Altera

altera

USA v. eBay Court Filing, retrieved on January 11, 2024, is part of . You can jump to any part in this filing . This is part 12 of 12. HackerNoon’s Legal PDF Series here COUNT SIX - Destruction, Alteration, and Falsification of Records in a Federal Investigation (18 U.S.C. § 1519) The United States Attorney further charges: 60. The United States Attorney re-alleges and incorporates by reference paragraphs 1 through 53 of this Information. 61. On or about the dates set forth below, in the District of Massachusetts, the defendant, eBay Inc., did knowingly alter, destroy, conceal and falsify the records, documents, and tangible objects below with the intent to impede, obstruct, and influence the investigation and proper administration of a matter within the jurisdiction of any department and agency of the United States, that is, the harassment and intimidation campaign charged in this Information. All in violation of 18 U.S.C. § 1519. About HackerNoon Legal PDF Series: We bring you the most important technical and insightful public domain court case filings. This court case retrieved on January 11, 2024, from is part of the public domain. The court-created documents are works of the federal government, and under copyright law, are automatically placed in the public domain and may be shared without legal restriction. ecommercebytes.com

eBay Charged With Destruction, Alteration and Falsification of Records in a Federal Investigation

The New Spirit by Havelock Ellis is part of the HackerNoon Books Series. You can jump to any chapter in this book here . PREFACE TO THE THIRD EDITION PREFACE TO THE THIRD EDITION No alterations have been made in this edition. It is true that three of the figures here studied were living when the book was written; but their genius had matured, their work was for the most part done. Nothing they could produce would seriously modify one’s conception of them as aboriginal personal forces, the outcome of the past, the initiators of the future. Apart from this, it seems to me a mistake to manipulate or add to one’s own completed work. If I were to re-write it, I should doubtless write it differently; the Conclusion, for instance, which is earliest in date, seems to me now rather formal and metaphysical. But for the most part I have nothing seriously to alter or to omit. I have sometimes been asked why, in a discussion of some of the new influences of the past century, I have left out representative men who have made so great a stir in the world. Goethe, it may possibly be true, stalks through every page, but where are Kant, Hegel, Auguste Comte, John Stuart Mill, Herbert Spencer? I cannot remember ever proposing to include these names. The reason may be clearer if I mention other names I once wished to include, although—partly doubting my competence to discuss them, partly fearing that their introduction might seem to interfere with the unity of the book—I ultimately refrained. One was Burne Jones. I shall never forget how, as a youth in the Public Library at Sydney, I turned over the leaves of a volume of etchings and suddenly alighted on “Merlin and Vivien.” Something I knew of Botticelli, Lippi and the rest, and I had brooded over their antique mystery and charm; but here were all the mystery and the charm brought down among us from the world where saints stand stiff and aureoled, and angels walk tip-toe on lily cups. The fifteenth century artists of Flanders and Venice and Florence introduced us into a frankly supernatural world, and they delighted like children to scatter rich fruits on the golden floors, and to stick peacocks’ feathers into the bejewelled walls. It is a rarer and subtler art to suggest that infinitely remote world while accepting the austere conditions of our own earth. The pale ghosts of Puvis de Chavannes’ frescoes are a far-off suggestion of this art; and one thinks too of the modern magician who has brought before us the twinkling of Salome’s feet by the red blood from the Baptist’s head, curdling amid the flowers; the rich-robed daughters of Apollo among the olives; the mystic elephant in solemn festival, gathering the lotus with his trunk as his feet plash slowly in the clear waters of the sacred lake. But the shadowy art of Puvis, the wayward and limited art of Gustave Moreau, come short of the consistent and completely realised art which has been attained by the painter who stands forth in the eyes of Europe as the greatest imaginative artist of England. It is a new synthesis of the world of nature and the world of dreams. The three women who dance in the foreground of “The Mill” tell us of a country where human joys and sorrows, hopes and fears, are set to a different measure, and sung in unknown keys. A strange and troublous art, it seems sometimes,—like the sinuous melodies of Renan, which seem to belong to some far-haunted past, and yet contain the intimate secrets of our own hearts,—but it fascinates and holds us as though music became visible before our eyes. It opens before us a new and delightful pathway into the land of dreams. Another was Auguste Rodin. To mould the human figure has been an amusement for man since ever he carved wood or indented clay. It was left for the sculptors of Egypt and of Greece and of Italy to form human figures of stone, not as a mere symbol of the reality, but as a revelation of their own moods and visions of beauty or passion; and since then the amusement has fallen back into convention and symbol, although the plastic representation of the modern human body, etiolated and hidden, offers fewer difficulties than its representation in painting which Millet and Degas have in varying ways striven to achieve. Now even the great sculptors of old only suggest to us beauty or grace or strength that has become conventional; they reveal nothing. In this man’s work the form that is closest to us of all forms in the world, that we cling to from the day of birth, and that remains with us, half-seen or divined, until the day of death, has been revealed anew, just as new aspects of light have been revealed by Claude Monet. It is the ancient human way-worn and passion-used form, rendered with pathetic truth, and yet we feel that we have never truly seen the human body before. We marvel how expression can be carried so far without passing the bounds of nature and simplicity. It is far from the method of Michelangelo, Rodin’s immediate predecessor, with whom it has been the fashion to compare him. Michelangelo’s stupendous fantasy twisted the human body into the strange or lovely shapes of his own inverted dreams. In Rodin’s work, it is through a relentless love of nature that we are led to a new and intimate vision of the body. The quiet artist in his simple work-room has been building up through long years his great Gate of Hell; it is the gate of the joy and beauty and terror of life, expressed otherwise than those sober stories of the old world so charmingly told on that gate that was thought worthy of Heaven. But through this gate we are led to a new insight of that figure in the world which is closest to us and most precious, such an insight, it may well be, as Pheidias and Donatello brought to the men of their time. Another personality that I desired to analyse, and perhaps the greatest, was Richard Wagner. The Leipzig youth, who hated the tawdry tinsel of the theatre, and was so little of a musical prodigy that he could never learn to play the piano, impelled by a strange instinct has yet wrought music and the stage to a poetic height never before approached. Just as our arts rise out of our industries, so the manifold art of Wagner—woven of music and poetry and drama—rises to something that is beyond art. Wagner has made the largest impersonal synthesis yet attainable of the personal influences that thrill our lives, and has built it on the broadest physiological basis of our senses, so that faith has here become sight. Such harmony is what we are accustomed to call Heaven, and such art—to the mere musician cacophony and confusion—is truly called religion. It will take some time yet before we understand its place in life as a new expression of the human soul. Generations must pass before it will be possible for a greater artist, by a still wider sensory appeal, to lift us to any higher Heaven. It is not the men of one idea—important as these are—who most truly represent the spirit of an age. Such men most often represent the spirit of some earlier generation, which in them has become definitely crystallised. It is the men whose ideas are still free in pungent, penetrating, often confused solution that we may count nearest to the natural forces of an age, and it is these that are most interesting to analyse. In such men the feebler instincts of their fellows are concentrated, and the flaming energy of their spirits attracts few, repels most, of their fellows. It is, no doubt, because of this high degree of emotional exaltation that these men bring us to religion. It all comes to religion. I would point out to those who think that this result needs apology, that such men do not bring before us the pale, animistic children of dreams, who for so many ages have sought with their shadowy arms to beckon men away from the world to a home on the other side of the sky, but the robust children of our working life, the offspring of our living energies and emotions, the harmonised satisfaction of all that we have lived, of all that we have felt. So the “new spirit” brings us to one of the most ancient modes of human emotion. I sought to emphasise this in my Introduction as well as in the Conclusion, not altogether successfully for some of my readers, who have been led to credit me with virtues of modernity to which I can make no claim. So far from being “an apostle of modernity,” the “new spirit” that I am concerned with is but a quickening in the pulse of life such as may take place in any age, though my tracings are only of a recent acceleration. The greatest manifestation of the new spirit that I know of took place long since in the zoological history of the race when the immediate ancestor of man began to walk on his hind legs, so developing the skilful hands and restless brain that brought sin into the world. That strange and perilous method of locomotion—which carried other diseases and disabilities in its train, more concrete than sin—marked a revolutionary outburst of new life worth contemplating. Yet even among the later and minor movements of life it is not the most recent that to me personally are the most attractive. The Eiffel Tower does not thrill me like the gray towers of Chartres; I find the streets of Zaragoza more interesting than those of Manchester. And, on the other hand, there are modernities which seem to me old, very old, older than life itself. To say this is no doubt to confess that the personal element has a large place in this study of the “New Spirit.” And it is true that, however honest a piece of mechanism your sphygmograph may be, if it is alive there is a very considerable personal equation which you must make up your mind to reckon with. I believe I am not altogether incapable of slinging facts at the head of the British Goliath (with purely benevolent intentions), but on this occasion I wrote for my own pleasure: let me apologise to Goliath for any annoyance I may so have caused him. I wished to speak for once, so far as might be, in my own voice, glad if here and there a reader cared to follow my impatient track, furnishing from the stores of his own knowledge and intelligence what was lacking in commentaries and pièces justificatives. I wished at the outset to take a bird’s-eye view of the world as it presented itself to me personally, only indicating by mere hints those parts of the field in which I was more specially concerned. And I wished also to indicate—perhaps once for all—my own faith in those large facts of nature which are unaffected by personal equation, and which harmonise all our petty individual activities. Nature is bent on her own ends, and with infinite ingenuity uses all our energies to carry out her idea of increasing and multiplying the countless forms of life. Death itself is but an accidental after-thought, a beneficial adaptation—as Weismann would have us express it—only affecting the body, that servant of the immortal germ-cells which has grown so large and arrogant since the days when we Metazoa were young in the world. That is the one master-thought of Nature, or—shall we say?—her systematised delusion, her délire à forme chronique. But the malady, if it is one, is incurable. A friend of mine, under the influence of nitrous oxide, once found himself face to face with the Almighty. Being a man of earnest and philosophic temperament, he took advantage of the opportunity to demand passionately the meaning and aim of this tangled skein of things in which we find ourselves: “Why have You placed us here? For what purpose have You submitted us to all this strife and misery? What is the solution of the riddle of life?” And then, uttered in a characteristic bass, came in one word the awful reply which my friend will never forget: “ .” I fear that that voice is, or might well have been, divine. Procreation And yet why should one “fear”? We have our brief triumph. Seeking out many curious things, we learn to know and to enjoy the earth. Nature’s naughty children—whether artists or scientists or mystics—we may stand aside, contemplate her great object, and impudently elevate our fingers to our nose. It amuses us, and scarcely hurts her. She cannot refuse us the by-play of her own adaptations. For it all comes of that primitive manifestation of the new spirit, the “Fall,” which raised us on to our hind limbs and enabled us to drink of the Cider of Paradise. H. E. 7th October, 1892. About HackerNoon Book Series: We bring you the most important technical, scientific, and insightful public domain books. This book is part of the public domain. Havelock Ellis (2017). The New Spirit. Urbana, Illinois: Project Gutenberg. Retrieved October 2022 https://www.gutenberg.org/cache/epub/55878/pg55878-images.html This eBook is for the use of anyone anywhere at no cost and with almost no restrictions whatsoever. You may copy it, give it away or re-use it under the terms of the Project Gutenberg License included with this eBook or online at www.gutenberg.org , located at https://www.gutenberg.org/policy/license.html.

