DOCSLIB.ORG

Is the Brain a Good Model for Machine Intelligence?

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Is the Brain a Good Model for Machine Intelligence? COMMENT equivalent to Turing’s finite-state machine with an infinite tape and a finite symbol set, and that does computation. In 1943, Warren McCulloch and Walter Pitts2 noted the “all-or-none” nature of the firing of neurons in a nervous system, and suggested that networks of neurons could be modelled as logical propositions. They POTTS ANDY BY ILLUSTRATION modelled a network of neurons as circuits of logic gates, noting that these may “compute only such numbers as can a Turing machine”. But more, they proposed that everything at a psychological level happens in these net- works. Over the decades, such ideas begat more studies in neural networks, which in turn begat computational neuroscience. Now those metaphors and models pervade explanations of how the brain ‘computes’. But these binary abstractions do not capture all the complexities inherent in the brain. So now I see circles before my eyes. The brain has become a digital computer; yet we are still trying to make our machines intelli- gent. Should those machines be modelled on the brain, given that our models of the brain are performed on such machines? That will probably not be enough. When you are stuck, you are stuck. We will get out of this cul-de-sac, but it will take some brave and bright souls to break out of our circular confusions of models. DEMIS HASSABIS Is the brain a good model Model the brain’s algorithms for machine intelligence? Neuroscientist, computer-game producer and chess master, To celebrate the centenary of the year of Alan Turing’s University College London birth, four scientists and entrepreneurs assess the Alan Turing looked to the human brain as divide between neuroscience and computing. the prototype for intelligence. If he were alive today, he would surely be working at the inter- section of natural and artificial intelligence. model. He abstracted the actions of a human Yet to date, artificial intelligence (AI) RODNEY BROOKS ‘computer’ using paper and pencil to per- researchers have mostly ignored the brain form a calculation (as the word meant then) as a source of algorithmic ideas. Although Avoid the cerebral into a formalized machine, manipulating in Turing’s time we lacked the means to look symbols on an infinite paper tape. inside this biological ‘black box’, we now blind alley But there is a worry that his version of have a host of tools, from functional mag- computation, based on functions of inte- netic resonance imaging to optogenetics, Emeritus professor of robotics, gers, is limited. Biological systems clearly with which to do so. Massachusetts Institute of Technology differ. They must respond to varied stimuli Neuroscience has two key contributions over long periods of time; those responses to make towards progress in AI. First, the I believe that we are in an intellectual cul-de- in turn alter their environment and subse- many structures being discovered in the sac, in which we model brains and computers quent stimuli. The individual behaviours of brain — such as grid cells used for naviga- on each other, and so prevent ourselves from social insects, for example, are affected by tion, or hierarchical cell layers for vision having deep insights that would come with the structure of the home they build and the processing — may inspire new computer new models. behaviour of their siblings within it. The first step in this back and forth was Nevertheless, for 70 years, those people 1 TURING AT 00 made by Alan Turing. In his 1936 paper working in what is now called computa- A legacy that spans science: laying the foundations of computation, tional neuroscience have assumed that the nature.com/turing Turing used a person as the basis for his brain is a computer — a machine that is 462 | NATURE | VOL 482 | 23 FEBRUARY 2012 © 2012 Macmillan Publishers Limited. All rights reserved COMMENT algorithms and architectures. Second, anatomy or physiology. Even biologically Two of the many fundamental differences neuroscience findings may validate the plau- inspired approaches such as cellular autom- between the brain and the computer are sibility of existing algorithms being integral ata, genetic algorithms and neural networks memory and processing speed. The analogue parts of a general AI system. have only a tenuous link to living tissue. of long-term memory in a computer is To advance AI, we need to better under- In 1944, Turing confessed his dream of the hard disk, which can store practically stand the brain’s workings at the algorithmic building a brain, and many people continue unlimited amounts of data. Short-term infor- level — the representations and processes in that endeavour to this day. Yet any neuro­ mation is held in its random access memory that the brain uses to portray the world biologist will view such attempts as naive. (RAM), the capacity of which is astronomical around us. For