Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Chile: Semantics, Deep Learning,
and the
Transformation of Business
Steve Omohundro, Ph.D.
PossibilityResearch.com
SteveOm...
Economic Impact
Deep Learning
Neats vs. Scruffies
Semantics
The Future
https://www.flickr.com/photos/danielfoster/14758510...
Multi-Billion Dollar Investments
• 2013 Facebook – AI lab, DeepFace
• 2013 Yahoo - LookFlow
• 2013 Ebay – AI lab
• 2013 Al...
McKinsey: $50 Trillion to 2025
http://www.mckinsey.com/insights/business_technology/disruptive_technologies
http://thismasquerade.me/wp-content/uploads/blogger/-84ftTg_azwY/Um8up3j4zTI/AAAAAAAABJ8/9_-dxg23UaY/s1600/9979402_ml.jpg
...
Internet of Things: $15 Trillion to 2025
100 Billion devices by 2025
Cars, Appliances, Cameras, Meters, Wearables, etc.
ht...
Robot Manufacturing: $10 Trillion to 2025
Work 24 hours/day
No breaks, food, medical
Don’t quit, get bored, get depressed
...
Foxconn Technology Group
• World’s largest contract
manufacturer
• Assembles 40% of all consumer
electronics
• iPhone, iPa...
420 Chinese Robot Companies
http://thestack.com/china-robot-market-overstimulated-291014 http://www.reuters.com/article/20...
March 2015: China Brain
http://www.scmp.com/lifestyle/technology/article/1728422/head-chinas-google-wants-country-take-lea...
https://osuwmcdigital.osu.edu/sitetool/sites/urologypublic/images/Robotics/robotic_surgery_table.jpg
Health Care: $10 Tril...
Self-Driving Vehicles: $10 Trillion by 2025
Disrupt Dealers, Insurance, Parking, Finance, Trucking, Taxis
10 million jobs
...
Tesla: “Autopilot” mode
http://zackkanter.com/2015/01/23/how-ubers-autonomous-cars-will-destroy-10-million-jobs-by-2025/ht...
https://d185ox70mr1pkc.cloudfront.net/post_image_teaser/1403883171000-uber-force.png
World’s largest job creator: 50,000 p...
http://airwolf3d.com/wp-content/uploads/2012/05/3d-printer-v.5.5-airwolf3d1.jpg
3D Printing: $2 Trillion by 2025
April 2014: Chinese WinSun 3D printed
10 houses, 2100 sq ft, $4800
http://m.newsru.co.il/realty/20jan2015/3d_house_i101.ht...
WinSun 3D printed 12,000 sq ft villa
http://3dprint.com/38144/3d-printed-apartment-building/
US Building construction: $1 ...
https://venturescannerinsights.files.wordpress.com/2015/01/artificial-intelligence-map.jpg
https://venturescannerinsights.files.wordpress.com/2015/09/ai4.jpeg
Deep Learning Neural Nets
http://www.nicta.com.au/content/uploads/2015/02/deep.jpg
Deep Learning Successes
• Speech Recognition TIMIT 2009: Cortana,
Skype, Google Now, Siri, Baidu, Nuance, etc.
• Image Rec...
http://www-formal.stanford.edu/jmc/history/dartmouth/dartmouth.html
• Natural Language
• Neural Nets
• Self-Improvement
• ...
“Neats” vs. “Scruffies”
http://news.stanford.edu/news/2003/june18/mccarthy-618.html http://www.bbc.co.uk/timelines/zq376fr...
1957: Rosenblatt’s “Perceptron”
http://www.rutherfordjournal.org/article040101.html
“The embryo of an
electronic computer ...
1969: Perceptrons can’t do XOR!
http://www.i-programmer.info/images/stories/BabBag/AI/book.jpg
Minsky & Papert
https://con...
Multilayer Neural Nets
http://opticalengineering.spiedigitallibrary.org/article.aspx?articleid=1714547
Backpropagation
198...
“Neat” Software Compiler
Source
Code
Lexical
Analysis
Parsing
Semantic
Analysis
Code
Generation
Code
Optimization
“Neat” Language Translator?
Tokenization Stemming Tagging
Parsing
Logical
Representation
Translation
http://www.amazon.com/Comprehensive-Grammar-English-Language-General/dp/0582517346/ref=sr_1_2
1957 Chomsky Grammar
"Englis...
http://www.espressoenglish.net/order-of-adjectives-in-english/
Linguistic Rules are Complicated!
http://www.computer.org/csdl/mags/ex/2009/02/mex2009020008-abs.html
http://www.mitpressjournals.org/doi/abs/10.1162/coli.2...
