人工智能行业研究:·20160528《百度和清华都在投入!31页报告解读全球深度学习产业链》报告原文
Copyright © 2016 Tractica 1 How Deep Learning Is Enabling Computer Vision Markets Bruce Daley May 2, 2016 Copyright © 2016 Tractica 2 What the Future Promised 1956 1958 1960 1962 1964 1966 1968 1970 First AI Winter 1970 Term “Artificial Intelligence” Coined at Dartmouth Conferences Summer 1956 Mark 1 Perceptron Artificial Neural Net 1958 Perceptrons published 1969 2001 Released April 3 1968 Calculus solver 1961 Checkers program 1962 Knowledge based Chess 1967 Natural Language Algebra Solver 1964 ATOMIC CARS ARTIFICIAL INTELLIGENCE VACATIONS ON THE MOON Copyright © 2016 Tractica 3 What We Got Instead REALITY TV STAYCATIONS YUGO Google Unsupervised Deep Learning Test June 2012 2008 2009 2010 2011 2012 2013 2014 2015 Checkers solved by researchers at University of Alberta 2007 Google builds self driving car 2010 Artificial Intelligence for Enterprise Applications published Apr 2015 Her Released Dec 18 2013 Narrative Science creates Quill to write news stories 2010 IBM Watson wins Jeopardy! 2011 Apple launches Siri 2011 Copyright © 2016 Tractica 4 Why We Became Interested $- $10,000 $20,000 $30,000 $40,000 $50,000 $60,000 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 Expected Growth in Deep Learning Hardware Services Software WHAT IS DIFFERENT NOW? $10.4 billion for software alone in 2024 Hardware and services revenue will be many times software sales • Increases in data • Improvements in hardware • Advancements in algorithms Copyright © 2016 Tractica 5 Introductions BRUCE DALEY Principal Analyst Artificial Intelligence Conclusion Deep Learning Market Use Cases and Forecasts Market Drivers and Market Barriers About Tractica What is Different About Deep Learning? Copyright © 2016 Tractica 6 About Tractica Tractica is a market intelligence firm that focuses on human interaction with technology The firm’s market research and consulting services provide industry participants and stakeholders with in-depth analysis of emerging technology trends, business issues, market drivers, and end-user demand dynamics across multiple application domains Sector Focus • User Interface Technologies • Biometrics • Digital Health • Wearable Devices • Automation identify soil health, wet spots in fields, or irrigation and waste runoff; and monitor crop canopies Copyright © 2016 Tractica 20 Market Forecast: Agriculture Industry $0 $ 5 0 0 $ 1 ,0 0 0 $ 1 ,5 0 0 $2 ,0 00 $ 2 ,5 0 0 $3 ,0 00 $ 3 ,5 0 0 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 2 0 1 9 2 0 2 0 2 0 2 1 2 0 2 2 2 0 2 3 2 0 2 4 N ort h Ame r i ca We s te rn Eu rop e Ea s te rn Eu rope Asi a Pa c ifi c La ti n Ame ric a M iddl e East Afr ic a ($ M illi o n s) Deep Learning Software Revenue in the Agriculture Industry by Region, World Markets: 2015-2024 (Source: Tractica) $0 $ 5 0 $ 1 0 0 $ 1 5 0 $2 00 $ 2 5 0 20 15 20 16 20 17 20 18 20 19 20 20 2 0 2 1 20 22 20 23 20 24 North Am e r i c a We s te rn Eu rop e Eas tern Eur ope Asi a Pa c ifi c La ti n Ame ric a M id dl e Ea s t Afr ic a( $ M illi o n s) Companies to Watch First world agriculture has some of the best use cases of any industry Key Takeaway Copyright © 2016 Tractica 21 Clinical Trial Medication Compliance Use Case The need — patients not taking the right medications, at the right time, and in the right doses is a major problem in medicine. The problem is particularly acute in clinical trials Use Case 2.4 from Deep Learning Use Cases for Computer Vision Copyright © 2016 Tractica 22 Market Forecast: Healthcare Deep Learning Software Revenue in the Healthcare Industry by Region, World Markets: 2015-2024 (Source: Tractica) $0 $10 $20 $30 $40 $50 $60 $70 $80 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 North America Western Europe Eastern Europe Asia Pacific Latin America Middle East Africa Companies to Watch The more administrative aspects of healthcare will benefit from deep learning long before patients will see any meaningful benefit Key Takeaway Copyright © 2016 Tractica 23 Clothes Sizing and Fitting Use Case Use Case 2.5 from Deep Learning Use Cases for Computer Vision The need — during the dot-com boom, online clothing sales were predicted to grow to 50% of total sales. Other kinds of merchandise have reached that mark but online clothing sales hovers around 20% Copyright © 2016 Tractica 24 Market Forecast: Retail Industry $0 $ 5 0 0 $ 1 ,0 0 0 $ 1 ,5 0 0 $2 ,0 