No alterations have been made in this edition

High entry barriers, complex transaction processes, and geographical boundaries have historically constrained the real estate and fine art markets. These difficulties though aren't slowing down the management of traditional assets—especially on the blockchain. According to a report by Standard Chartered, tokenized Real-World Assets (RWAs) will reach $30 trillion by 2034. This significant potential growth points to the untapped potential of real-world assets. Tokenization through the blockchain is revolutionizing how traditional assets are managed today. It is transforming global trade by improving accessibility and liquidity. Blockchain technology is shaking up traditional asset management by improving transparency and immutability, increasing efficiency, enhancing security, and providing global accessibility. There is less intermediary involvement in tokenized assets, unlike what is obtainable with traditional asset management. Eliminating middlemen makes the process more transparent (reducing fraud and increasing trust) and less complex. In this article, we will explore how traditional assets are being bridged to the digital space using tokenization. Real-World Assets Tokenization Real-world asset tokenization is the process of issuing digital tokens based on the blockchain to physical or traditional assets like gold, real estate, machinery, etc. Essentially, the tokenization of these assets involves creating tokens that are typically issued as smart contracts on blockchain networks like Ethereum, Solana, Polygon, etc. Every token issued represents a fractional ownership of the underlying asset, and this is backed by a legal framework ensuring the connection between the token and the physical asset. A wide variety of assets can be tokenized, they include: Financial Instruments like stocks, bonds, and other structured products
Real Estate like commercial and residential properties
Commodities like gold, silver, oil, etc.
Assets like arts and collectibles RWA Tokenization and the Opportunities in the Market We are at a transformative phase in the financial markets with RWA tokenization. This use case of blockchain technology has seen renowned financial institutions and fintech innovators actively developing and partnering with tokenization platforms. According to forecasts, it is predicted that 7-9% of investors’ portfolios will be allocated to tokenized assets by 2027 and the industry is on course to reach that. A pointer to this is the market experiencing traction in securities tokenization, with major players like BlackRock, Goldman Sachs, Franklin Templeton, and JPMorgan launching dedicated tokenization initiatives. $10 trillion BlackRock for instance recently partnered with Securitize to provide better access to traditional financial products via digitization. BlackRock’s tokenized fund BUIDL, is leading the tokenized Treasury category with a market cap of $541 million. Franklin Templeton’s tokenized Treasury FOBXX is the third-largest with a market capitalization of $410 million. The Total Value Locked (TVL) in the RWA sector at the time of writing is $6.4 billion. This represents the industry's economic value and its universal acceptance. As institutional-standard infrastructure continues to mature for trading, custody, and other products and services, private market assets like real estate and private equity have emerged as early adoption leaders. Some key drivers of this adoption include the demand for access to premium investment opportunities without intermediaries, the push for more liquidity in otherwise traditionally illiquid assets, and major cost reductions in asset management and transactions. The sector however faces significant challenges still, some of which include the complexity of integrating traditional financial products with blockchain infrastructure and regulatory uncertainty. Other challenges include education and institutional adoption curves, and technicalities around interoperability, scalability, and security. With regulations, it varies across jurisdictions. Certain regions are emerging as clear leaders in providing regulatory frameworks for tokenized assets. Countries like Switzerland and Singapore have established progressive environments that support the tokenization of RWAs while protecting investors and their investments. Pioneers at the Forefront of RWA Tokenization Enter RWA Inc. and Others The RWA sector is gaining momentum thanks to the work of projects in the space. RWA Inc. is one such key player and pioneering platform leading the charge in the RWA tokenization sector and redefining how we interact with RWAs on the blockchain. This is the first comprehensive RWA ecosystem offering end-to-end RWA tokenization through a cutting-edge multi-asset platform that includes tokenization-as-a-service, a launchpad, and a marketplace. RWA Inc. isn't only digitizing assets, it's also unlocking an entirely new standard for asset ownership, trading, and management. The multi-asset platform seamlessly integrates a launchpad and a marketplace, while offering tokenization as a service, bridging the gap between traditional finance and a digital future on the blockchain. Operating in a potential $30 trillion market, RWA Inc. is well-positioned to be a dominant force in the RWA sector as it leverages unmatched regulatory compliance (already established 6 regulated trading licenses in the UAE) and transformative asset accessibility. The licenses RWA Inc. holds positions it as the premier onramp for traditional investment firms, banks, and hedge fund managers. $RWA is the native utility token that fuels the RWA Inc. ecosystem. Other projects at the forefront of RWA tokenization include: Propy: A decentralized real estate protocol that leverages blockchain technology to facilitate real estate transactions. YieldBricks: a company that provides seamless DeFi pools for tokenizing yield via real estate assets. EstateX: A blockchain-based company that democratizes access to real estate investments with increased liquidity, lower investment minimums, and portfolio diversification. Metamovers: An innovative blockchain platform engineered to transform the secondary market for real-world assets, specifically focusing on real estate. Future of RWA Tokenization Speaking on tokenization and the RWA industry, David Henderson, the Head of Marketing at Backed Finance, a significant player in the tokenization of government securities, said, “The tokenization revolution is in full swing. Financial institutions are embracing this technology, recognizing its potential to reshape the global financial landscape. The distinction between ‘real-world’ and digital assets will blur as blockchains become the settlement layer for all financial transactions, democratizing access to markets worldwide. The future of finance is borderless and inclusive.” RWA tokenization is at a focal intersection between traditional finance and blockchain innovation. Integrations between tokenized RWAs and DeFi protocols are creating new avenues for yield generation and lending markets. The development of institutional-standard infrastructures also increases adoption by major traditional financial players. As the technology continues to mature, we will see the emergence of sophisticated systems that combine blockchain's efficiency with the mechanisms of the conventional financial market. The tokenization of real-world assets is homogenizing access to asset classes that were previously exclusive, reducing market friction, and automating compliance processes. These are potentially profound impacts on traditional finance. The success of these alterations, however, depends largely on continued technology advancements, regulatory clarity, and institutional adoption. Conclusion One of the most significant innovations in modern finance in recent times is the tokenization of RWAs. They play a major role in bridging the gap between traditional assets and blockchain financial infrastructure. Even though the technology and market frameworks are still evolving, the foundations for massive transformations are being laid by projects like RWA Inc. They are changing how we view asset ownership and trading. The convergence of technological advancement, increase in institutional interest, and proper regulatory development corroborate that RWA tokenization is a rudimentary shift in the financial markets. And they are set to play a major part in this bull run. So, as an investor or an institution, this is the best time to develop strategic approaches to RWA tokenization. High entry barriers, complex transaction processes, and geographical boundaries have historically constrained the real estate and fine art markets. These difficulties though aren't slowing down the management of traditional assets—especially on the blockchain. According to a report by Standard Chartered, tokenized Real-World Assets (RWAs) will reach $30 trillion by 2034. High entry barriers, complex transaction processes, and geographical boundaries have historically constrained the real estate and fine art markets. These difficulties though aren't slowing down the management of traditional assets—especially on the blockchain. According to a report by Standard Chartered, tokenized Real-World Assets (RWAs) will reach $30 trillion by 2034. report report This significant potential growth points to the untapped potential of real-world assets. Tokenization through the blockchain is revolutionizing how traditional assets are managed today. It is transforming global trade by improving accessibility and liquidity. This significant potential growth points to the untapped potential of real-world assets. Tokenization through the blockchain is revolutionizing how traditional assets are managed today. It is transforming global trade by improving accessibility and liquidity. Blockchain technology is shaking up traditional asset management by improving transparency and immutability, increasing efficiency, enhancing security, and providing global accessibility. There is less intermediary involvement in tokenized assets, unlike what is obtainable with traditional asset management. Eliminating middlemen makes the process more transparent (reducing fraud and increasing trust) and less complex. In this article, we will explore how traditional assets are being bridged to the digital space using tokenization. Real-World Assets Tokenization Real-World Assets Tokenization Real-world asset tokenization is the process of issuing digital tokens based on the blockchain to physical or traditional assets like gold, real estate, machinery, etc. Essentially, the tokenization of these assets involves creating tokens that are typically issued as smart contracts on blockchain networks like Ethereum, Solana, Polygon, etc. Every token issued represents a fractional ownership of the underlying asset, and this is backed by a legal framework ensuring the connection between the token and the physical asset. A wide variety of assets can be tokenized, they include: Financial Instruments like stocks, bonds, and other structured products Real Estate like commercial and residential properties Commodities like gold, silver, oil, etc. Assets like arts and collectibles Financial Instruments like stocks, bonds, and other structured products Real Estate like commercial and residential properties Commodities like gold, silver, oil, etc. Assets like arts and collectibles RWA Tokenization and the Opportunities in the Market RWA Tokenization and the Opportunities in the Market We are at a transformative phase in the financial markets with RWA tokenization. This use case of blockchain technology has seen renowned financial institutions and fintech innovators actively developing and partnering with tokenization platforms. According to forecasts, it is predicted that 7-9% of investors’ portfolios will be allocated to tokenized assets by 2027 and the industry is on course to reach that. allocated allocated A pointer to this is the market experiencing traction in securities tokenization, with major players like BlackRock, Goldman Sachs, Franklin Templeton, and JPMorgan launching dedicated tokenization initiatives. $10 trillion BlackRock for instance recently partnered with Securitize to provide better access to traditional financial products via digitization. BlackRock’s tokenized fund BUIDL, is leading the tokenized Treasury category with a market cap of $541 million . Franklin Templeton’s tokenized Treasury FOBXX is the third-largest with a market capitalization of $410 million . $541 million $541 million $410 million $410 million The Total Value Locked (TVL) in the RWA sector at the time of writing is $6.4 billion . This represents the industry's economic value and its universal acceptance. As institutional-standard infrastructure continues to mature for trading, custody, and other products and services, private market assets like real estate and private equity have emerged as early adoption leaders. $6.4 billion $6.4 billion Some key drivers of this adoption include the demand for access to premium investment opportunities without intermediaries, the push for more liquidity in otherwise traditionally illiquid assets, and major cost reductions in asset management and transactions. The sector however faces significant challenges still, some of which include the complexity of integrating traditional financial products with blockchain infrastructure and regulatory uncertainty. Other challenges include education and institutional adoption curves, and technicalities around interoperability, scalability, and security. With regulations, it varies across jurisdictions. Certain regions are emerging as clear leaders in providing regulatory frameworks for tokenized assets. Countries like Switzerland and Singapore have established progressive environments that support the tokenization of RWAs while protecting investors and their investments. Switzerland and Singapore Switzerland and Singapore Pioneers at the Forefront of RWA Tokenization Enter RWA Inc. and Others Pioneers at the Forefront of RWA Tokenization Enter RWA Inc. and Others The RWA sector is gaining momentum thanks to the work of projects in the space. RWA Inc. is one such key player and pioneering platform leading the charge in the RWA tokenization sector and redefining how we interact with RWAs on the blockchain. This is the first comprehensive RWA ecosystem offering end-to-end RWA tokenization through a cutting-edge multi-asset platform that includes tokenization-as-a-service, a launchpad, and a marketplace. RWA Inc. isn't only digitizing assets, it's also unlocking an entirely new standard for asset ownership, trading, and management. The multi-asset platform seamlessly integrates a launchpad and a marketplace, while offering tokenization as a service, bridging the gap between traditional finance and a digital future on the blockchain. Operating in a potential $30 trillion market, RWA Inc. is well-positioned to be a dominant force in the RWA sector as it leverages unmatched regulatory compliance (already established 6 regulated trading licenses in the UAE) and transformative asset accessibility. The licenses RWA Inc. holds positions it as the premier onramp for traditional investment firms, banks, and hedge fund managers. $RWA is the native utility token that fuels the RWA Inc. ecosystem. Other projects at the forefront of RWA tokenization include: Propy: A decentralized real estate protocol that leverages blockchain technology to facilitate real estate transactions. Propy: YieldBricks: a company that provides seamless DeFi pools for tokenizing yield via real estate assets. YieldBricks: EstateX: A blockchain-based company that democratizes access to real estate investments with increased liquidity, lower investment minimums, and portfolio diversification. EstateX: Metamovers: An innovative blockchain platform engineered to transform the secondary market for real-world assets, specifically focusing on real estate. Metamovers: Future of RWA Tokenization Future of RWA Tokenization Speaking on tokenization and the RWA industry, David Henderson, the Head of Marketing at Backed Finance, a significant player in the tokenization of government securities, said , “The tokenization revolution is in full swing. Financial institutions are embracing this technology, recognizing its potential to reshape the global financial landscape. The distinction between ‘real-world’ and digital assets will blur as blockchains become the settlement layer for all financial transactions, democratizing access to markets worldwide. The future of finance is borderless and inclusive.” said said RWA tokenization is at a focal intersection between traditional finance and blockchain innovation. Integrations between tokenized RWAs and DeFi protocols are creating new avenues for yield generation and lending markets. The development of institutional-standard infrastructures also increases adoption by major traditional financial players. As the technology continues to mature, we will see the emergence of sophisticated systems that combine blockchain's efficiency with the mechanisms of the conventional financial market. The tokenization of real-world assets is homogenizing access to asset classes that were previously exclusive, reducing market friction, and automating compliance processes. These are potentially profound impacts on traditional finance. The success of these alterations, however, depends largely on continued technology advancements, regulatory clarity, and institutional adoption. Conclusion Conclusion One of the most significant innovations in modern finance in recent times is the tokenization of RWAs. They play a major role in bridging the gap between traditional assets and blockchain financial infrastructure. Even though the technology and market frameworks are still evolving, the foundations for massive transformations are being laid by projects like RWA Inc. They are changing how we view asset ownership and trading. The convergence of technological advancement, increase in institutional interest, and proper regulatory development corroborate that RWA tokenization is a rudimentary shift in the financial markets. And they are set to play a major part in this bull run. So, as an investor or an institution, this is the best time to develop strategic approaches to RWA tokenization.