1962: Roger Shepard Cognitive Geometry
http://link.springer.com/article/10.1007/BF02289630
https://psychlopedia.wikispaces...
Word2Vec – Mikolov 2013
• Distributional Semantics – Firth 1957
• Represent words by vectors
• Close vectors represent sim...
Why? Same context shift for all male -> female
https://drive.google.com/file/d/0B7XkCwpI5KDYRWRnd1RzWXQ2TWc/edit?usp=shari...
More Semantic Relations
• Paris – France + Italy = Rome
• Human – Animal = Ethics
• Obama – USA + Russia = Putin
• Library...
Marr’s “Neat” Vision Pipeline
http://homepages.inf.ed.ac.uk/rbf/CVonline/LOCAL_COPIES/GOMES1/marr.html
http://www.amazon.c...
Deep Neural Net Face Recognition
Google FaceNet, June 2015
Record accuracy 99.63% on Labeled Faces in the Wild dataset
Cut...
Cheap Cameras
+
Face Recognition
+
Body Recognition
=
Brin’s “Transparent Society”
http://www.ebay.com/sch/i.html?_nkw=cmo...
https://github.com/Newmu/dcgan_code/tree/gh-pages
Biological Networks are Recurrent
Gene network
Chromosome 22
Human Metabolome
Brain Connectome
http://journals.plos.org/pl...
http://karpathy.github.io/2015/05/21/rnn-effectiveness/
Image Classification
Image Captioning
Sentence Sentiment
English->...
Recurrent Net Hallucinates C Code
http://karpathy.github.io/2015/05/21/rnn-effectiveness/
Karpathy: 464MB of C code, 3 lay...
The rat escaped.
The rat the cat attacked escaped.
The rat the cat the dog chased attacked escaped.
http://arlingtonva.s3....
NeuralTalk and Walk Demo
https://vimeo.com/146492001
https://www.youtube.com/watch?v=08Cl7ii6viY&feature=youtu.be&t=15m31s
DeepMind Deep-Q Networkshttp://www.nature.com/nature...
Aerial Drones: $98 Billion by 2025
Delivery, Surveillance, Agriculture, Military, Police
http://mint-tek.com/wp-content/up...
Deep Learning Has Blindspots
http://arxiv.org/abs/1412.1897
Other Issues
• Typically have problems to solve rather than
reinforcement signals
• Want confidence that system solves pro...
Technology Needs Semantics!
• Analyzing camera, sensor, weather data
• Better search, question answering, info
• Analysis ...
Approaches to Semantics
• Montague – map into Typed Lambda Calculus
• Denotational – map into CS Domains
• Mathematical – ...
New Possibilities Coming Soon!
PossibilityResearch.com
http://flinttown.com/wp-content/uploads/2015/07/Fireworks.jpg
Upcoming SlideShare
Loading in …5
×

Exosphere Chile Talk: Semantics, Deep Learning, and the Transformation of Business

787 views

Published on

McKinsey predicts that AI and robotics will create $50 trillion of value over the next 10 years. Many predict that the recent technology of “deep learning” will be a big part of the transformation. Over 250 deep learning startup companies have attracted more than $1 billion of venture investment in the past year. Deep learning systems have recently broken records in speech recognition, image recognition, image captioning, translation, drug discovery and other tasks. Why is this happening now and how is it likely to play out? We review the development of AI and the pendulum swings between the “neats” and the “scruffies”. We describe traditional approaches to semantics through logics and grammars and the new deep learning vector semantics. We relate it to Roger Shepard’s cognitive geometry and the structure of biological networks. We also describe limitations of deep learning for safety and regulation. We show how it fits into the rational agent framework and discuss what the next steps may be.