example, if we knew how How can you represent a neuronal synapse — compared with the human brain. Such quan- conceptual knowledge was formed from per- a complex structure containing hundreds of titative differences become qualitative when ceptual inputs, it would crucially allow for the different proteins, each a chemical prodigy in considering strategies meaning of symbols in an artificial language its own right and arranged in a mare’s nest of “Signals in for intelligence. system to be grounded in sensory ‘reality’. interactions — with a single line of code? We the brain are Intelligence is mani- AI researchers should not only immerse still do not know the detailed circuitry of any transmitted fested by the ability to themselves in the latest brain research, but region of the brain well enough to reproduce at a snail’s learn. Machine-learning also conduct neuroscience experiments to its structure. Brains are special. They steer us pace.” practitioners use ‘stat­ address key questions such as: “How is con- through the world, tell us what to do or say, istical learning’ which ceptual knowledge acquired?” Conversely, and perform myriad vital functions. Brains requires a very large collection of examples from a neuroscience perspective, attempt- are the source of our emotions, motivation, on which to generalize. This ‘frequentist’ ing to distil intelligence into an algorithmic creativity and consciousness. Because no one approach to probabilistic reasoning needs construct may prove to be the best path to knows how to reproduce any of these features vast memory capacity and algorithms that are understanding some of the enduring mys- in an artificial machine, we must consider at odds with available data on how the brain teries of our minds, such as consciousness that something important is missing from works. For example, IBM computer Watson and dreams. the canonical microchip. needed to consume terabytes of reference Brains differ from computers in a number material to beat human contestants on Jeop- of key respects. They operate in cycles rather ardy!. Volvo’s pedestrian-detection system DENNIS BRAY than in linear chains of causality, sending (developed by Mobileye) learned to identify and receiving signals back and forth. Unlike people by using millions of pictures. In both Brain emulation the hardware and software of a machine, the cases, the human brain is considerably more mind and brain are not distinct entities. And parsimonious in the reliance on data — some- requires cells then there is the question of chemistry. thing that does not constrain the computer. Living cells process incoming sensory In terms of processing power, the brain Department of Physiology, information and generate not just electri- can reach about 10–50 petaflops — equiva- Development and Neuroscience, cal signals but subtle biochemical changes. lent to hundreds of thousands of the most University of Cambridge Cells are soft, malleable and built from an advanced Intel Core i7 CPUs. Yet signals essentially infinite variety of macromolecular in the brain are transmitted at a snail’s pace Machines can match us in many tasks, but species quite unlike silicon chips. Organisms — five or six orders of magnitude slower they work differently from networks of nerve encode past experiences in distinct cellular than modern CPUs. This huge difference in cells. If our aim is to build machines that are states — in humans these are the substrate communication speed drives vastly different ever more intelligent and dexterous, then we of goal-oriented movements and the sense architectures. should use circuits of copper and silicon. But of self. Perhaps machines built from cell-like The brain compensates for the slow if our aim is to reproduce the human brain, components would be more like us. signal speed by adopting a hierarchical paral- with its quirky brilliance, capacity for multi- lel structure, involving successive layers with tasking and sense of self, we have to look for increasing receptive field and complexity. By other materials and different designs. AMNON SHASHUA comparison, a computer architecture is usu- Computers outperform us in complex ally flat and, because of its much faster clock mathematical calculations and are better at Speed will trump rate, can employ brute-force techniques. storing and retrieving data. We accept that Computer chess systems such as Deep Blue they can beat us at chess — once regarded brain’s advantages use pattern-recognition strategies, such as as the apogee of human intellect. But the libraries of opening moves and completely success of a computer called Watson in US Sachs Professor of Computer Science, solved end-games, complemented by their television quiz show Jeopardy! in 2011 was Hebrew University of Jerusalem, and ability to evaluate the outcomes of some a nail in the coffin of human superiority. co-founder and chairman of Mobileye 200 million moves per second. This is way The machine beat two human contestants beyond the best grandmaster.