Bridging Traditional Assets to the Blockchain Through Tokenization

The Building Blocks of UI/UX: How Atomic Design Transforms Your Workflow

Authors:
(1) Nora Schneider, Computer Science Department, ETH Zurich, Zurich, Switzerland (nschneide@student.ethz.ch);
(2) Shirin Goshtasbpour, Computer Science Department, ETH Zurich, Zurich, Switzerland and Swiss Data Science Center, Zurich, Switzerland (shirin.goshtasbpour@inf.ethz.ch);
(3) Fernando Perez-Cruz, Computer Science Department, ETH Zurich, Zurich, Switzerland and Swiss Data Science Center, Zurich, Switzerland (fernando.perezcruz@sdsc.ethz.ch). Table of Links Abstract and 1 Introduction 2 Background 2.1 Data Augmentation 2.2 Anchor Regression 3 Anchor Data Augmentation 3.1 Comparison to C-Mixup and 3.2 Preserving nonlinear data structure 3.3 Algorithm 4 Experiments and 4.1 Linear synthetic data 4.2 Housing nonlinear regression 4.3 In-distribution Generalization 4.4 Out-of-distribution Robustness 5 Conclusion, Broader Impact, and References A Additional information for Anchor Data Augmentation B Experiments Abstract We propose a novel algorithm for data augmentation in nonlinear over-parametrized regression. Our data augmentation algorithm borrows from the literature on causality and extends the recently proposed Anchor regression (AR) method for data augmentation, which is in contrast to the current state-of-the-art domain-agnostic solutions that rely on the Mixup literature. Our Anchor Data Augmentation (ADA) uses several replicas of the modified samples in AR to provide more training examples, leading to more robust regression predictions. We apply ADA to linear and nonlinear regression problems using neural networks. ADA is competitive with state-of-the-art C-Mixup solutions. [1] 1 Introduction Data augmentation is one of the key ingredients of any successful application of a machine learning classifier. The first example that typically comes to mind is the in-depth description of the data augmentation in the now-famous Alexnet paper [26]. Data augmentation algorithms come in different flavors, and they mostly rely on the expectation that small perturbations, invariances, or symmetries applied to the input will not change the class label. That way, we can present ‘fresh new’ samples as alterations of the available examples for training. These transformations modify the input distribution to make the algorithm more robust for cases where the distribution of the test set may differ from that of the training set. We refer the reader to the related work section (Section 2.1) for an overview and description of different data augmentation strategies. The literature for data augmentation in regression is slim. The paper on Mixup augmentation [51] proposes a simple and general scheme for data augmentation using convex combinations of samples. The authors only apply their data augmentation proposal to classification problems. They conjecture in the discussion that the application to regression is straightforward, however, this is not the case in practice. Mixup is theoretically analyzed in [5, 52] as a regularization technique for classification and regression problems. However, it is only illustrated in classification problems. The Mixup algorithm has been extended to regression problems in [18, 49], in which the authors explain that Mixup cannot be blindly applied to regression problems. To our knowledge, these are the only two papers in which data augmentation for regression is proposed. RegMix [18] relies on a hard-to-train prior neural network controller before augmenting the data using a Mixup strategy. C-Mixup [49], a method proposed more recently, solves some of the issues limiting the standard Mixup algorithm for regression problems. The authors propose to mix only closeby samples in the output space (i.e., samples which have close enough labels). This strategy is only valid when the target variables are monotonic with the input and is applied in a transformed space. The authors present comprehensive results in data augmentation for in-distribution generalization, task generalization and out-of-distribution robustness. In this paper, we rely on the causality literature to provide a different avenue for augmenting data in regression problems. Causal discovery finds the causes of a response variable among a given set of observations or helps to recognize the causal relations between a set of variables [39]. These causes allow us to understand how these relations will change if we were to intervene in a subset of the (input) variables or what would be the effect on the output. So, in general, the regression model will be robust to perturbations in the input variables making the prediction less sensitive to changes in the distribution of the test set. For example, the authors in [40] use the invariance property for prediction to perform causal inference. In turn, Anchor Regression (AR) builds upon the causality literature to obtain robust regression solutions when the input variables have been perturbed [42]. The procedure relies on anchor variables capturing the heterogeneity within a dataset and a parameter γ that measures the deviation with respect to the least square solution. Once the values of the anchors are known, AR modifies the data and obtains the least square solution, as detailed in Section 2.2. In this paper, we propose Anchor Data Augmentation (ADA) to augment the training dataset with several replicas of the available data. We use a simple clustering of the data to encode a homogeneous group of observations and use different values of γ to robustify the solution to different strengths of potential distribution shifts. In every minibatch, we sample γ from a predetermined range around γ = 1. As AR was developed for linear regression, the data augmentation strategy needs to be modified for nonlinear regression accordingly. We validate ADA for in-distribution generalization and out-of-distribution robustness under the same conditions proposed in C-Mixup [49], as well as some illustrative linear and nonlinear regression examples. In the replicated experiments, ADA is competitive or superior to other augmentation strategies such as C-Mixup, although on some datasets the performance gain is marginal. The rest of the paper is organized as follows: First, we provide background information in Section 2. We give a brief overview of related work on data augmentation in Section 2.1 and summarize the key concepts on Anchor Regression in Section 2.2. Second, Section 3 shows how we extend Anchor Regression and introduces ADA. Section 4 reports empirical evidence that our approach can improve predictions, especially in over-parameterized settings. We conclude the paper in Section 5. This paper is available on arxiv under CC0 1.0 DEED license. [1] Our Python implementation of ADA is available at: https://github.com/noraschneider/ anchordataaugmentation/ Authors: (1) Nora Schneider, Computer Science Department, ETH Zurich, Zurich, Switzerland (nschneide@student.ethz.ch); (2) Shirin Goshtasbpour, Computer Science Department, ETH Zurich, Zurich, Switzerland and Swiss Data Science Center, Zurich, Switzerland (shirin.goshtasbpour@inf.ethz.ch); (3) Fernando Perez-Cruz, Computer Science Department, ETH Zurich, Zurich, Switzerland and Swiss Data Science Center, Zurich, Switzerland (fernando.perezcruz@sdsc.ethz.ch). Authors: Authors: (1) Nora Schneider, Computer Science Department, ETH Zurich, Zurich, Switzerland (nschneide@student.ethz.ch); (2) Shirin Goshtasbpour, Computer Science Department, ETH Zurich, Zurich, Switzerland and Swiss Data Science Center, Zurich, Switzerland (shirin.goshtasbpour@inf.ethz.ch); (3) Fernando Perez-Cruz, Computer Science Department, ETH Zurich, Zurich, Switzerland and Swiss Data Science Center, Zurich, Switzerland (fernando.perezcruz@sdsc.ethz.ch). Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 Background 2.1 Data Augmentation 2.1 Data Augmentation 2.2 Anchor Regression 2.2 Anchor Regression 3 Anchor Data Augmentation 3 Anchor Data Augmentation 3.1 Comparison to C-Mixup and 3.2 Preserving nonlinear data structure 3.1 Comparison to C-Mixup and 3.2 Preserving nonlinear data structure 3.3 Algorithm 3.3 Algorithm 4 Experiments and 4.1 Linear synthetic data 4 Experiments and 4.1 Linear synthetic data 4.2 Housing nonlinear regression 4.2 Housing nonlinear regression 4.3 In-distribution Generalization 4.3 In-distribution Generalization 4.4 Out-of-distribution Robustness 4.4 Out-of-distribution Robustness 5 Conclusion, Broader Impact, and References 5 Conclusion, Broader Impact, and References A Additional information for Anchor Data Augmentation A Additional information for Anchor Data Augmentation B Experiments B Experiments Abstract We propose a novel algorithm for data augmentation in nonlinear over-parametrized regression. Our data augmentation algorithm borrows from the literature on causality and extends the recently proposed Anchor regression (AR) method for data augmentation, which is in contrast to the current state-of-the-art domain-agnostic solutions that rely on the Mixup literature. Our Anchor Data Augmentation (ADA) uses several replicas of the modified samples in AR to provide more training examples, leading to more robust regression predictions. We apply ADA to linear and nonlinear regression problems using neural networks. ADA is competitive with state-of-the-art C-Mixup solutions. [1] 1 Introduction Data augmentation is one of the key ingredients of any successful application of a machine learning classifier. The first example that typically comes to mind is the in-depth description of the data augmentation in the now-famous Alexnet paper [26]. Data augmentation algorithms come in different flavors, and they mostly rely on the expectation that small perturbations, invariances, or symmetries applied to the input will not change the class label. That way, we can present ‘fresh new’ samples as alterations of the available examples for training. These transformations modify the input distribution to make the algorithm more robust for cases where the distribution of the test set may differ from that of the training set. We refer the reader to the related work section (Section 2.1) for an overview and description of different data augmentation strategies. The literature for data augmentation in regression is slim. The paper on Mixup augmentation [51] proposes a simple and general scheme for data augmentation using convex combinations of samples. The authors only apply their data augmentation proposal to classification problems. They conjecture in the discussion that the application to regression is straightforward, however, this is not the case in practice. Mixup is theoretically analyzed in [5, 52] as a regularization technique for classification and regression problems. However, it is only illustrated in classification problems. The Mixup algorithm has been extended to regression problems in [18, 49], in which the authors explain that Mixup cannot be blindly applied to regression problems. To our knowledge, these are the only two papers in which data augmentation for regression is proposed. RegMix [18] relies on a hard-to-train prior neural network controller before augmenting the data using a Mixup strategy. C-Mixup [49], a method proposed more recently, solves some of the issues limiting the standard Mixup algorithm for regression problems. The authors propose to mix only closeby samples in the output space (i.e., samples which have close enough labels). This strategy is only valid when the target variables are monotonic with the input and is applied in a transformed space. The authors present comprehensive results in data augmentation for in-distribution generalization, task generalization and out-of-distribution robustness. In this paper, we rely on the causality literature to provide a different avenue for augmenting data in regression problems. Causal discovery finds the causes of a response variable among a given set of observations or helps to recognize the causal relations between a set of variables [39]. These causes allow us to understand how these relations will change if we were to intervene in a subset of the (input) variables or what would be the effect on the output. So, in general, the regression model will be robust to perturbations in the input variables making the prediction less sensitive to changes in the distribution of the test set. For example, the authors in [40] use the invariance property for prediction to perform causal inference. In turn, Anchor Regression (AR) builds upon the causality literature to obtain robust regression solutions when the input variables have been perturbed [42]. The procedure relies on anchor variables capturing the heterogeneity within a dataset and a parameter γ that measures the deviation with respect to the least square solution. Once the values of the anchors are known, AR modifies the data and obtains the least square solution, as detailed in Section 2.2. In this paper, we propose Anchor Data Augmentation (ADA) to augment the training dataset with several replicas of the available data. We use a simple clustering of the data to encode a homogeneous group of observations and use different values of γ to robustify the solution to different strengths of potential distribution shifts. In every minibatch, we sample γ from a predetermined range around γ = 1. As AR was developed for linear regression, the data augmentation strategy needs to be modified for nonlinear regression accordingly. We validate ADA for in-distribution generalization and out-of-distribution robustness under the same conditions proposed in C-Mixup [49], as well as some illustrative linear and nonlinear regression examples. In the replicated experiments, ADA is competitive or superior to other augmentation strategies such as C-Mixup, although on some datasets the performance gain is marginal. Anchor Data Augmentation (ADA) The rest of the paper is organized as follows: First, we provide background information in Section 2. We give a brief overview of related work on data augmentation in Section 2.1 and summarize the key concepts on Anchor Regression in Section 2.2. Second, Section 3 shows how we extend Anchor Regression and introduces ADA. Section 4 reports empirical evidence that our approach can improve predictions, especially in over-parameterized settings. We conclude the paper in Section 5. This paper is available on arxiv under CC0 1.0 DEED license. This paper is available on arxiv under CC0 1.0 DEED license. available on arxiv [1] Our Python implementation of ADA is available at: https://github.com/noraschneider/ anchordataaugmentation/