Published in: Technology
  • Be the first to comment

  • Be the first to like this

Exosphere Chile Talk: Semantics, Deep Learning, and the Transformation of Business

  1. 1. Chile: Semantics, Deep Learning, and the Transformation of Business Steve Omohundro, Ph.D. PossibilityResearch.com SteveOmohundro.com SelfAwareSystems.com http://discovermagazine.com/~/media/Images/Issues/2013/Jan-Feb/connectome.jpg
  2. 2. Economic Impact Deep Learning Neats vs. Scruffies Semantics The Future https://www.flickr.com/photos/danielfoster/14758510078/
  3. 3. Multi-Billion Dollar Investments • 2013 Facebook – AI lab, DeepFace • 2013 Yahoo - LookFlow • 2013 Ebay – AI lab • 2013 Allen Institute for AI • 2013 Google – DNNresearch, SCHAFT, Industrial Perception, Redwood Robotics, Meka Robotics, Holomni, Bot & Dolly, Boston Dynamics • 2014 IBM - $1 billion in Watson • 2014 Google - DeepMind $500 million • 2014 Vicarious - $70 million • 2014 Microsoft – Project Adam, Cortana • 2015 Fanuc – Machine Learning for Robotics • 2015 Toyota – $1 billion AI and Robotics Lab, Silicon Valley
  4. 4. McKinsey: $50 Trillion to 2025 http://www.mckinsey.com/insights/business_technology/disruptive_technologies
  5. 5. http://thismasquerade.me/wp-content/uploads/blogger/-84ftTg_azwY/Um8up3j4zTI/AAAAAAAABJ8/9_-dxg23UaY/s1600/9979402_ml.jpg AI Knowledge Work: $25 Trillion to 2025 Marketing, ERP, Big Data, Smart Assistants
  6. 6. Internet of Things: $15 Trillion to 2025 100 Billion devices by 2025 Cars, Appliances, Cameras, Meters, Wearables, etc. https://www.summitbusiness.net/images/Internet.jpg http://www.forbes.com/sites/gilpress/2014/08/22/internet-of-things-by-the-numbers-market-estimates-and-forecasts/
  7. 7. Robot Manufacturing: $10 Trillion to 2025 Work 24 hours/day No breaks, food, medical Don’t quit, get bored, get depressed Work anywhere Hazards OK Don’t leak secrets Work well with others Easy to replicate http://thisisrealmedia.com/2014/06/19/robotics-and-ethics-the-smart-car-by-ron-parlato/
  8. 8. Foxconn Technology Group • World’s largest contract manufacturer • Assembles 40% of all consumer electronics • iPhone, iPad, Kindle, Xbox, Playstation 4, etc. • 1.3 million employees, $8K salary • Employee suicides • “Foxbot” robots, cost $25K, 2nd generation now • Building 30K robots/year http://www.tomshardware.com/news/foxcponn-apple-iphone-ipad-robot,19088.html
  9. 9. 420 Chinese Robot Companies http://thestack.com/china-robot-market-overstimulated-291014 http://www.reuters.com/article/2014/10/28/china-robots-idUSL3N0RB2WX20141028 1500 Dongguan “Robot Replace Human” factories
  10. 10. March 2015: China Brain http://www.scmp.com/lifestyle/technology/article/1728422/head-chinas-google-wants-country-take-lead-developing Robin Li Yanhong, CEO of Baidu proposed a state-level Chinese initiative to develop AI “comparable to the Apollo space programme”.
  11. 11. https://osuwmcdigital.osu.edu/sitetool/sites/urologypublic/images/Robotics/robotic_surgery_table.jpg Health Care: $10 Trillion to 2025 Robot surgery, medical records, AI diagnosis
  12. 12. Self-Driving Vehicles: $10 Trillion by 2025 Disrupt Dealers, Insurance, Parking, Finance, Trucking, Taxis 10 million jobs http://zackkanter.com/2015/01/23/how-ubers-autonomous-cars-will-destroy-10-million-jobs-by-2025/ http://www.theverge.com/2014/5/28/5756852/googles-self-driving-car-isnt-a-car-its-the-future
  13. 13. Tesla: “Autopilot” mode http://zackkanter.com/2015/01/23/how-ubers-autonomous-cars-will-destroy-10-million-jobs-by-2025/http://www.flickr.com/photos/quikbeam/6896564084/ Google: Fully Self-Driving in 2020 Mercedes, GM, Volvo, Apple, Uber,… http://en.wikipedia.org/wiki/Autonomous_car