Load more

Recommended publications
  • Lecture Note on Deep Learning and Quantum Many-Body Computation
    Lecture Note on Deep Learning and Quantum Many-Body Computation Jin-Guo Liu, Shuo-Hui Li, and Lei Wang∗ Institute of Physics, Chinese Academy of Sciences Beijing 100190, China November 23, 2018 Abstract This note introduces deep learning from a computa- tional quantum physicist’s perspective. The focus is on deep learning’s impacts to quantum many-body compu- tation, and vice versa. The latest version of the note is at http://wangleiphy.github.io/. Please send comments, suggestions and corrections to the email address in below. ∗ [email protected] CONTENTS 1 introduction2 2 discriminative learning4 2.1 Data Representation 4 2.2 Model: Artificial Neural Networks 6 2.3 Cost Function 9 2.4 Optimization 11 2.4.1 Back Propagation 11 2.4.2 Gradient Descend 13 2.5 Understanding, Visualization and Applications Beyond Classification 15 3 generative modeling 17 3.1 Unsupervised Probabilistic Modeling 17 3.2 Generative Model Zoo 18 3.2.1 Boltzmann Machines 19 3.2.2 Autoregressive Models 22 3.2.3 Normalizing Flow 23 3.2.4 Variational Autoencoders 25 3.2.5 Tensor Networks 28 3.2.6 Generative Adversarial Networks 29 3.3 Summary 32 4 applications to quantum many-body physics and more 33 4.1 Material and Chemistry Discoveries 33 4.2 Density Functional Theory 34 4.3 “Phase” Recognition 34 4.4 Variational Ansatz 34 4.5 Renormalization Group 35 4.6 Monte Carlo Update Proposals 36 4.7 Tensor Networks 37 4.8 Quantum Machine Leanring 38 4.9 Miscellaneous 38 5 hands on session 39 5.1 Computation Graph and Back Propagation 39 5.2 Deep Learning Libraries 41 5.3 Generative Modeling using Normalizing Flows 42 5.4 Restricted Boltzmann Machine for Image Restoration 43 5.5 Neural Network as a Quantum Wave Function Ansatz 43 6 challenges ahead 45 7 resources 46 BIBLIOGRAPHY 47 1 1 INTRODUCTION Deep Learning (DL) ⊂ Machine Learning (ML) ⊂ Artificial Intelli- gence (AI).
    [Show full text]
  • AI Computer Wraps up 4-1 Victory Against Human Champion Nature Reports from Alphago's Victory in Seoul
    The Go Files: AI computer wraps up 4-1 victory against human champion Nature reports from AlphaGo's victory in Seoul. Tanguy Chouard 15 March 2016 SEOUL, SOUTH KOREA Google DeepMind Lee Sedol, who has lost 4-1 to AlphaGo. Tanguy Chouard, an editor with Nature, saw Google-DeepMind’s AI system AlphaGo defeat a human professional for the first time last year at the ancient board game Go. This week, he is watching top professional Lee Sedol take on AlphaGo, in Seoul, for a $1 million prize. It’s all over at the Four Seasons Hotel in Seoul, where this morning AlphaGo wrapped up a 4-1 victory over Lee Sedol — incidentally, earning itself and its creators an honorary '9-dan professional' degree from the Korean Baduk Association. After winning the first three games, Google-DeepMind's computer looked impregnable. But the last two games may have revealed some weaknesses in its makeup. Game four totally changed the Go world’s view on AlphaGo’s dominance because it made it clear that the computer can 'bug' — or at least play very poor moves when on the losing side. It was obvious that Lee felt under much less pressure than in game three. And he adopted a different style, one based on taking large amounts of territory early on rather than immediately going for ‘street fighting’ such as making threats to capture stones. This style – called ‘amashi’ – seems to have paid off, because on move 78, Lee produced a play that somehow slipped under AlphaGo’s radar. David Silver, a scientist at DeepMind who's been leading the development of AlphaGo, said the program estimated its probability as 1 in 10,000.