A Novel Anchor Data Augmentation Algorithm for Nonlinear Models

Building an API with Go, PostgreSQL, Google Cloud and CockroachDB

AI’s ability to mimic human behavior, emotional intelligence, creativity, and morality while exploring the future of artificial intelligence Artificial intelligence, especially that afforded by ChatGPT, has changed our manner of dealing with machines. AI now puts us in a state of conversation, learns from us, and even attempts to understand the context behind the questions we ask. However, the significant issue that people keep asking is whether ChatGPT can be trained to become human. At the core of AIs like ChatGPT is a neural network, a complex web of algorithms meant to mimic the functioning of the human brain. The algorithms learn about patterns, make predictions, and improve over time through a process known as machine learning. However, there is quite a gap between teaching patterns and being human. But being human isn’t just about patterns or predictions and it is about emotions, experience, understanding of culture, making moral judgments, and the ability for truly original creativity. Interacting with AI today may often generate human-like responses that are interesting and even insightful at times. Yet, what generates those responses — patterns found in large datasets, not quite the sort of personal contemplation or emotive depth carried into every encounter by humans. To answer if ChatGPT can be trained to be human, we must consider what exactly it means for something to be human. For this argument, let’s break it down into several such components: emotional intelligence, creativity, morality, memory, and learning from experiences. Some of these are important facets of human behavior and cognition. Can AI, particularly ChatGPT, ever be imbued with these so that we consider the AI “human” to that degree? Emotional Intelligence Emotional intelligence is the ability to sense how others feel and manage one’s own emotions. Our words usually carry emotional intensity from subtle sarcasm to overt anger or happiness. Humans use tone, context, and body language to communicate and interpret emotions. ChatGPT will simulate aspects of emotional intelligence because it can generate text that carries an emotional flavor for a conversation. If you express frustration, then it produces a response that would appear empathetic; if you are ebullient, it can echo some lighthearted repartee. Does this mean it “knows” or understands emotions? No. Instead, what ChatGPT does is recognize patterns in language as they reflect emotion, but it does not feel emotions. It doesn’t have the physiological and psychological mechanisms that underlie human emotion. Training ChatGPT to “be human” in this way would take more than feeding it datasets on emotional expression. It would rather require an architecture that allowed for some authentic experiences of emotions — that is, something beyond the scope of current developments in AI. Even if we can, in some way, simulate emotional reactions in AI, they will still be fundamentally different from human emotions, which are tied closely with consciousness, biology, and experience. Creativity Creativity- the mark of human intelligence: the ability to come up with new ideas, solve problems in innovative ways, and create art that vibrates at a deeply emotional or intellectual level. Is ChatGPT creative? In this respect, yes. ChatGPT can write, for instance, poems, short stories, code- whatever you want it to do. All the same, this creativity is based on the pattern rather than on actual thinking outside of the box that may reflect human creativity. Some of the most creative solutions or artistic expressions are often born out of personal experience, emotional ups and downs, moments of inspiration, those things that cannot easily be expressed and replicated in an algorithm. Although AI can generate creative content, it remains without experiential or emotional depth to inform human creativity. This means it offers content based on learned patterns rather than any personal experience or emotional journey. That is the main point of explaining why, despite impressive outputs, ChatGPT is fundamentally different from a human mind. Morality The other very important element of being human is morality, for which decisions are generally guided by the moral compass of and through upbringing, culture, experiences, and societal norms. More often than not, humans struggle with conflicting values, weigh the consequences of their deeds, and sometimes, change their beliefs over time. Could ChatGPT make moral decisions? Not at all. While it may be able to be programmed to adhere to certain guidelines of morality or avoid harmful content, its understanding of morality is superficial. It does not possess personal values or understand right and wrong in the way humans do. Instead, it responds based on patterns in data that often simply mimic moral reasoning without truly engaging in the complex, human process of ethical decision-making. It is a huge task to train ChatGPT to be human in this regard since even humans are unable to accept one sense of moral principle. Morality also relates to consciousness, which AI lacks. Though AI systems can follow rules, they are not able to understand the sense of moral dilemmas that define human experience. Memory and Experience Humans are a product of their memories and experiences. Every conversation, every relationship, and triumph and every failure leave an impression on who we are and how we approach the world. AI does not have any kind of real memory of past interactions unless it has been explicitly programmed to retain specific data. Each interaction with ChatGPT is unique, and while it can process historical data, it doesn’t “remember” interactions the same way a human does. It would give this form so that ChatGPT can learn through experiences, creating a sense of identity and personal history over time. Now, it does not necessarily imply that it should remember anything but interpret it to affect the subsequent actions; the same way humans do in life, growing and changing according to their past. Current AI models, including ChatGPT, do not have this ability. Though they might learn from big datasets, they lack personal experiences and do not change according to individual interaction. Because of this, for ChatGPT to be closer to “being human,” it requires something more than the capability to store and accumulate information but must also develop the capability to reflect on that same information and to change in meaningful ways relative to past experiences. Learning by Experience Human intelligence is reinforced by the ability to reflect upon mistakes and utilize those as learning experiences. We sometimes reflect on our actions, question our decisions, and make alterations in behavior based on our experiences. ChatGPT, though, can be trained through machine learning models, though it learns very differently compared to humans. Improvement of AI in machine learning is through wide training on various datasets and adjustment of weights and biases for better outcomes. However, the methodology itself is not like that of human learning based on trial and error and is contextual, emotionally and socially tuned, and based on personal experience and interactions with other humans. In contrast, the learning process of ChatGPT is abstract and data-driven. While AI may learn to make improvements in its responses, it is without a human ability to reflect on itself. Human learning isn’t just adding new information, but building that new information off of what we already know and thinking about what the implications are. Sometimes, what we learn is questioned. ChatGPT learns to do this task better at getting statistically likely responses without any understanding or reflection about what it is learning. The Future of AI So, can ChatGPT be made human? Currently, as of the technology, the answer is no. In full, that is. While still able to mimic and even generate human-like conversations, produce creative content, and maintain adherence to all sorts of ethical directives, AI continues to lack depth, consciousness, and emotional complexities of what it means to be human. However, the future of AI is no longer so static. Researchers are more and more pushing the frontiers of what could be done with AI: find new approaches to making machines more intelligent, adaptive, and even emotionally intelligent. Some believe that with domains such as neuro-symbolic AI-machine learning coupled with symbolic reasoning or even quantum computation, we may eventually build one day the essence of human-like cognition in an AI. But yet, all of this raises a basic question: Do we want AI to be human? The development of machines that could think and feel like humans raises profound questions in the realms of ethics and philosophy. Where do we draw that line between the machine and the person if AI becomes too human?
Should AI have rights when it can feel? These are questions that society will have to grapple with as AI continues its evolution. AI’s ability to mimic human behavior, emotional intelligence, creativity, and morality while exploring the future of artificial intelligence Artificial intelligence, especially that afforded by ChatGPT, has changed our manner of dealing with machines. AI now puts us in a state of conversation, learns from us, and even attempts to understand the context behind the questions we ask. However, the significant issue that people keep asking is whether ChatGPT can be trained to become human. However, the significant issue that people keep asking is whether ChatGPT can be trained to become human. However, the significant issue that people keep asking is whether ChatGPT can be trained to become human. At the core of AIs like ChatGPT is a neural network, a complex web of algorithms meant to mimic the functioning of the human brain. The algorithms learn about patterns, make predictions, and improve over time through a process known as machine learning. However, there is quite a gap between teaching patterns and being human. But being human isn’t just about patterns or predictions and it is about emotions, experience, understanding of culture, making moral judgments, and the ability for truly original creativity. Interacting with AI today may often generate human-like responses that are interesting and even insightful at times. Yet, what generates those responses — patterns found in large datasets, not quite the sort of personal contemplation or emotive depth carried into every encounter by humans. To answer if ChatGPT can be trained to be human, we must consider what exactly it means for something to be human. For this argument, let’s break it down into several such components: emotional intelligence, creativity, morality, memory, and learning from experiences. Some of these are important facets of human behavior and cognition. Can AI, particularly ChatGPT, ever be imbued with these so that we consider the AI “human” to that degree? Can AI, particularly ChatGPT, ever be imbued with these so that we consider the AI “human” to that degree? Can AI, particularly ChatGPT, ever be imbued with these so that we consider the AI “human” to that degree? Emotional Intelligence Emotional intelligence is the ability to sense how others feel and manage one’s own emotions. Our words usually carry emotional intensity from subtle sarcasm to overt anger or happiness. Humans use tone, context, and body language to communicate and interpret emotions. ChatGPT will simulate aspects of emotional intelligence because it can generate text that carries an emotional flavor for a conversation. If you express frustration, then it produces a response that would appear empathetic; if you are ebullient, it can echo some lighthearted repartee. Does this mean it “knows” or understands emotions? No. Instead, what ChatGPT does is recognize patterns in language as they reflect emotion, but it does not feel emotions. It doesn’t have the physiological and psychological mechanisms that underlie human emotion. Training ChatGPT to “be human” in this way would take more than feeding it datasets on emotional expression. It would rather require an architecture that allowed for some authentic experiences of emotions — that is, something beyond the scope of current developments in AI. Even if we can, in some way, simulate emotional reactions in AI, they will still be fundamentally different from human emotions, which are tied closely with consciousness, biology, and experience. Creativity Creativity- the mark of human intelligence: the ability to come up with new ideas, solve problems in innovative ways, and create art that vibrates at a deeply emotional or intellectual level. Is ChatGPT creative? In this respect, yes. ChatGPT can write, for instance, poems, short stories, code- whatever you want it to do. All the same, this creativity is based on the pattern rather than on actual thinking outside of the box that may reflect human creativity. Some of the most creative solutions or artistic expressions are often born out of personal experience, emotional ups and downs, moments of inspiration, those things that cannot easily be expressed and replicated in an algorithm. Although AI can generate creative content, it remains without experiential or emotional depth to inform human creativity. This means it offers content based on learned patterns rather than any personal experience or emotional journey. That is the main point of explaining why, despite impressive outputs, ChatGPT is fundamentally different from a human mind. Morality The other very important element of being human is morality, for which decisions are generally guided by the moral compass of and through upbringing, culture, experiences, and societal norms. More often than not, humans struggle with conflicting values, weigh the consequences of their deeds, and sometimes, change their beliefs over time. Could ChatGPT make moral decisions? Not at all. While it may be able to be programmed to adhere to certain guidelines of morality or avoid harmful content, its understanding of morality is superficial. It does not possess personal values or understand right and wrong in the way humans do. Instead, it responds based on patterns in data that often simply mimic moral reasoning without truly engaging in the complex, human process of ethical decision-making. It is a huge task to train ChatGPT to be human in this regard since even humans are unable to accept one sense of moral principle. Morality also relates to consciousness, which AI lacks. Though AI systems can follow rules, they are not able to understand the sense of moral dilemmas that define human experience. Memory and Experience Humans are a product of their memories and experiences. Every conversation, every relationship, and triumph and every failure leave an impression on who we are and how we approach the world. AI does not have any kind of real memory of past interactions unless it has been explicitly programmed to retain specific data. Each interaction with ChatGPT is unique, and while it can process historical data, it doesn’t “remember” interactions the same way a human does. It would give this form so that ChatGPT can learn through experiences, creating a sense of identity and personal history over time. Now, it does not necessarily imply that it should remember anything but interpret it to affect the subsequent actions; the same way humans do in life, growing and changing according to their past. Current AI models, including ChatGPT, do not have this ability. Though they might learn from big datasets, they lack personal experiences and do not change according to individual interaction. Because of this, for ChatGPT to be closer to “being human,” it requires something more than the capability to store and accumulate information but must also develop the capability to reflect on that same information and to change in meaningful ways relative to past experiences. Learning by Experience Human intelligence is reinforced by the ability to reflect upon mistakes and utilize those as learning experiences. We sometimes reflect on our actions, question our decisions, and make alterations in behavior based on our experiences. ChatGPT, though, can be trained through machine learning models, though it learns very differently compared to humans. Improvement of AI in machine learning is through wide training on various datasets and adjustment of weights and biases for better outcomes. However, the methodology itself is not like that of human learning based on trial and error and is contextual, emotionally and socially tuned, and based on personal experience and interactions with other humans. In contrast, the learning process of ChatGPT is abstract and data-driven. While AI may learn to make improvements in its responses, it is without a human ability to reflect on itself. Human learning isn’t just adding new information, but building that new information off of what we already know and thinking about what the implications are. Sometimes, what we learn is questioned. ChatGPT learns to do this task better at getting statistically likely responses without any understanding or reflection about what it is learning. The Future of AI So, can ChatGPT be made human? Currently, as of the technology, the answer is no. So, can ChatGPT be made human? Currently, as of the technology, the answer is no. So, can ChatGPT be made human? Currently, as of the technology, the answer is no. In full, that is. While still able to mimic and even generate human-like conversations, produce creative content, and maintain adherence to all sorts of ethical directives, AI continues to lack depth, consciousness, and emotional complexities of what it means to be human. However, the future of AI is no longer so static. Researchers are more and more pushing the frontiers of what could be done with AI: find new approaches to making machines more intelligent, adaptive, and even emotionally intelligent. Some believe that with domains such as neuro-symbolic AI-machine learning coupled with symbolic reasoning or even quantum computation, we may eventually build one day the essence of human-like cognition in an AI. But yet, all of this raises a basic question: Do we want AI to be human? The development of machines that could think and feel like humans raises profound questions in the realms of ethics and philosophy. Where do we draw that line between the machine and the person if AI becomes too human? Should AI have rights when it can feel? Where do we draw that line between the machine and the person if AI becomes too human? Should AI have rights when it can feel? These are questions that society will have to grapple with as AI continues its evolution.

Can We Train AI to Become More Human-Like?