  14. 14. https://d185ox70mr1pkc.cloudfront.net/post_image_teaser/1403883171000-uber-force.png World’s largest job creator: 50,000 per month http://www.businessinsider.com/uber-offering-50000-jobs-per-month-to-drivers-2015-3 Uber valuation: $51 billion, 20% of fares http://www.wsj.com/articles/ubers-new-funding-values-it-at-over-41-billion-1417715938 Center for research on self-driving cars http://bits.blogs.nytimes.com/2015/02/02/uber-to-open-center-for-research-on-self-driving-cars/?_r=0 36 second wait, $.50/mile, 100% of fares http://zackkanter.com/2015/01/23/how-ubers-autonomous-cars-will-destroy-10-million-jobs-by-2025/
  15. 15. http://airwolf3d.com/wp-content/uploads/2012/05/3d-printer-v.5.5-airwolf3d1.jpg 3D Printing: $2 Trillion by 2025
  16. 16. April 2014: Chinese WinSun 3D printed 10 houses, 2100 sq ft, $4800 http://m.newsru.co.il/realty/20jan2015/3d_house_i101.html
  17. 17. WinSun 3D printed 12,000 sq ft villa http://3dprint.com/38144/3d-printed-apartment-building/ US Building construction: $1 Trillion/yr 5.8 million employees
  18. 18. https://venturescannerinsights.files.wordpress.com/2015/01/artificial-intelligence-map.jpg
  19. 19. https://venturescannerinsights.files.wordpress.com/2015/09/ai4.jpeg
  20. 20. Deep Learning Neural Nets http://www.nicta.com.au/content/uploads/2015/02/deep.jpg
  21. 21. Deep Learning Successes • Speech Recognition TIMIT 2009: Cortana, Skype, Google Now, Siri, Baidu, Nuance, etc. • Image Recognition ImageNet 2012 • Image Captioning 2014 • Natural Language: Sentiment 2013, Translation 2014, Semantics 2014 • Drug Discovery: Merck Challenge 2012 • DeepMind 49 Atari Video Games 2015
  22. 22. http://www-formal.stanford.edu/jmc/history/dartmouth/dartmouth.html • Natural Language • Neural Nets • Self-Improvement • Abstraction • Creativity 1955: “Artificial Intelligence” proposed
  23. 23. “Neats” vs. “Scruffies” http://news.stanford.edu/news/2003/june18/mccarthy-618.html http://www.bbc.co.uk/timelines/zq376fr 1963: John McCarthy Stanford AI Lab Mathematically Precise Thinking = Logical Inference Semantic Representations 1963: Marvin Minsky MIT MAC AI Group Self-Organized Adaptive Elements Machine Learning Emergent Semantics
  24. 24. 1957: Rosenblatt’s “Perceptron” http://www.rutherfordjournal.org/article040101.html “The embryo of an electronic computer that [the Navy] expects will be able to walk, talk, see, write, reproduce itself and be conscious of its existence." https://en.wikipedia.org/wiki/Perceptron https://upload.wikimedia.org/wikipedia/commons/3/31/Perceptron.svg http://bio3520.nicerweb.com/Locked/chap/ch03/3_11-neuron.jpg
  25. 25. 1969: Perceptrons can’t do XOR! http://www.i-programmer.info/images/stories/BabBag/AI/book.jpg Minsky & Papert https://constructingkids.files.wordpress.com/2013/05/minsky-papert-71-csolomon-x640.jpg http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/ietron/xor.gif
  26. 26. Multilayer Neural Nets http://opticalengineering.spiedigitallibrary.org/article.aspx?articleid=1714547 Backpropagation 1986 Rumelhart (1963 Bryson 1974 Werbos) Deep Learning 2007 Hinton (1989 LeCun 1992 Schmidhuber) http://www.nature.com/polopoly_fs/7.14689.1389093731!/image/deep-learning-graphic.jpg_gen/derivatives/landscape_400/deep-learning-graphic.jpg
  27. 27. “Neat” Software Compiler Source Code Lexical Analysis Parsing Semantic Analysis Code Generation Code Optimization
  28. 28. “Neat” Language Translator? Tokenization Stemming Tagging Parsing Logical Representation Translation
  29. 29. http://www.amazon.com/Comprehensive-Grammar-English-Language-General/dp/0582517346/ref=sr_1_2 1957 Chomsky Grammar "English as a Formal Language". In: Bruno Visentini (ed.): Linguaggi nella società e nella tecnica. Mailand 1970, 189–223. 1970 Montague Semantics 1792 Pages!
  30. 30. http://www.espressoenglish.net/order-of-adjectives-in-english/ Linguistic Rules are Complicated!
  31. 31. http://www.computer.org/csdl/mags/ex/2009/02/mex2009020008-abs.html http://www.mitpressjournals.org/doi/abs/10.1162/coli.2006.32.4.527#.VjWP0_mfM-U 2006: Simple n-gram models with lots of data beat complicated hand built linguistic models! 2009: And data is cheap and plentiful! Much cheaper than linguists or programmers!
