{"product_id":"140202","title":"The structure of generative AI ","description":"\u003ccenter\u003e\u003cdiv style=\"text-align:center\"\u003e\u003cimg src=\"https:\/\/tmgdisk01.cafe24.com\/images\/vs\/4172\/sv\/3jYGWruL1RJ3SXRgs26M7UnSFqgHWT.png?v=1765078695\" style=\"max-width:100%;max-height:10px\"\u003e\u003c\/div\u003e\u003c\/center\u003e\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\u003ccenter\u003e\n\n\u003cdiv style=\"width:95%\"\u003e\n\n\u003cdiv style=\"text-align:center;font-size:30px;font-weight:bolder;line-height:1.6em\"\u003e The structure of generative AI \u003c\/div\u003e\n\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\n\u003cdiv style=\"border-bottom:1px;border-bottom-style:dotted;border-color:;padding-bottom:20px\"\u003e\u003ccenter\u003e\u003ctable align=\"center\" width=\"100%\"\u003e\u003ctbody style=\"border:0px\"\u003e\n\n\u003ctr\u003e\u003ctd align=\"center\" style=\"line-height:1.2em;text-align:center;font-size:18px;color:black;font-weight:bold;padding-bottom:20px;\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\n\n\u003ctr\u003e\u003ctd style=\"text-align:center\"\u003e\u003cimg src=\"https:\/\/image.yes24.com\/goods\/145386781\/XL\" style=\"max-width:100%;height:auto\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\n\n\n\u003c\/tbody\u003e\u003c\/table\u003e\u003c\/center\u003e\u003c\/div\u003e\n\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\n\u003cdiv style=\"width:95%;{split_style6}padding-top:20px;padding-bottom:20px\"\u003e\n\n\u003cdiv style=\"text-align:left;font-size:16px;font-weight:bold;padding-bottom:20px\"\u003e Description \u003c\/div\u003e\n\n\u003cdiv style=\"text-align:left;word-break:break-all;font-size:14px;line-height:1.6em;\"\u003e\n\n\u003cdiv\u003e\u003ch5\u003e \u003cb\u003eBook Introduction\u003c\/b\u003e\n\u003c\/h5\u003e\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv\u003e\u003cdiv\u003e \u003cb\u003eThe core principles of generative AI data generation technology that allows for easy reading and understanding without formulas.\u003c\/b\u003e\u003cbr\u003e\u003cbr\u003e Daisuke Okanohara, Japan's top AI expert, wrote this book, explaining the structure of generative AI using only text and pictures, without equations, so that even non-experts can understand it. \u003cbr\u003eFrom the history of generative AI to flow, diffusion models, flow matching, and optimal transportation, let's properly understand the core of generative AI, which is at the center of today's IT, with a friendly commentary by a veteran author who is praised for \"providing rich context surrounding the technology.\"\u003cbr\u003e\n\u003c\/div\u003e\u003c\/div\u003e\n\u003cdiv\u003e\u003cul\u003e\u003cli\u003e You can preview some of the book's contents.\u003cbr\u003e \u003cspan\u003ePreview\u003c\/span\u003e\n\n\u003c\/li\u003e\u003c\/ul\u003e\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cbr\u003e\u003cdiv\u003e\u003ch5\u003e \u003cb\u003eindex\u003c\/b\u003e\n\u003c\/h5\u003e\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv\u003e Translator's Preface ix\u003cbr\u003e To Korean readers x\u003cbr\u003e Preface xi\u003cbr\u003e\u003cbr\u003e \u003cb\u003eCHAPTER 1 Generative AI 1\u003c\/b\u003e\u003cbr\u003e\u003cbr\u003e What is Generative AI? Part 1\u003cbr\u003e Creation according to instructions and conditions 2\u003cbr\u003e Ability to create data that was previously difficult to create 3\u003cbr\u003e From Rule-Based to Machine Learning 5\u003cbr\u003e Generation tasks are a particularly difficult machine learning problem 7.\u003cbr\u003e Generating data is like finding an island in a vast ocean.\u003cbr\u003e Vast and Strange High-Dimensional Space 11\u003cbr\u003e Generation has more than one correct output 13\u003cbr\u003e Manifold Hypothesis: Data in Low Dimensions 15 \u003cbr\u003eSymmetry: Data that is invariant to transformations 18\u003cbr\u003e Composition: Data made up of a combination of multiple parts 20\u003cbr\u003e [COLUMN] Are data's characteristics provided by humans, or are they self-learning? 