Material10 is a next-generation game publisher developed by Co-Founder and CEO Maitham Mohamed. Material10 is focused on creating engaging multiplayer games with a new business model pioneering open economies, redefining the gaming industry by taking advantage of their unique business model to solve core problems in the traditional free-to-play model. Having raised over $2 million and put together a team with experience working on titles such as Ghost Recon, Tom Clancy's Rainbow Six Siege, and GTA, among others, Material10 is poised to make a major impact on the gaming industry. How does Material10 Work? Material10 got its start when the company recognized a need for a new business model in gaming monetization. Current models either charge an up-front purchase cost or rely on pay-to-win microtransactions. Recent trends in the industry have been attempting to combine these two common methods, much to the frustration of players. With advancements in blockchain technology, they realized they could tap into entirely new business models since it was now possible to create secondary markets where players could trade in-game virtual items with each other, and royalties/taxes could be enforced. They aim to challenge traditional models by focusing completely on secondary monetization, generating revenue by taxing player-to-player trades rather than selling in-game items directly. With this model, Material10 is optimizing for a robust player economy that operates far more independently than direct primary monetization. Barriers to Entry in Traditional Systems Most free-to-play models tend to create artificial barriers for players, such as paywalls and reliance on microtransactions to progress or gain a competitive edge in the game. These models often foster discontent and quickly decrease engagement. The best kinds of free-to-play models often rely on “skins,” which players can purchase for a simple cosmetic alteration. However, these models struggle to stay afloat without big spenders, falling flat if the game doesn’t take off. Material10’s model is a solution to the problems that have plagued free-to-play models since the beginning. Material10 simply makes all in-game items accessible for free, allowing players to engage deeply with the game without any financial barriers to play. The primary form of monetization will be a tax on player-to-player trades, creating a sustainable revenue model that emphasizes player engagement with both the game and each other. This system is designed to naturally create a player community and build overall satisfaction. An Experienced Vision for Accessibility As a gamer himself, Maitham recognized the potential of an in-game economy where the items were accessible to everyone. “Growing up, I spent a lot of my time playing video games and was such a huge FIFA Ultimate Team Player,” Maitham recalls. “The game had a deep economy where you could trade and purchase in-game players, but they were often quite expensive. I was only 12 years old, and obviously, my parents didn’t give me money for in-game items, especially after already buying the game for me, so I had to figure out a way to make money trading these in-game items within their markets within the game to make enough ‘FIFA coins’ so I could purchase my favorite players. This really gave me an intuitive understanding of how open economies within video games can work, and it’s exactly why I'm working on Material10.” Maitham’s Success with Material10 Maitham’s work has earned him a spot on Forbes 30 Under 30 in the Sports & Gaming category. His background in investment banking, corporate finance, and leading digital asset projects has provided him with the experience he needs to position Material10 as a leader in the gaming space. Maitham worked hard to raise funding for his new venture using strategic thinking and industry knowledge and secured $2 million from leading investors, including Blockchain Founders Fund, Big Brain Holdings, GSR, Trousdale Ventures, and Monoceros. New Model, Fresh Experience Building off of their unique and innovative approach to in-game economies, Material10 has now set about creating a fresh gameplay experience. The company is developing games like Project Fallen, a third-person shooter and real-time strategy lite experience where players control squads of soldiers while leveraging a drone-based aerial view. The game will emphasize pulling off tactical moves in the third-person while executing high-level strategies from above. The project will utilize Material10’s new business model, aiming to fundamentally change the free-to-play experience and realize the full potential of monetization. Material10 expects this approach to enhance player engagement while aligning with modern expectations of fairness and transparency in gaming, a far cry from traditional pay-to-win models. Leading Gaming into the Future Material10 aims to be a leader in next-generation gaming, continuously reinventing what is possible in their space with new economic systems and community engagement tools. The company is committed to creating immersive, player-driven experiences that oppose traditional pay-to-win models, redefining the gaming landscape with a focus on sustainability, player satisfaction, and long-term engagement. Material10 is a next-generation game publisher developed by Co-Founder and CEO Maitham Mohamed. Material10 is focused on creating engaging multiplayer games with a new business model pioneering open economies, redefining the gaming industry by taking advantage of their unique business model to solve core problems in the traditional free-to-play model. Having raised over $2 million and put together a team with experience working on titles such as Ghost Recon, Tom Clancy's Rainbow Six Siege , and GTA , among others, Material10 is poised to make a major impact on the gaming industry. Ghost Recon, Tom Clancy's Rainbow Six Siege GTA How does Material10 Work? Material10 got its start when the company recognized a need for a new business model in gaming monetization. Current models either charge an up-front purchase cost or rely on pay-to-win microtransactions. Recent trends in the industry have been attempting to combine these two common methods, much to the frustration of players. With advancements in blockchain technology, they realized they could tap into entirely new business models since it was now possible to create secondary markets where players could trade in-game virtual items with each other, and royalties/taxes could be enforced. They aim to challenge traditional models by focusing completely on secondary monetization, generating revenue by taxing player-to-player trades rather than selling in-game items directly. With this model, Material10 is optimizing for a robust player economy that operates far more independently than direct primary monetization. Barriers to Entry in Traditional Systems Most free-to-play models tend to create artificial barriers for players, such as paywalls and reliance on microtransactions to progress or gain a competitive edge in the game. These models often foster discontent and quickly decrease engagement. The best kinds of free-to-play models often rely on “skins,” which players can purchase for a simple cosmetic alteration. However, these models struggle to stay afloat without big spenders, falling flat if the game doesn’t take off. Material10’s model is a solution to the problems that have plagued free-to-play models since the beginning. Material10 simply makes all in-game items accessible for free, allowing players to engage deeply with the game without any financial barriers to play. The primary form of monetization will be a tax on player-to-player trades, creating a sustainable revenue model that emphasizes player engagement with both the game and each other. This system is designed to naturally create a player community and build overall satisfaction. An Experienced Vision for Accessibility As a gamer himself, Maitham recognized the potential of an in-game economy where the items were accessible to everyone. “Growing up, I spent a lot of my time playing video games and was such a huge FIFA Ultimate Team Player,” Maitham recalls. “The game had a deep economy where you could trade and purchase in-game players, but they were often quite expensive. I was only 12 years old, and obviously, my parents didn’t give me money for in-game items, especially after already buying the game for me, so I had to figure out a way to make money trading these in-game items within their markets within the game to make enough ‘FIFA coins’ so I could purchase my favorite players. This really gave me an intuitive understanding of how open economies within video games can work, and it’s exactly why I'm working on Material10.” Maitham’s Success with Material10 Maitham’s work has earned him a spot on Forbes 30 Under 30 in the Sports & Gaming category. His background in investment banking, corporate finance, and leading digital