    [Show full text]
  • Accelerators and Obstacles to Emerging ML-Enabled Research Fields
    Life at the edge: accelerators and obstacles to emerging ML-enabled research fields Soukayna Mouatadid Steve Easterbrook Department of Computer Science Department of Computer Science University of Toronto University of Toronto [email protected] [email protected] Abstract Machine learning is transforming several scientific disciplines, and has resulted in the emergence of new interdisciplinary data-driven research fields. Surveys of such emerging fields often highlight how machine learning can accelerate scientific discovery. However, a less discussed question is how connections between the parent fields develop and how the emerging interdisciplinary research evolves. This study attempts to answer this question by exploring the interactions between ma- chine learning research and traditional scientific disciplines. We examine different examples of such emerging fields, to identify the obstacles and accelerators that influence interdisciplinary collaborations between the parent fields. 1 Introduction In recent decades, several scientific research fields have experienced a deluge of data, which is impacting the way science is conducted. Fields such as neuroscience, psychology or social sciences are being reshaped by the advances in machine learning (ML) and the data processing technologies it provides. In some cases, new research fields have emerged at the intersection of machine learning and more traditional research disciplines. This is the case for example, for bioinformatics, born from collaborations between biologists and computer scientists in the mid-80s [1], which is now considered a well-established field with an active community, specialized conferences, university departments and research groups. How do these transformations take place? Do they follow similar paths? If yes, how can the advances in such interdisciplinary fields be accelerated? Researchers in the philosophy of science have long been interested in these questions.
    [Show full text]
  • The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design
    The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design Jeffrey Dean Google Research [email protected] Abstract The past decade has seen a remarkable series of advances in machine learning, and in particular deep learning approaches based on artificial neural networks, to improve our abilities to build more accurate systems across a broad range of areas, including computer vision, speech recognition, language translation, and natural language understanding tasks. This paper is a companion paper to a keynote talk at the 2020 International Solid-State Circuits Conference (ISSCC) discussing some of the advances in machine learning, and their implications on the kinds of computational devices we need to build, especially in the post-Moore’s Law-era. It also discusses some of the ways that machine learning may also be able to help with some aspects of the circuit design process. Finally, it provides a sketch of at least one interesting direction towards much larger-scale multi-task models that are sparsely activated and employ much more dynamic, example- and task-based routing than the machine learning models of today. Introduction The past decade has seen a remarkable series of advances in machine learning (ML), and in particular deep learning approaches based on artificial neural networks, to improve our abilities to build more accurate systems across a broad range of areas [LeCun et al. 2015]. Major areas of significant advances ​ ​ include computer vision [Krizhevsky et al. 2012, Szegedy et al. 2015, He et al. 2016, Real et al. 2017, Tan ​ ​ ​ ​ ​ ​ and Le 2019], speech recognition [Hinton et al.