00 $ 2 ,5 0 0 $3 ,0 00 $ 3 ,5 0 0 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 2 0 1 9 2 0 2 0 2 0 2 1 2 0 2 2 2 0 2 3 2 0 2 4 N ort h Ame r i ca We s te rn Eu rop e Ea s te rn Eu rope Asi a Pa c ifi c La ti n Ame ric a M iddl e East Afr ic a ($ M illi o n s) Deep Learning Software Revenue in the Retail Industry by Region, World Markets: 2015-2024 (Source: Tractica) $0 $ 1 0 0 $ 2 0 0 $ 3 0 0 $ 4 0 0 $ 5 0 0 $ 6 0 0 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 2 0 1 9 2 0 2 0 2 0 2 1 2 0 2 2 2 0 2 3 2 0 2 4 North Am e r i c a We s te rn Eu rop e Ea s te rn Eu rope Asi a Pa c ifi c La ti n Americ a M id dl e Ea s t A fr ic a($ M il li o n s) Companies to Watch Retail is likely to be an industry sector where deep learning has a profound influence Key Takeaway Copyright © 2016 Tractica 25 Manufacturing Use Case — Quality Control Use Case Use Case 2.6 from Deep Learning Use Cases for Computer Vision The need — in 2012, the number of automobile recalls almost equaled the number of sales, and a high volume of warranty payments impacted automakers’ profits Copyright © 2016 Tractica 26 $0 $ 2 0 0 $ 4 0 0 $ 6 0 0 $ 8 0 0 $1 ,0 00 $ 1 ,2 0 0 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 2 0 1 9 2 0 2 0 2 0 2 1 2 0 2 2 2 0 2 3 2 0 2 4 North Am e r i c a We s te rn Eu rop e Ea s te rn Eu rope Asi a Pa c ifi c La ti n Ame ric a M iddl e East Afr ic a ($ M illi o n s) Market Forecast: Manufacturing Deep Learning Software Revenue in the Manufacturing Industry by Region, World Markets: 2015-2024 (Source: Tractica) Companies to Watch Surprisingly, this sector lags behind others in research and adoption. The U.S. and China lead, but a strong Japanese resurgence should not be ruled out Key Takeaway Copyright © 2016 Tractica 27 • Deep learning is limited • Enabling technology • Mimics the brain but not exactly • Dependent on good data • Deep learning is real • More data • Faster hardware • Better software • Deep learning is different • Data agnostic • New programming methodologies • Companies with the best business models will win Its the business models, not the technology, that will prove most disruptive Growth in Value of Data Introduction Growth Maturity Decline Sa les Product Life Cycle Fore ca sts Summary Copyright © 2016 Tractica 28 “In all industries, especially the technology industry, people overestimate what you can do in one year, and they underestimate what you can do in ten.” Deep Learning Will Enable New Business Models, Disrupt Old Ones, and Create New Sources of Wealth Largest taxi Company (Source: Tom Goodwin) Most valuable retailer Largest hotel company Most popular media company Marc Benioff Chairman and CEO Copyright © 2016 Tractica 29 TITLE Sub title Quarter B R U C E D A L E Y P r i nc i pa l A na l y s t C L I N T W H E E L O C K M a na g i ng D i r e c t or D e e p L e a r n i n g f o r E n t e r p r i s e A p p l i c a t i o n s A dv e r t i s i ng T e c hnol og y , F i na nc i a l S e r v i c e s , M e di a , M a nuf a c t ur i ng , O i l & G a s , R e t a i l , a nd O t he r E nt e r pr i s e M a r k e t s f or D e e p L e a r ni ng S of t w a r e a nd S y s t e m s P u b l i s h e d 4 Q 2 0 1 5 R E S E A R C H R E P O R T • Industries • Ad Service Technology • Agriculture • Automotive • Consumer Finance • Data Storage and Networking • Education • Healthcare • Investment • Legal • Manufacturing • Media • Medical Diagnostics • Oil and Gas • Philanthropies • Retail Additional Reading • Hardware Product Categories • Cloud Services • Compute Products • GPU Chips • Network Products • Storage Devices Copyright © 2016 Tractica 30 Additional Reading Where Data is Wealth Profiting from data storage in a digital society Published 4Q 2015 TITLE Sub title Quarter BRU CE D A LEY Pr inc ipa l A na lyst CLIN T W H EELO CK M ana ging D ir ect or D eep Learning for Enterprise A pplications A dv ert ising Technol og y, Fina nc ial Services, M edi a, M anuf act ur ing , O il & Gas, Ret ail, and O t he r Ent erpr ise M ar ket s for D eep Learni ng Sof t w are and Syst em s Published 4 Q 2 0 1 5 RESEA RCH REPO RT Deep Learning Use Cases for Computer Vision Six Deep Learning-Enabled Vision Applications in Digital Media, Healthcare, Agriculture, Retail, Manufacturing, and Other Industries Published 2Q 2016 Copyright © 2016 Tractica 31 1111 Pearl Street, Suite 201 Boulder, CO 80302 USA +1.303.248.3000 www.tractica.com Contact Us