asset projects has provided him with the experience he needs to position Material10 as a leader in the gaming space. Maitham worked hard to raise funding for his new venture using strategic thinking and industry knowledge and secured $2 million from leading investors, including Blockchain Founders Fund, Big Brain Holdings, GSR, Trousdale Ventures, and Monoceros. New Model, Fresh Experience Building off of their unique and innovative approach to in-game economies, Material10 has now set about creating a fresh gameplay experience. The company is developing games like Project Fallen , a third-person shooter and real-time strategy lite experience where players control squads of soldiers while leveraging a drone-based aerial view. The game will emphasize pulling off tactical moves in the third-person while executing high-level strategies from above. Project Fallen The project will utilize Material10’s new business model, aiming to fundamentally change the free-to-play experience and realize the full potential of monetization. Material10 expects this approach to enhance player engagement while aligning with modern expectations of fairness and transparency in gaming, a far cry from traditional pay-to-win models. Leading Gaming into the Future Material10 aims to be a leader in next-generation gaming, continuously reinventing what is possible in their space with new economic systems and community engagement tools. The company is committed to creating immersive, player-driven experiences that oppose traditional pay-to-win models, redefining the gaming landscape with a focus on sustainability, player satisfaction, and long-term engagement.

Material10: The Startup That’s Breaking All the Rules in Gaming

Material10: الشركة الناشئة التي تكسر كل القواعد في مجال الألعاب

উপাদান 10: স্টার্টআপ যা গেমিংয়ের সমস্ত নিয়ম ভঙ্গ করছে

Material10: La startup que rompe todas las reglas en el mundo de los videojuegos

Material10: ゲーム業界のあらゆるルールを破るスタートアップ

10 medžiaga: visas žaidimų taisykles pažeidžiantis startuolis

Material10: Startup-ul care încalcă toate regulile în domeniul jocurilor

Material10: සූදු ක්‍රීඩාවේ සියලුම නීති කඩ කරන ආරම්භය

Material10: Bilawga Jabiya Dhamaan Shuruucda Ciyaaraha

Material10: Startupen som bryter mot alla spelregler

பொருள் 10: கேமிங்கில் உள்ள அனைத்து விதிகளையும் மீறும் தொடக்கம்

Material10: Оғоз, ки тамоми қоидаҳоро дар бозиҳо вайрон мекунад

Material10: O'yindagi barcha qoidalarni buzadigan startap

Machine learning experiments require extensive parametrization, including optimizer parameters, network architecture, and data augmentation. However, we strive for concise, readable code instead of a cumbersome 200 lines dedicated to argparse. Our goal is to focus on programming logic rather than threading new parameters through function signatures. Additionally, we seek a structure that is easily expandable without burdening the project, ensuring the reproducibility of experiments. Hydra offers a solution to these challenges. Below, you will find a basic guide on how to use it. What is Hydra? Hydra is a library with rich capabilities for managing configurations. The main site describes the name like this: “The key feature is the ability to dynamically create a hierarchical configuration by composition and override it through config files and the command line.The name Hydra comes from its ability to run multiple similar jobs - much like a Hydra with multiple heads.” But I have my own interpretation of its name. It is just a combination of some tools that work together impressively well! Hydra offers a seamless solution to the common headaches faced by ML engineers when attempting to replicate experiments. It elegantly replaces the need for argparse or YAML configurations, allowing access to parameters both from the command line and YAML files. Consider the pain points: Replicating experiments using argparse forces reliance on string inputs and complicates YAML config launches. Relying solely on YAML files for configuration leads to duplication and potential errors when only a single parameter needs alteration. Hydra addresses these issues by enabling dynamic configuration adjustments without the need for multiple bulky files or rigid command-line arguments. Furthermore, it simplifies the process of passing complex configurations, such as model architectures or functions, directly from the config file to the model. This capability eliminates the tedious task of manually feeding parameters into the model, streamlining the workflow and reducing the margin for error. Basic Setup Let's imagine the simplest setup: multiclass classification on MNIST using an MLP. We have a configuration and a training script. .
├── configs
│   └── config.yaml
└── main.py The main script can look like this. import torch
import torch.nn as nn
import torch.optim as optim
from torchvision import datasets, transforms
from torch.utils.data import DataLoader

import hydra
from omegaconf import DictConfig

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.model = nn.Sequential(
            nn.Linear(28*28, 128),
            nn.ReLU(),
            nn.Linear(128, 64),
            nn.ReLU(),
            nn.Linear(64, 10)
        )
    
    def forward(self, x):
        return self.model(x.view(-1, 28*28))

@hydra.main(version_base=None, config_path=None, config_name="config")
def main(cfg: DictConfig):
    # Load MNIST dataset
    transform = transforms.Compose([
        transforms.ToTensor(), 
        transforms.Normalize((0.5,), (0.5,))
    ])
    train_dataset = datasets.MNIST(root='./data', train=True, download=True, transform=transform)
    train_loader = DataLoader(train_dataset, batch_size=cfg.batch_size, shuffle=True)

    # Initialize the network, loss function, and optimizer
    model = Net()
    criterion = nn.CrossEntropyLoss()
    optimizer = optim.SGD(model.parameters(), lr=cfg.lr, momentum=cfg.momentum)

    # Train the network
    for epoch in range(cfg.epochs):  # loop over the dataset multiple times
        for i, (inputs, labels) in enumerate(train_loader, 0):
            optimizer.zero_grad()  # zero the parameter gradients
            outputs = model(inputs)  # forward pass
            loss = criterion(outputs, labels)  # calculate loss
            loss.backward()  # backward pass
            optimizer.step()  # optimize

    print('Finished Training')

if __name__ == "__main__":
    main() The configuration has the following structure. It is the default, and parameters from it will be used by default if not specified otherwise. # conf/config.yaml
batch_size: 64
lr: 0.01
momentum: 0.9
epochs: 1 To start training you can use: python main.py And for changing parameters, you can do it not only from YAML but also from cli: python main.py lr=0.03 If you use it, parameters from cli override the first parameters. As you can see, there is no argparse or additional middleware. It's easy to change parameters from the command line. Pass Class Object With Hydra We see that in class Net, the architecture can be highly customizable. Even more, we can use also another net for this, e.g. CNN. But we do not want to change our pipeline. Hydra can construct almost any Python object with specified parameters. Let’s describe our net in another way, using YAML. model:
  _target_: torch.nn.Sequential
  layers:
    - _target_: torch.nn.Flatten
    - _target_: torch.nn.Linear
      in_features: 784  # 28x28 images are flattened into 784
      out_features: 128
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.Linear
      in_features: 128
      out_features: 64
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.Linear
      in_features: 64
      out_features: 10 and for CNN _target_: torch.nn.Sequential
  layers:
    - _target_: torch.nn.Conv2d
      in_channels: 1  # MNIST images are grayscale, so 1 input channel
      out_channels: 32  # Number of output channels
      kernel_size: 3  # Size of the convolutional kernel
      stride: 1
      padding: 1
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.MaxPool2d
      kernel_size: 2  # Pooling window size
      stride: 2
    - _target_: torch.nn.Conv2d
      in_channels: 32
      out_channels: 64
      kernel_size: 3
      stride: 1
      padding: 1
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.MaxPool2d
      kernel_size: 2
      stride: 2
    - _target_: torch.nn.Flatten  # Flatten the output for the fully connected layer
    - _target_: torch.nn.Linear
      in_features: 7*7*64  # Size after convolutions and pooling
      out_features: 128
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.Dropout
      p: 0.5  # Dropout rate
    - _target_: torch.nn.Linear
      in_features: 128
      out_features: 10  # Number of classes in MNIST Pass Function Through Config Sometimes, our parameters can also function. For example, How to Access Config Without Decorating Main - e.g., in ipython Notebook. So, you can use Compose API. from hydra import compose, initialize from omegaconf import OmegaConf