    [Show full text]
  • The Power and Promise of Computers That Learn by Example
    Machine learning: the power and promise of computers that learn by example MACHINE LEARNING: THE POWER AND PROMISE OF COMPUTERS THAT LEARN BY EXAMPLE 1 Machine learning: the power and promise of computers that learn by example Issued: April 2017 DES4702 ISBN: 978-1-78252-259-1 The text of this work is licensed under the terms of the Creative Commons Attribution License which permits unrestricted use, provided the original author and source are credited. The license is available at: creativecommons.org/licenses/by/4.0 Images are not covered by this license. This report can be viewed online at royalsociety.org/machine-learning Cover image © shulz. 2 MACHINE LEARNING: THE POWER AND PROMISE OF COMPUTERS THAT LEARN BY EXAMPLE Contents Executive summary 5 Recommendations 8 Chapter one – Machine learning 15 1.1 Systems that learn from data 16 1.2 The Royal Society’s machine learning project 18 1.3 What is machine learning? 19 1.4 Machine learning in daily life 21 1.5 Machine learning, statistics, data science, robotics, and AI 24 1.6 Origins and evolution of machine learning 25 1.7 Canonical problems in machine learning 29 Chapter two – Emerging applications of machine learning 33 2.1 Potential near-term applications in the public and private sectors 34 2.2 Machine learning in research 41 2.3 Increasing the UK’s absorptive capacity for machine learning 45 Chapter three – Extracting value from data 47 3.1 Machine learning helps extract value from ‘big data’ 48 3.2 Creating a data environment to support machine learning 49 3.3 Extending the lifecycle
    [Show full text]
  • ARCHITECTS of INTELLIGENCE for Xiaoxiao, Elaine, Colin, and Tristan ARCHITECTS of INTELLIGENCE
    MARTIN FORD ARCHITECTS OF INTELLIGENCE For Xiaoxiao, Elaine, Colin, and Tristan ARCHITECTS OF INTELLIGENCE THE TRUTH ABOUT AI FROM THE PEOPLE BUILDING IT MARTIN FORD ARCHITECTS OF INTELLIGENCE Copyright © 2018 Packt Publishing All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, without the prior written permission of the publisher, except in the case of brief quotations embedded in critical articles or reviews. Every effort has been made in the preparation of this book to ensure the accuracy of the information presented. However, the information contained in this book is sold without warranty, either express or implied. Neither the author, nor Packt Publishing or its dealers and distributors, will be held liable for any damages caused or alleged to have been caused directly or indirectly by this book. Packt Publishing has endeavored to provide trademark information about all of the companies and products mentioned in this book by the appropriate use of capitals. However, Packt Publishing cannot guarantee the accuracy of this information. Acquisition Editors: Ben Renow-Clarke Project Editor: Radhika Atitkar Content Development Editor: Alex Sorrentino Proofreader: Safis Editing Presentation Designer: Sandip Tadge Cover Designer: Clare Bowyer Production Editor: Amit Ramadas Marketing Manager: Rajveer Samra Editorial Director: Dominic Shakeshaft First published: November 2018 Production reference: 2201118 Published by Packt Publishing Ltd. Livery Place 35 Livery Street Birmingham B3 2PB, UK ISBN 978-1-78913-151-2 www.packt.com Contents Introduction ........................................................................ 1 A Brief Introduction to the Vocabulary of Artificial Intelligence .......10 How AI Systems Learn ........................................................11 Yoshua Bengio .....................................................................17 Stuart J.
    [Show full text]
  • Technology Innovation and the Future of Air Force Intelligence Analysis
    C O R P O R A T I O N LANCE MENTHE, DAHLIA ANNE GOLDFELD, ABBIE TINGSTAD, SHERRILL LINGEL, EDWARD GEIST, DONALD BRUNK, AMANDA WICKER, SARAH SOLIMAN, BALYS GINTAUTAS, ANNE STICKELLS, AMADO CORDOVA Technology Innovation and the Future of Air Force Intelligence Analysis Volume 2, Technical Analysis and Supporting Material RR-A341-2_cover.indd All Pages 2/8/21 12:20 PM For more information on this publication, visit www.rand.org/t/RRA341-2 Library of Congress Cataloging-in-Publication Data is available for this publication. ISBN: 978-1-9774-0633-0 Published by the RAND Corporation, Santa Monica, Calif. © Copyright 2021 RAND Corporation R® is a registered trademark. Cover: U.S. Marine Corps photo by Cpl. William Chockey; faraktinov, Adobe Stock. Limited Print and Electronic Distribution Rights This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited. Permission is given to duplicate this document for personal use only, as long as it is unaltered and complete. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial use. For information on reprint and linking permissions, please visit www.rand.org/pubs/permissions. The RAND Corporation is a research organization that develops solutions to public policy challenges to help make communities throughout the world safer and more secure, healthier and more prosperous. RAND is nonprofit, nonpartisan, and committed to the public interest. RAND’s publications do not necessarily reflect the opinions of its research clients and sponsors.