21\u003cbr\u003e Summary 22\u003cbr\u003e\u003cbr\u003e \u003cb\u003eCHAPTER 2: A History of Generative AI 23\u003c\/b\u003e\u003cbr\u003e\u003cbr\u003e Mechanisms of Memory 23\u003cbr\u003e From the Easing model to the Hopfield network 24\u003cbr\u003e Energy-based model 28\u003cbr\u003e Energy-Based Model 29 for Naturally Realizing Associative Memory\u003cbr\u003e The Relationship Between Energy and Probability: Boltzmann Distribution 31\u003cbr\u003e Principles of the Langevin Monte Carlo Method 32\u003cbr\u003e 33 Fatal Problems with Energy-Based Models\u003cbr\u003e [COLUMN] The Real World is a Giant Simulator 34\u003cbr\u003e Distribution function 35 that governs information throughout space\u003cbr\u003e Data generated from hidden information 37\u003cbr\u003e Awareness is required for creation 38\u003cbr\u003e Variational Autoencoder (VAE) 40\u003cbr\u003e Problem 42 of the Latent Variable Model\u003cbr\u003e [COLUMN] Generative Adversarial Networks (GANs) 43\u003cbr\u003e [COLUMN] Autoregressive Model 43\u003cbr\u003e [COLUMN] 2024 Nobel Prize 44\u003cbr\u003e Summary 45\u003cbr\u003e \u003cbr\u003e\u003cb\u003eCHAPTER 3 Creating Using Flow 47\u003c\/b\u003e\u003cbr\u003e\u003cbr\u003e Floran 47\u003cbr\u003e Continuity Equation: Matter Does Not Suddenly Disappear or Warp 49\u003cbr\u003e Complex probability distributions created using flow 51\u003cbr\u003e Flow-based model 53 that does not require a distribution function\u003cbr\u003e Normalization Flow and Continuous Normalization Flow 55\u003cbr\u003e Learning to maximize the likelihood obtained along the flow 55\u003cbr\u003e Generate data according to flow 57\u003cbr\u003e Flow 58: Decomposing complex generation problems into simpler sub-generation problems\u003cbr\u003e Flow Modeling 60\u003cbr\u003e Flow result calculation 62\u003cbr\u003e Normalization Flow Challenge 64\u003cbr\u003e Summary 65\u003cbr\u003e\u003cbr\u003e \u003cb\u003eCHAPTER 4 Diffusion Models and Flow Matching 67\u003c\/b\u003e\u003cbr\u003e\u003cbr\u003e Discovery of the Diffusion Model 67\u003cbr\u003e General diffusion phenomenon 68\u003cbr\u003e [COLUMN] Brownian Motion 69\u003cbr\u003e The diffusion model is 70\u003cbr\u003e Flow created by the diffusion process = Score 72\u003cbr\u003e The Relationship Between Score and Energy 73\u003cbr\u003e Score 74, which changes with time\u003cbr\u003e Denoising Score Matching 76\u003cbr\u003e Simulation-free learning is only available to some 78\u003cbr\u003e Summary of Learning and Generation by Diffusion Models 79 \u003cbr\u003eCharacteristics of flow generated by diffusion models 79\u003cbr\u003e Relationship between diffusion models and latent variable models 80\u003cbr\u003e Automatically learning phylogenetic trees for data generation 81\u003cbr\u003e The diffusion model is an energy-based model 82\u003cbr\u003e The diffusion model is a generative model using flow 82\u003cbr\u003e Flow Matching: Complex Flow 83 created by collecting flows\u003cbr\u003e Optimal Transportation 83\u003cbr\u003e Generation 85 using optimal transport\u003cbr\u003e Finding the optimal transport directly is too computationally intensive 85\u003cbr\u003e Learning Flow Matching 86\u003cbr\u003e The evolution of flow matching 88\u003cbr\u003e Conditional generation is realized with conditional flow 88\u003cbr\u003e Latent Diffusion Model: Transforming Original Data into Latent Space to Improve Quality 90\u003cbr\u003e Summary 91\u003cbr\u003e\u003cbr\u003e \u003cb\u003eCHAPTER 5: THE FUTURE PROSPECTS OF FLOW-BASED TECHNOLOGY 93\u003c\/b\u003e\u003cbr\u003e\u003cbr\u003e Solving the Mystery of Generalization 93\u003cbr\u003e Generation 95 with symmetry in mind\u003cbr\u003e Attention Mechanism and Flow 96\u003cbr\u003e Numerical Optimization by Flow 96\u003cbr\u003e Generating discrete data such as language 97\u003cbr\u003e 99 Contact with the brain's computational mechanisms\u003cbr\u003e The Future of Creation by Flow 99\u003cbr\u003e \u003cbr\u003e\u003cb\u003eAPPENDIX A Machine Learning Keywords 101\u003c\/b\u003e\u003cbr\u003e\u003cbr\u003e Probability and Generative Models 101\u003cbr\u003e Maximum likelihood method 102\u003cbr\u003e Machine Learning 103\u003cbr\u003e Machine Learning Mechanisms 104\u003cbr\u003e Parameter tuning = learning 105\u003cbr\u003e Neural Network 106\u003cbr\u003e Generalization 106: Obtaining rules applicable to infinite data from finite training data\u003cbr\u003e\u003cbr\u003e \u003cb\u003eAPPENDIX B REFERENCES 109\u003c\/b\u003e\u003cbr\u003e\u003cbr\u003e