with initialize(version_base=None, config_path="conf", job_name="run_0001"):
  cfg = compose(config_name="config", overrides= ["parameter=value"])

print (OmegaConf.to_yaml(cfg)) Multi Runs For example, you want to find some good hyperparam and Hydra allows a lot for that. Basically, you can launch several experiments sequentially like. python train_model.py -m "batch_size=16,32,64" It launches sequentially, but it is easy to extend to parallel if turned on some launcher like joblib. python train_model.py -m "batch_size=16,32,64 hydra/launcher=joblib" Run All Experiments From the Folder for config_file in configs/*.yaml; do python main.py --config-name="${config_file}"; done Resources hydra - https://hydra.cc/
good template for DL - https://github.com/ashleve/lightning-hydra-template/tree/main Machine learning experiments require extensive parametrization, including optimizer parameters, network architecture, and data augmentation. However, we strive for concise, readable code instead of a cumbersome 200 lines dedicated to argparse. Our goal is to focus on programming logic rather than threading new parameters through function signatures. Additionally, we seek a structure that is easily expandable without burdening the project, ensuring the reproducibility of experiments. Hydra offers a solution to these challenges. Below, you will find a basic guide on how to use it. What is Hydra? Hydra is a library with rich capabilities for managing configurations. The main site describes the name like this: “The key feature is the ability to dynamically create a hierarchical configuration by composition and override it through config files and the command line.The name Hydra comes from its ability to run multiple similar jobs - much like a Hydra with multiple heads.” “The key feature is the ability to dynamically create a hierarchical configuration by composition and override it through config files and the command line. The name Hydra comes from its ability to run multiple similar jobs - much like a Hydra with multiple heads.” But I have my own interpretation of its name. It is just a combination of some tools that work together impressively well! Hydra offers a seamless solution to the common headaches faced by ML engineers when attempting to replicate experiments. It elegantly replaces the need for argparse or YAML configurations, allowing access to parameters both from the command line and YAML files. Consider the pain points: Replicating experiments using argparse forces reliance on string inputs and complicates YAML config launches. Replicating experiments using argparse forces reliance on string inputs and complicates YAML config launches. Relying solely on YAML files for configuration leads to duplication and potential errors when only a single parameter needs alteration. Relying solely on YAML files for configuration leads to duplication and potential errors when only a single parameter needs alteration. Hydra addresses these issues by enabling dynamic configuration adjustments without the need for multiple bulky files or rigid command-line arguments. Furthermore, it simplifies the process of passing complex configurations, such as model architectures or functions, directly from the config file to the model. This capability eliminates the tedious task of manually feeding parameters into the model, streamlining the workflow and reducing the margin for error. Basic Setup Let's imagine the simplest setup: multiclass classification on MNIST using an MLP. We have a configuration and a training script. .
├── configs
│   └── config.yaml
└── main.py .
├── configs
│   └── config.yaml
└── main.py The main script can look like this. import torch
import torch.nn as nn
import torch.optim as optim
from torchvision import datasets, transforms
from torch.utils.data import DataLoader

import hydra
from omegaconf import DictConfig

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.model = nn.Sequential(
            nn.Linear(28*28, 128),
            nn.ReLU(),
            nn.Linear(128, 64),
            nn.ReLU(),
            nn.Linear(64, 10)
        )
    
    def forward(self, x):
        return self.model(x.view(-1, 28*28))

@hydra.main(version_base=None, config_path=None, config_name="config")
def main(cfg: DictConfig):
    # Load MNIST dataset
    transform = transforms.Compose([
        transforms.ToTensor(), 
        transforms.Normalize((0.5,), (0.5,))
    ])
    train_dataset = datasets.MNIST(root='./data', train=True, download=True, transform=transform)
    train_loader = DataLoader(train_dataset, batch_size=cfg.batch_size, shuffle=True)

    # Initialize the network, loss function, and optimizer
    model = Net()
    criterion = nn.CrossEntropyLoss()
    optimizer = optim.SGD(model.parameters(), lr=cfg.lr, momentum=cfg.momentum)

    # Train the network
    for epoch in range(cfg.epochs):  # loop over the dataset multiple times
        for i, (inputs, labels) in enumerate(train_loader, 0):
            optimizer.zero_grad()  # zero the parameter gradients
            outputs = model(inputs)  # forward pass
            loss = criterion(outputs, labels)  # calculate loss
            loss.backward()  # backward pass
            optimizer.step()  # optimize

    print('Finished Training')

if __name__ == "__main__":
    main() import torch
import torch.nn as nn
import torch.optim as optim
from torchvision import datasets, transforms
from torch.utils.data import DataLoader

import hydra
from omegaconf import DictConfig

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.model = nn.Sequential(
            nn.Linear(28*28, 128),
            nn.ReLU(),
            nn.Linear(128, 64),
            nn.ReLU(),
            nn.Linear(64, 10)
        )
    
    def forward(self, x):
        return self.model(x.view(-1, 28*28))

@hydra.main(version_base=None, config_path=None, config_name="config")
def main(cfg: DictConfig):
    # Load MNIST dataset
    transform = transforms.Compose([
        transforms.ToTensor(), 
        transforms.Normalize((0.5,), (0.5,))
    ])
    train_dataset = datasets.MNIST(root='./data', train=True, download=True, transform=transform)
    train_loader = DataLoader(train_dataset, batch_size=cfg.batch_size, shuffle=True)

    # Initialize the network, loss function, and optimizer
    model = Net()
    criterion = nn.CrossEntropyLoss()
    optimizer = optim.SGD(model.parameters(), lr=cfg.lr, momentum=cfg.momentum)

    # Train the network
    for epoch in range(cfg.epochs):  # loop over the dataset multiple times
        for i, (inputs, labels) in enumerate(train_loader, 0):
            optimizer.zero_grad()  # zero the parameter gradients
            outputs = model(inputs)  # forward pass
            loss = criterion(outputs, labels)  # calculate loss
            loss.backward()  # backward pass
            optimizer.step()  # optimize

    print('Finished Training')

if __name__ == "__main__":
    main() The configuration has the following structure. It is the default, and parameters from it will be used by default if not specified otherwise. # conf/config.yaml
batch_size: 64
lr: 0.01
momentum: 0.9
epochs: 1 # conf/config.yaml
batch_size: 64
lr: 0.01
momentum: 0.9
epochs: 1 To start training you can use: python main.py python main.py And for changing parameters, you can do it not only from YAML but also from cli: python main.py lr=0.03 python main.py lr=0.03 If you use it, parameters from cli override the first parameters. As you can see, there is no argparse or additional middleware. It's easy to change parameters from the command line. Pass Class Object With Hydra We see that in class Net, the architecture can be highly customizable. Even more, we can use also another net for this, e.g. CNN. But we do not want to change our pipeline. Hydra can construct almost any Python object with specified parameters. Let’s describe our net in another way, using YAML. model:
  _target_: torch.nn.Sequential
  layers:
    - _target_: torch.nn.Flatten
    - _target_: torch.nn.Linear
      in_features: 784  # 28x28 images are flattened into 784
      out_features: 128
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.Linear
      in_features: 128
      out_features: 64
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.Linear
      in_features: 64
      out_features: 10 model:
  _target_: torch.nn.Sequential
  layers:
    - _target_: torch.nn.Flatten
    - _target_: torch.nn.Linear
      in_features: 784  # 28x28 images are flattened into 784
      out_features: 128
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.Linear
      in_features: 128
      out_features: 64
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.Linear
      in_features: 64
      out_features: 10 and for CNN _target_: torch.nn.Sequential
  layers:
    - _target_: torch.nn.Conv2d
      in_channels: 1  # MNIST images are grayscale, so 1 input channel
      out_channels: 32  # Number of output channels
      kernel_size: 3  # Size of the convolutional kernel
      stride: 1
      padding: 1
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.MaxPool2d
      kernel_size: 2  # Pooling window size
      stride: 2
    - _target_: torch.nn.Conv2d
      in_channels: 32
      out_channels: 64
      kernel_size: 3
      stride: 1
      padding: 1
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.MaxPool2d
      kernel_size: 2
      stride: 2
    - _target_: torch.nn.Flatten  # Flatten the output for the fully connected layer
    - _target_: torch.nn.Linear
      in_features: 7*7*64  # Size after convolutions and pooling
      out_features: 128
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.Dropout
      p: 0.5  # Dropout rate
    - _target_: torch.nn.Linear
      in_features: 128
      out_features: 10  # Number of classes in MNIST _target_: torch.nn.Sequential
  layers:
    - _target_: torch.nn.Conv2d
      in_channels: 1  # MNIST images are grayscale, so 1 input channel
      out_channels: 32  # Number of output channels
      kernel_size: 3  # Size of the convolutional kernel
      stride: 1
      padding: 1
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.MaxPool2d
      kernel_size: 2  # Pooling window size
      stride: 2
    - _target_: torch.nn.Conv2d
      in_channels: 32
      out_channels: 64
      kernel_size: 3
      stride: 1
      padding: 1
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.MaxPool2d
      kernel_size: 2
      stride: 2
    - _target_: torch.nn.Flatten  # Flatten the output for the fully connected layer
    - _target_: torch.nn.Linear
      in_features: 7*7*64  # Size after convolutions and pooling
      out_features: 128
    - _target_: torch.nn.ReLU
    - _target_: torch.nn.Dropout
      p: 0.5  # Dropout rate
    - _target_: torch.nn.Linear
      in_features: 128
      out_features: 10  # Number of classes in MNIST Pass Function Through Config Sometimes, our parameters can also function. For example, How to Access Config Without Decorating Main - e.g., in ipython Notebook. So, you can use Compose API . Compose API from hydra import compose, initialize from omegaconf import OmegaConf

with initialize(version_base=None, config_path="conf", job_name="run_0001"):
  cfg = compose(config_name="config", overrides= ["parameter=value"])

print (OmegaConf.to_yaml(cfg)) from hydra import compose, initialize from omegaconf import OmegaConf

with initialize(version_base=None, config_path="conf", job_name="run_0001"):
  cfg = compose(config_name="config", overrides= ["parameter=value"])

print (OmegaConf.to_yaml(cfg)) Multi Runs For example, you want to find some good hyperparam and Hydra allows a lot for that. Basically, you can launch several experiments sequentially like. python train_model.py -m "batch_size=16,32,64" python train_model.py -m "batch_size=16,32,64" It launches sequentially, but it is easy to extend to parallel if turned on some launcher like joblib. python train_model.py -m "batch_size=16,32,64 hydra/launcher=joblib" python train_model.py -m "batch_size=16,32,64 hydra/launcher=joblib" Run All Experiments From the Folder for config_file in configs/*.yaml; do python main.py --config-name="${config_file}"; done for config_file in configs/*.yaml; do python main.py --config-name="${config_file}"; done Resources hydra - https://hydra.cc/ good template for DL - https://github.com/ashleve/lightning-hydra-template/tree/main hydra - https://hydra.cc/ https://hydra.cc/ good template for DL - https://github.com/ashleve/lightning-hydra-template/tree/main https://github.com/ashleve/lightning-hydra-template/tree/main

