models on mobile. We saw how to convert our ML models to TfLite format here. For those of you who came here first, I recommend you click on the above link to get the whole picture.  If you just want the android part ,the demo app we are building has a GAN model generating handwritten digits and a classifier model predicting the generated digit. 31. From TF to TFLite: Deploying ML Models on Mobile [Part 1] tl;dr - Link to code: TensorFlow GAN model. So the other day I was talking to my rubber ducky about how G-Board predicts my next word, even when those words are entirely made up by me, in that how it actually learns on-device. How amazingly Netflix, Amazon, Google Maps make use of machine learning in their apps. How does machine learning on apps even work? Does the model learn even after being deployed? Can I deploy a GAN model on mobile? 32. How to Use TensorFlow in Python: Google‘s Open-Source Library For Deep Learning You might not always know it, but Deep Learning is everywhere. We explain how to use TensorFlow, Google's Library For Deep Learning, in Python. 33. Build Your Own Voice Recognition Model with Tensorflow While I'm usually a JavaScript person, there are plenty of things that Python makes easier to do. Doing voice recognition with machine learning is one of those. 34. C++ to WebAssembly using Bazel and Emscripten How to get Bazel and Emscripten to compile C++ to WebAssembly or JavaScript 35. Approach Pre-Trained Deep Learning Models With Caution Pre-trained models are easy to use, but are you glossing over details that could impact your model performance? 36. Tensorflow Vs. Keras: Comparison by building a model for image classification Yes, as the title says, it has been very usual talk among data-scientists (even you!) where a few say, TensorFlow is better and some say Keras is way good! Let’s see how this thing actually works out in practice in the case of image classification. 37. Human Detection System Using RaspberryPi, Thermal Camera and Machine Learning Triggering reliable events based on the presence of people has been the dream of many geeks and DIY automators for a while. Having your house to turn the lights on or off when you enter or exit your living room is an interesting application, for instance. Most of the solutions out there to solve these kinds of problems, even more high-end solutions like the Philips Hue sensors, detect motion, not actual people presence — which means that the lights will switch off once you lay on your couch like a sloth. 38. How Blockchain & AI Integration is Changing Business Landscape? The potential of Blockchain is no lesser than Artificial intelligence. If you have taken a look at them, you must already know the impacts of these technologies on various industries. 39. Lessons for Improving Training Performance — Part 1 Part 1: Lower precision & larger batch size are standard now 40. [Hacking Tinder] Train an AI to Auto-Swipe for You 🖖 Auto-tinder was created to train an AI using Tensorflow and Python3 that learns your interests in the other sex and automatically plays the tinder swiping-game for you. 41. Fabio Manganiello on Home-Made Computer Vision, IoT, Automation, AI Fabio Manganiello writes about solutions he's discovered while building a platform, library of plugins and an API to connect/manage any device and service through any backend, allowing users to easily set up any kind of automation. Fabio is based in Amsterdam, the Netherlands, and has been nominated for a 2020 #Noonie for exceptional contributions to the IoT tag category on Hacker Noon. 42. Efficient Model Training in the Cloud with Kubernetes, TensorFlow, and Alluxio Open Source This article presents the collaboration of Alibaba, Alluxio, and Nanjing University in tackling the problem of Deep Learning model training in the cloud. Various performance bottlenecks are analyzed with detailed optimizations of each component in the architecture. This content was previously published on Alluxio's Engineering Blog, featuring Alibaba Cloud Container Service Team's case study (White Paper here). Our goal was to reduce the cost and complexity of data access for Deep Learning training in a hybrid environment, which resulted in over 40% reduction in training time and cost. 43. The Full Story behind Convolutional Neural Networks and the Math Behind it Convolutional Neural Networks became really popular after 2010 because they outperformed any other network architecture on visual data, but the concept behind CNN is not new. In fact, it is very much inspired by the human visual system. In this article, I aim to explain in very details how researchers came up with the idea of CNN, how they are structured, how the math behind them works and what techniques are applied to improve their performance. [44. Differential Privacy with Tensorflow 2.0 :  Multi class Text Classification Privacy](https://hackernoon.com/differential-privacy-with-tensorflow-20-multi-class-text-classification-privacy-yk7a37uh) Introduction 45. What Is Edge AI? Edge AI—also referred to as on-device AI—commonly refers to the components required to run an AI algorithm locally on a hardware device. 46. Mueller Report for Nerds! Spark meets NLP with TensorFlow and BERT (Part 1) Photo by Michael on Unsplash 47. [Tutorial] Build a Gender Classifier for Live Webcam Stream using Tensorflow and OpenCV Training a Neural Network from scratch suffers two main problems. First, a very large, classified input dataset is needed so that the Neural Network can learn the different features it needs for the classification. 48. Objects Classification Using CNN-based Model — All the images (plots) are generated and modified by the Author. 49. Can I Grade Loans Better Than LendingClub? In case you missed it, I built a neural network to predict loan risk using a public dataset from LendingClub. Then I built a public API to serve the model’s predictions. That’s nice and all, but… how good is my model? 50. C++ to WebAssembly using Bazel and Emscripten How to get Bazel and Emscripten to compile C++ to WebAssembly or JavaScript 51. Training Your Own Text Classification Model From Scratch With Tensorflow Is As Easy As ABC Hello ML Newb! In this article, you will learn to train your own text classification model from scratch using Tensorflow in just a few lines of code. 52. RethNet Model: Object-by-Object Learning for Detecting Facial Skin Problems In August 2019, a group of researchers from lululab Inc propose the state-of-the-art concept using a semantic segmentation method to detect the most common facial skin problems accurately. The work is accepted to ICCV 2019 Workshop. 53. Uber AI Labs Senior Research Scientist Talks TensorFlow 2.0 [Interview] There’s no doubt that TensorFlow is one of the most popular machine learning libraries right now. However, newbie developers who want to experiment with TensorFlow often face difficulties in learning TensorFlow; the framework has a not unjustified reputation for having a steep learning curve that can make it hard for developers to get to grips with quickly. 54. Training Machine Learning Models Using TensorFlow or PyTorch I will show you how gradient descent works, which is in the deepest deep of machine learning. 55. Building Machine Learning Models With TensorFlow In this article, I will share with you some useful tips and guidelines that you can use to better build better deep learning models. 56. Build A Smart Baby Monitor Using a RaspberryPi and Tensorflow Some of you may have noticed that it’s been a while since my last article, despite winning this year's IoT Noonies award (btw thanks to all of you who voted, that means a lot to me!). 57. #Switch2Swift for Deep Learning If you are interested, what the recent fast.ai advanced and closed Deep Learning Class had to say about Google’s Swift for Tensorflow project, you might find this post interesting. Even if you attended the class, you should find here hopefully a good overview (with links into the class, presentations, and additional material), what Swift for Tensorflow is and why it might be relevant. 58. Scale Vision Transformers (ViT) Beyond Hugging Face Speed up state-of-the-art ViT models in Hugging Face 🤗 up to 2300% (25x times faster ) with Databricks, Nvidia, and Spark NLP 🚀 59. Loan Risk Prediction Using Neural Networks A Step-by-Step Guide (With a Healthy Dose of Data Cleaning) Thank you for checking out the 59 most read stories about Tensorflow on HackerNoon. to find the most read stories about any technology. Visit the /Learn Repo