Chapter 2, 110\u003cbr\u003e Chapter 3, 112\u003cbr\u003e Chapter 4, 112\u003cbr\u003e Chapter 5, 114\u003cbr\u003e\u003cbr\u003e Search 117\u003c\/div\u003e\n\u003cdiv\u003e\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cbr\u003e\u003cdiv\u003e\u003ch5\u003e \u003cb\u003eDetailed image\u003c\/b\u003e \u003c\/h5\u003e\u003c\/div\u003e\n\u003cdiv\u003e\u003cdiv\u003e\u003cimg src=\"https:\/\/image.yes24.com\/momo\/TopCate5267\/MidCate004\/526637671.jpg\" border=\"0\" alt=\"Detailed Image 1\"\u003e\u003c\/div\u003e\u003c\/div\u003e\n\u003cbr\u003e\u003cdiv\u003e\u003ch5\u003e \u003cb\u003eInto the book\u003c\/b\u003e\n\u003c\/h5\u003e\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv\u003e\n\u003cdiv\u003e As problems become more complex, rules-based solutions become more difficult.\u003cbr\u003e The generative work discussed in this book is exactly that kind of work.\u003cbr\u003e \/ For example, let's say we want to generate an image given the instruction 'Silhouette of a dog and its owner running next to the waves on a beach at sunset'.\u003cbr\u003e In this case, you should tell them in advance what color the sunset sky is, what shape the waves are, and what the possible movements of the dog and owner are.\u003cbr\u003e We also need to teach them what happens when they are combined. \u003cbr\u003eThe sunset should be reflected in the color of the sea, and that reflected light should also affect the color of dogs and people.\u003cbr\u003e According to the laws of physics, the silhouettes of the dog and owner should be opposite the sun.\u003cbr\u003e\u003cbr\u003e --- p.6\u003cbr\u003e\u003cbr\u003e Let's review probability distributions.\u003cbr\u003e A probability distribution assigns a probability greater than or equal to 0 to each possible event.\u003cbr\u003e And the sum of the probabilities assigned to all events must be exactly 1.\u003cbr\u003e For example, if you roll a die and it tells you what will happen, it is a probability distribution with a probability of 1\/6.\u003cbr\u003e Also, the probability that the weather will be sunny, cloudy, or rainy tomorrow will have probability distributions such as 1\/2, 1\/3, and 1\/6, respectively.\u003cbr\u003e\u003cbr\u003e --- p.35\u003cbr\u003e\u003cbr\u003e There are various flows around us, such as air flow and water flow.\u003cbr\u003e Generally, the state of matter is classified into three types: solid, liquid, and gas, depending on temperature and pressure, and flow can be seen in liquid and gas. \u003cbr\u003eFor example, water or steam obtained by heating water has flow.\u003cbr\u003e By flow, matter can freely change shape and move along the flow.\u003cbr\u003e \/ Flow has various properties, but among them, 'continuity' is especially important when dealing with generative models.\u003cbr\u003e Continuity means that matter does not suddenly appear or disappear without reason, or that matter does not suddenly warp and appear in a different location when it moves.\u003cbr\u003e --- pp.47-48\u003cbr\u003e\u003cbr\u003e Let's say you write with ink on the surface of water.\u003cbr\u003e The letters written with this ink will gradually dissolve over time, and eventually the ink will mix evenly throughout the water.\u003cbr\u003e (…) If we could reproduce this ink diffusion process in the opposite direction, we could revert from a state where the ink was evenly mixed in water to a state where the letters were written in ink again. \u003cbr\u003eThat is, the idea is that by reversing the process of adding noise to an object with order, gradually destroying it and turning it into complete disorder, we can create order from disorder, that is, realize creation.\u003cbr\u003e\u003cbr\u003e --- pp.68-69\u003cbr\u003e\u003cbr\u003e Understanding generalization is crucial to understanding how training data is referenced to generate new data and why unintended results may occur.\u003cbr\u003e For example, generalizations can sometimes lead to a phenomenon called hallucination.\u003cbr\u003e This is a problem of generating unrealistic data that does not exist in the training data.\u003cbr\u003e While this phenomenon often isn't a problem when generating images, audio, and video, it can cause significant problems when you're trying to generate based on facts. \u003cbr\u003eTherefore, it is desirable to be able to control generalization more precisely.