How to Configure Experiments With Hydra - From an ML Enginner Perspective

Authors:
(1) Yinwei Dai, Princeton University (Equal contributions);
(2) Rui Pan, Princeton University (Equal contributions);
(3) Anand Iyer, Georgia Institute of Technology;
(4) Ravi Netravali, Georgia Institute of Technology. Table of Links Abstract and 1 Introduction 2 Background and Motivation and 2.1 Model Serving Platforms 2.2 Early-Exit Models 2.3 Challenges 3 Design 3.1 Preparing Models with Early Exits 3.2 Accuracy-Aware Threshold Tuning 3.3 Latency-Focused Ramp Adjustments 4 Implementation 5 Evaluation and 5.1 Methodology 5.2 Overall Results 5.3 Comparison with Existing EE Strategies 5.4 Microbenchmarks 6 Additional Related Work 7 Conclusion, References, Appendix 3.2 Accuracy-Aware Threshold Tuning To avoid accuracy drops as workload characteristics change over time, Apparate’s controller employs frequent and fast tuning of thresholds for already-enabled ramps. The reason is that threshold tuning for any set of ramps is sufficient to ensure that user-specified accuracy constraints are not violated – at the extreme, all thresholds could be set to zero, which precludes any early exiting. Altering only the set of active ramps fails to provide this property. To enable threshold tuning, as requests pass through a model, Apparate’s controller continually records exiting information at each active ramp, as well as the final result that the original model predicts. More precisely, for each ramp, Apparate records the result (e.g., classification label) with the lowest error rate, i.e., the highest-confidence result for the ramp, even if the error exceeds the ramp’s threshold (precluding exiting). Importantly, since inputs always pass fully through models with Apparate, this information is recorded for all inputs at each active ramp, irrespective of upstream exiting decisions. This is paramount since the information serves not only as a signal for when to tune thresholds, but also provides guidance for how to do so. Triggering tuning. Apparate maintains an average achieved accuracy over the past 16 samples by comparing exiting results with the deployed configuration to results of the original model. Threshold tuning is triggered any time a window’s accuracy falls below the user-specified accuracy constraint. The threshold tuning process (described below) runs asynchronously on a CPU, without any disruptions to ongoing jobs. This is possible since thresholds are anyway enforced only by Apparate’s controller; GPUs are agnostic to threshold values, and instead simply stream ramp results to the Apparate controller which determines exiting decisions. Evaluating threshold configurations. Threshold tuning requires insight into how any alterations to active ramp thresholds would affect overall model exiting behavior (and accuracies). The aforementioned per-request, per-ramp monitoring information grants this visibility, enabling Apparate to rapidly evaluate any threshold values across active ramps without requiring additional inference, and while accounting for inter-ramp dependencies. In particular, to evaluate new threshold values, Apparate simply identifies the earliest ramp whose top prediction now has an error rate below its threshold. Comparing these results with those of the original model indicates the achieved accuracy for the new configuration; latency wins for these exit patterns are computed using the one-time profiling data described in §3.3. Instead, Apparate employs a greedy heuristic (Algorithm 1 in §A) that leverages a fundamental property of EEs when evaluated against an original model: higher thresholds result in monotonic decreases in accuracy and monotonic increases in latency savings. This prunes the space of threshold values to consider by providing a clear boundary in the Rdimensional space that separates configurations that are sufficiently accurate from those that are not. Additionally, for accurate configurations, maximum latency savings must fall on that boundary. Figure 10 illustrates this. These properties inform Apparate’s hill climbing strategy [47] for threshold tuning. Starting with threshold values of 0 for each active ramp, and a step size of 0.1, threshold tuning runs in a series of (incremental) exploration rounds. In each round, we increase the threshold of each ramp in isolation (leaving the others unchanged), and evaluate the achieved accuracy and latency savings as described above. Apparate then chooses the single ramp threshold change that delivered the largest additional latency savings per unit of additional accuracy loss. This process repeats until no ramp’s threshold can be increased without an accuracy violation. To enhance this process, Apparate follows a multiplicative increase, multiplicative decrease policy on step sizes to balance search speed and granularity. Specifically, each time a step increase results in an accuracy violation, Apparate halves that ramp’s step size for subsequent rounds to hone in on the boundary at fine granularity; step sizes are lower-bounded at 0.01. Conversely, selection of a ramp for threshold alteration suggests a potentially promising path of exploration; in this case, for a speedup, Apparate doubles that ramp’s step size for the following round. Overall, as shown in Figure 11, Apparate’s threshold tuning algorithm runs up to 3 orders of magnitude faster than a pure grid search (11.9ms vs. 3.0s on average). Note that these results maximally parallelize grid search across a 30-core machine. Further, selected threshold values achieve within 0-3.8% of the latency savings of the optimal configurations. This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. Authors: (1) Yinwei Dai, Princeton University (Equal contributions); (2) Rui Pan, Princeton University (Equal contributions); (3) Anand Iyer, Georgia Institute of Technology; (4) Ravi Netravali, Georgia Institute of Technology. Authors: Authors: (1) Yinwei Dai, Princeton University (Equal contributions); (2) Rui Pan, Princeton University (Equal contributions); (3) Anand Iyer, Georgia Institute of Technology; (4) Ravi Netravali, Georgia Institute of Technology. Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 Background and Motivation and 2.1 Model Serving Platforms 2 Background and Motivation and 2.1 Model Serving Platforms 2.2 Early-Exit Models 2.2 Early-Exit Models 2.3 Challenges 2.3 Challenges 3 Design 3 Design 3.1 Preparing Models with Early Exits 3.1 Preparing Models with Early Exits 3.2 Accuracy-Aware Threshold Tuning 3.2 Accuracy-Aware Threshold Tuning 3.3 Latency-Focused Ramp Adjustments 3.3 Latency-Focused Ramp Adjustments 4 Implementation 4 Implementation 5 Evaluation and 5.1 Methodology 5 Evaluation and 5.1 Methodology 5.2 Overall Results 5.2 Overall Results 5.3 Comparison with Existing EE Strategies 5.3 Comparison with Existing EE Strategies 5.4 Microbenchmarks 5.4 Microbenchmarks 6 Additional Related Work 6 Additional Related Work 7 Conclusion, References, Appendix 7 Conclusion, References, Appendix 3.2 Accuracy-Aware Threshold Tuning To avoid accuracy drops as workload characteristics change over time, Apparate’s controller employs frequent and fast tuning of thresholds for already-enabled ramps. The reason is that threshold tuning for any set of ramps is sufficient to ensure that user-specified accuracy constraints are not violated – at the extreme, all thresholds could be set to zero, which precludes any early exiting. Altering only the set of active ramps fails to provide this property. To enable threshold tuning, as requests pass through a model, Apparate’s controller continually records exiting information at each active ramp, as well as the final result that the original model predicts. More precisely, for each ramp, Apparate records the result (e.g., classification label) with the lowest error rate, i.e., the highest-confidence result for the ramp, even if the error exceeds the ramp’s threshold (precluding exiting). Importantly, since inputs always pass fully through models with Apparate, this information is recorded for all inputs at each active ramp, irrespective of upstream exiting decisions. This is paramount since the information serves not only as a signal for when to tune thresholds, but also provides guidance for how to do so. Triggering tuning . Apparate maintains an average achieved accuracy over the past 16 samples by comparing exiting results with the deployed configuration to results of the original model. Threshold tuning is triggered any time a window’s accuracy falls below the user-specified accuracy constraint. The threshold tuning process (described below) runs asynchronously on a CPU, without any disruptions to ongoing jobs. This is possible since thresholds are anyway enforced only by Apparate’s controller; GPUs are agnostic to threshold values, and instead simply stream ramp results to the Apparate controller which determines exiting decisions. Triggering tuning Evaluating threshold configurations. Threshold tuning requires insight into how any alterations to active ramp thresholds would affect overall model exiting behavior (and accuracies). The aforementioned per-request, per-ramp monitoring information grants this visibility, enabling Apparate to rapidly evaluate any threshold values across active ramps without requiring additional inference, and while accounting for inter-ramp dependencies. In particular, to evaluate new threshold values, Apparate simply identifies the earliest ramp whose top prediction now has an error rate below its threshold. Comparing these results with those of the original model indicates the achieved accuracy for the new configuration; latency wins for these exit patterns are computed using the one-time profiling data described in §3.3. Evaluating threshold configurations. Instead, Apparate employs a greedy heuristic (Algorithm 1 in §A) that leverages a fundamental property of EEs when evaluated against an original model: higher thresholds result in monotonic decreases in accuracy and monotonic increases in latency savings. This prunes the space of threshold values to consider by providing a clear boundary in the Rdimensional space that separates configurations that are sufficiently accurate from those that are not. Additionally, for accurate configurations, maximum latency savings must fall on that boundary. Figure 10 illustrates this. These properties inform Apparate’s hill climbing strategy [47] for threshold tuning. Starting with threshold values of 0 for each active ramp, and a step size of 0.1, threshold tuning runs in a series of (incremental) exploration rounds. In each round, we increase the threshold of each ramp in isolation (leaving the others unchanged), and evaluate the achieved accuracy and latency savings as described above. Apparate then chooses the single ramp threshold change that delivered the largest additional latency savings per unit of additional accuracy loss. This process repeats until no ramp’s threshold can be increased without an accuracy violation. To enhance this process, Apparate follows a multiplicative increase, multiplicative decrease policy on step sizes to balance search speed and granularity. Specifically, each time a step increase results in an accuracy violation, Apparate halves that ramp’s step size for subsequent rounds to hone in on the boundary at fine granularity; step sizes are lower-bounded at 0.01. Conversely, selection of a ramp for threshold alteration suggests a potentially promising path of exploration; in this case, for a speedup, Apparate doubles that ramp’s step size for the following round. Overall, as shown in Figure 11, Apparate’s threshold tuning algorithm runs up to 3 orders of magnitude faster than a pure grid search (11.9ms vs. 3.0s on average). Note that these results maximally parallelize grid search across a 30-core machine. Further, selected threshold values achieve within 0-3.8% of the latency savings of the optimal configurations. This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. available on arxiv

Stories, News & Mentions about ALTERA

Altera’s Stories on HackerNoon