    [Show full text]
  • Machine Learning Conference Report
    Part of the conference series Breakthrough science and technologies Transforming our future Machine learning Conference report Machine learning report – Conference report 1 Introduction On 22 May 2015, the Royal Society hosted a unique, high level conference on the subject of machine learning. The conference brought together scientists, technologists and experts from across academia, industry and government to learn about the state-of-the-art in machine learning from world and UK experts. The presentations covered four main areas of machine learning: cutting-edge developments in software, the underpinning hardware, current applications and future social and economic impacts. This conference is the first in a new series organised This report is not a verbatim record, but summarises by the Royal Society, entitled Breakthrough Science the discussions that took place during the day and the and Technologies: Transforming our Future, which will key points raised. Comments and recommendations address the major scientific and technical challenges reflect the views and opinions of the speakers and of the next decade. Each conference will focus on one not necessarily that of the Royal Society. Full versions technology and cover key issues including the current of the presentations can be found on our website at: state of the UK industry sector, future direction of royalsociety.org/events/2015/05/breakthrough-science- research and the wider social and economic implications. technologies-machine-learning The conference series is being organised through the Royal Society’s Science and Industry programme, which demonstrates our commitment to reintegrate science and industry at the Society and to promote science and its value, build relationships and foster translation.
    [Show full text]
  • Arxiv:1709.02779V1 [Quant-Ph] 8 Sep 2017 I
    Machine learning & artificial intelligence in the quantum domain Vedran Dunjko Institute for Theoretical Physics, University of Innsbruck, Innsbruck 6020, Austria Max Planck Institute of Quantum Optics, Garching 85748, Germany Email: [email protected] Hans J. Briegel Institute for Theoretical Physics, University of Innsbruck Innsbruck 6020, Austria Department of Philosophy, University of Konstanz, Konstanz 78457, Germany Email: [email protected] Abstract. Quantum information technologies, on the one side, and intelligent learning systems, on the other, are both emergent technologies that will likely have a transforming impact on our society in the future. The respective underlying fields of basic research { quantum information (QI) versus machine learning and artificial intelligence (AI) { have their own specific questions and challenges, which have hitherto been investigated largely independently. However, in a growing body of recent work, researchers have been prob- ing the question to what extent these fields can indeed learn and benefit from each other. QML explores the interaction between quantum computing and machine learning, inves- tigating how results and techniques from one field can be used to solve the problems of the other. In recent time, we have witnessed significant breakthroughs in both directions of influence. For instance, quantum computing is finding a vital application in providing speed-ups for machine learning problems, critical in our \big data" world. Conversely, machine learning already permeates many cutting-edge technologies, and may become instrumental in advanced quantum technologies. Aside from quantum speed-up in data analysis, or classical machine learning optimization used in quantum experiments, quan- tum enhancements have also been (theoretically) demonstrated for interactive learning tasks, highlighting the potential of quantum-enhanced learning agents.