59 Stories To Learn About Tensorflow

, co-founder of began his journey as a University of Nevada, Las Vegas (UNLV) student and is now an influential figure in the global tech community. He embodies the essence of innovation and ambition so many founders aspires to. Tomasik most attended the AMD Advancing AI Event, where TensorWave announced a significant partnership. Piotr Tomasik TensorWave Inc. A UNLV College of Engineering alumnus, Piotr's journey is not just about personal success; it's a testament to the potential of Las Vegas's tech sector. His role in co-founding TensorWave Inc. and steering it towards groundbreaking collaborations is a beacon of inspiration for upcoming tech enthusiasts in the city. Under Piotr’s guidance, TensorWave has not only made strides in AI technology but also cemented Las Vegas's position on the global tech map. The recent partnership with AMD and GigaIO, leading to the development of the TensorNODE infrastructure powered by AMD Instinct MI300X accelerators, is a milestone in AI technology. This collaboration isn't just a technological triumph; it's a narrative of how local talent can propel a city into the limelight of the global tech arena. A Testament to Local Talent: Piotr Tomasik’s Impact on Las Vegas Las Vegas Chief Information Officer Michael Lee Sherwood's commendation of Piotr Tomasik underscores the significant impact he has had on the city’s technology landscape. Sherwood’s praise highlights how Tomasik's innovative approach and commitment to excellence have not only accelerated the growth of the local tech sector but also enhanced Las Vegas's reputation as a hub for advanced AI solutions. “In my eight years as the CIO for the City of Las Vegas, I've had the privilege of witnessing firsthand the remarkable impact Piotr Tomasik has had on our technology landscape. Piotr's innovative approach and commitment to excellence have not only accelerated the growth of our local tech sector but have also placed Las Vegas on the map as a hub for advanced AI solutions. Celebrating our local tech leaders is a cornerstone of our city's evolution into a leading smart city, driven by cutting-edge technology and ambitious talent.” Beyond his achievements with TensorWave and the groundbreaking TensorNODE infrastructure, Piotr Tomasik's career spans various critical roles in the tech ecosystem. As a celebrated Las Vegas Top Tech Exec, a Vegas INC 40 under 40 awardee, and a proud UNLV Alumni of the Year, Piotr's contributions extend well beyond individual companies. His co-founding of StartUp Vegas and active role as a General Partner at Vegas Tech Ventures signify his dedication to cultivating a thriving tech community. Piotr's involvement as an advisor to numerous tech-focused organizations further underscores his commitment to guiding and nurturing the next generation of tech innovators. This multifaceted experience makes Piotr a beacon for aspiring entrepreneurs and investors alike, proving his expertise is not just in technological innovation but also in fostering a vibrant startup ecosystem. The Future of AI and Las Vegas: Piotr's Vision Piotr’s vision extends beyond technological advancements; it's about elevating Las Vegas in the global tech community. His excitement about bringing TensorWave's innovative solutions to the global market, in collaboration with industry giants like GigaIO and AMD, reflects his dedication to both his hometown and the broader field of AI.

POWER LEADER: TensorWave co-founder Piotr Tomasik secures key partnership at AMD Advancing AI event 

The Tansy Patch

Stories, News & Mentions about TANSO

Tanso’s Stories on HackerNoon