\u003cbr\u003e\n\n\u003c\/div\u003e\n\u003cdiv\u003e --- p.94\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cbr\u003e\u003cdiv\u003e\u003ch5\u003e \u003cb\u003ePublisher's Review\u003c\/b\u003e\n\u003c\/h5\u003e\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv\u003e \u003cb\u003eA book explaining generative AI using only text and pictures.\u003c\/b\u003e\u003cbr\u003e\u003cbr\u003e Flow-based generative techniques, especially diffusion models, have emerged in many fields, including image, audio, and video generation.\u003cbr\u003e Daisuke Okanohara, Japan's top AI expert, wrote \"The Mathematics of Diffusion Models\", which clearly explains the diffusion model mathematically, and this time he wrote \"The Structure of Generative AI\", which explains the entirety of generative AI using only text and pictures, without any mathematical formulas.\u003cbr\u003e\u003cbr\u003e\u003cbr\u003e It is written in an easy-to-read format that even non-experts can understand, covering everything from the history of generative AI to flow, diffusion models and flow matching, optimal transportation, and future prospects. \u003cbr\u003eInstead of avoiding formulas, I prefer to use everyday metaphors from the world we live in. For example, I begin my explanation of the diffusion model as follows:\u003cbr\u003e\u003cbr\u003e\u003cbr\u003e “Let’s say you write with ink on the surface of water.\u003cbr\u003e The letters written with this ink will gradually dissolve over time, and eventually the ink will mix evenly throughout the water.\u003cbr\u003e (…) If we could reproduce this ink diffusion process in the opposite direction, we could revert from a state where the ink was evenly mixed in water to a state where the letters were written in ink again.\u003cbr\u003e In other words, the idea is that by reversing the process of adding noise to an object of order, gradually destroying it and turning it into complete disorder, we can create order from disorder—that is, create it.\u003cbr\u003e \u003cbr\u003eDaisuke Okanohara is famous as the co-founder of Preferred Networks, Japan's largest AI unicorn, but he is also a veteran author who has written over a dozen professional books and is praised for \"providing rich context surrounding technology.\"\u003cbr\u003e Let's properly understand the structure of generative AI, which is at the center of today's IT, with his kind and accurate explanation. \u003cbr\u003e\n\n\u003c\/div\u003e\n\u003cdiv\u003e\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n\n\n\u003c\/div\u003e\n\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\n\u003cdiv style=\"width:95%;padding-top:20px;padding-bottom:20px\"\u003e\n\n\u003cdiv style=\"text-align:left;font-size:16px;font-weight:bold;padding-bottom:20px\"\u003e GOODS SPECIFICS \u003c\/div\u003e\n\n\u003cdiv style=\"text-align:left;font-size:14px;line-height:1.6em;\"\u003e\n\n\u003cdiv style=\"width:100%;margin-bottom:5px;line-height:1.6em;font-size:14px\"\u003e - \u003cstrong\u003eDate of issue:\u003c\/strong\u003e May 13, 2025\u003c\/div\u003e\n\n\u003cdiv style=\"width:100%;margin-bottom:5px;line-height:1.6em;font-size:14px\"\u003e - \u003cstrong\u003ePage count, weight, size:\u003c\/strong\u003e 132 pages | 170*225*8mm\u003c\/div\u003e\n\n\u003cdiv style=\"width:100%;margin-bottom:5px;line-height:1.6em;font-size:14px\"\u003e - \u003cstrong\u003eISBN13:\u003c\/strong\u003e 9791194587231 \u003c\/div\u003e\n\n\n\u003c\/div\u003e\n\n\n\u003c\/div\u003e\n\n\n\u003c\/div\u003e\n\n\u003ccenter\u003e\n\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\n\u003ccenter\u003e\u003ctable\u003e\u003ctr\u003e\u003ctd style=\"height:10px\"\u003e\u003c\/td\u003e\u003c\/tr\u003e\u003c\/table\u003e\u003c\/center\u003e\n\n\u003cspan\u003e\u003c\/span\u003e\n\n\u003c\/center\u003e\n\n\n\u003c\/center\u003e","brand":"LIBRAIRIE COREENNE","offers":[{"title":"Default Title","offer_id":43893405089834,"sku":"140202","price":30.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0683\/2750\/5962\/files\/6c82ceef82117ea3e1d0c82e393fce7e.jpg?v=1765400589","url":"https:\/\/librairie.coreenne.fr\/en\/products\/140202","provider":"LIBRAIRIE COREENNE","version":"1.0","type":"link"}