    [Show full text]
  • Updated Compressed Data Trends
    Trends in Data Science 2019/2020 Foreword Across six major conferences in Boston, NYC, Sao Paulo, Bangalore, London, and San Francisco, ODSC hosted an unprecedented 920+ speakers in 2019. It was another year of rapid advances and exciting developments in the fields of data science and artificial intelli- gence. The year was significant on several fronts. Frameworks like PyTorch, Keras, and TensorFlow saw major releases that further cemented their position as the leading machine and deep learning tools and featured in many of our hands-on training sessions. However, NLP transformer architectures was one of our speak- ers’ favorite topics in 2019 due to major advances such as OpenAI’s GPT-2, Google’s BERT, DiDI’s Elmo, and FaceBook’s RoBERTa. Hugging Face’s transformer library which exposes an API to these transformers got a lot of traction also. These pretrained models, especially BERT, have been especially hailed by many as NLPs Imagenet moment. Employing new techniques like bi-directional sequencing and transformers these models are saving data scientists the time and expenses normally required to train NLP models thus marking a major development. Our new MLOps and Data Engineering focus area tracks coincided with the massive ramp up of efforts to increase the percent of data science projects into production. Tools like MLFlow and Kubeflow continued to grow in popularity. Interest in AutoML grew exponen- tially in 2019, given its potential as a productivity tool in all stages of the machine learning life cycle. Sessions around Annotation, model interoperability, pipelines, deployment, and testing saw increased interest. This coupled with the fact that 2019 saw a significant drop in the cost of modeling helped accelerate deployment in production environments.
    [Show full text]
  • Brave New World
    Visions don’t come much loftier than DeepMind’s. Described as an Apollo programme for the 21st century, it’s helping to tackle the world’s biggest problems by developing learning machines more powerful than the human brain. “We’re on a scientific mission to push the boundaries of artificial intelligence (AI), developing programs that can learn to solve any complex problem without needing to be taught how.” That’s literally the mission statement of DeepMind, the world’s leading AI research firm. It’s the sheer breadth of that “any” that’s so thrilling. While most AI is “narrow”, able only to learn a single task, DeepMind’s goal is more akin to the brain: a “general-purpose learning machine” that can make sense of whatever it’s applied to – even utilising such human traits as the ability to “imagine” outcomes and reason about the future. As Dr Demis Hassabis, DeepMind’s co-founder and CEO, puts it: “First we solve intelligence, then we use that to solve everything else and make the world a better place.” “Time magazine listed Hassabis among the 100 Most Influential People for 2017” DeepMind honed its expertise in the world of games: “a useful training ground”, it now says. It created a program that taught itself how to play and win at 49 Atari titles, with just raw pixels as input. It developed AlphaGo, which beat the world’s best player at Go – one of the most complex and intuitive games ever devised, with more positions than there are atoms in the universe.
    [Show full text]
  • Global Artificial Intelligence Industry Whitepaper
    Global artificial intelligence industry whitepaper Global artificial intelligence industry whitepaper | 4. AI reshapes every industry 1. New trends of AI innovation and integration 5 1.1 AI is growing fully commercialized 5 1.2 AI has entered an era of machine learning 6 1.3 Market investment returns to reason 9 1.4 Cities become the main battleground for AI innovation, integration and application 14 1.5 AI supporting technologies are advancing 24 1.6 Growing support from top-level policies 26 1.7 Over USD 6 trillion global AI market 33 1.8 Large number of AI companies located in the Beijing-Tianjin-Hebei Region, Yangtze River Delta and Pearl River Delta 35 2. Development of AI technologies 45 2.1 Increasingly sophisticated AI technologies 45 2.2 Steady progress of open AI platform establishment 47 2.3 Human vs. machine 51 3. China’s position in global AI sector 60 3.1 China has larger volumes of data and more diversified environment for using data 61 3.2 China is in the highest demand on chip in the world yet relying heavily on imported high-end chips 62 3.3 Chinese robot companies are growing fast with greater efforts in developing key parts and technologies domestically 63 3.4 The U.S. has solid strengths in AI’s underlying technology while China is better in speech recognition technology 63 3.5 China is catching up in application 64 02 Global artificial intelligence industry whitepaper | 4. AI reshapes every industry 4. AI reshapes every industry 68 4.1 Financial industry: AI enhances the business efficiency of financial businesses
    [Show full text]