  32. 32. 1962: Roger Shepard Cognitive Geometry http://link.springer.com/article/10.1007/BF02289630 https://psychlopedia.wikispaces.com/mental+rotation
  33. 33. Word2Vec – Mikolov 2013 • Distributional Semantics – Firth 1957 • Represent words by vectors • Close vectors represent similar contexts • Certain relations represented by translation: King – Man + Woman = Queen • Also tense, temperature, location, plurals,… http://deeplearning4j.org/word2vec.html
  34. 34. Why? Same context shift for all male -> female https://drive.google.com/file/d/0B7XkCwpI5KDYRWRnd1RzWXQ2TWc/edit?usp=sharing 2013 Mikolov: The man ate his lunch. The king ate his lunch. The woman at her lunch. The queen ate her lunch.
  35. 35. More Semantic Relations • Paris – France + Italy = Rome • Human – Animal = Ethics • Obama – USA + Russia = Putin • Library – Books = Hall • Biggest – Big + Small = Smallest • Ethical – Possibly + Impossibly = Unethical • Picasso – Einstein + Scientist = Painter • Forearm – Leg + Knee = Elbow • Architect – Building + Software = Programmer https://code.google.com/p/word2vec/ http://byterot.blogspot.com/2015/06/five-crazy-abstractions-my- deep-learning-word2doc-model-just-did-NLP-gensim.html http://arxiv.org/pdf/1301.3781.pdf http://deeplearning4j.org/word2vec.html
  36. 36. Marr’s “Neat” Vision Pipeline http://homepages.inf.ed.ac.uk/rbf/CVonline/LOCAL_COPIES/GOMES1/marr.html http://www.amazon.com/Vision-Computational-Investigation-Representation-Information/dp/0262514621/ref=sr_1_2
  37. 37. Deep Neural Net Face Recognition Google FaceNet, June 2015 Record accuracy 99.63% on Labeled Faces in the Wild dataset Cuts best previous error rate by 30% 22 layer feedforward net, 140M weights, 1.6 GFLOP/image, conv/pool/norm Trained on triples pushing same faces together, different apart http://arxiv.org/abs/1503.03832 https://github.com/cmusatyalab/openface CMU OpenFace, Oct. 2015 Open Source version of FaceNet 84.83% accuracy, <.1 training faces
  38. 38. Cheap Cameras + Face Recognition + Body Recognition = Brin’s “Transparent Society” http://www.ebay.com/sch/i.html?_nkw=cmos+image+sensor&_sop=15http://www.aliexpress.com/cheap/cheap-image-sensor-module.html $3.20 on Alibaba $2.95 on ebay http://fossbytes.com/facebook-can-now-recognize-you-in-photos-without-even-seeing-your-face/ http://thenextweb.com/dd/2015/10/15/watch-this-open-source-program-recognize-faces-in-real-time/ https://www.newscientist.com/article/mg21528835-600-cameras-know-you-by-your-walk/ http://www.amazon.com/Transparent-Society-Technology-Between-Privacy/dp/0738201448/ref=sr_1_1
  39. 39. https://github.com/Newmu/dcgan_code/tree/gh-pages
  40. 40. Biological Networks are Recurrent Gene network Chromosome 22 Human Metabolome Brain Connectome http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0028213#pone-0028213-g010 https://41.media.tumblr.com/tumblr_m5l6rzIqwc1r1171mo1_1280.jpg https://en.wikipedia.org/wiki/Hub_(network_science_concept)
  41. 41. http://karpathy.github.io/2015/05/21/rnn-effectiveness/ Image Classification Image Captioning Sentence Sentiment English->French Translation Real-time Video Frame Classification
  42. 42. Recurrent Net Hallucinates C Code http://karpathy.github.io/2015/05/21/rnn-effectiveness/ Karpathy: 464MB of C code, 3 layer LSTM, 10 million parameters
  43. 43. The rat escaped. The rat the cat attacked escaped. The rat the cat the dog chased attacked escaped. http://arlingtonva.s3.amazonaws.com/wp-content/uploads/sites/25/2013/12/rat.jpg
  44. 44. NeuralTalk and Walk Demo https://vimeo.com/146492001
  45. 45. https://www.youtube.com/watch?v=08Cl7ii6viY&feature=youtu.be&t=15m31s DeepMind Deep-Q Networkshttp://www.nature.com/nature/journal/v518/n7540/full/nature14236.html Feb. 2015: 49 Atari 2600 Games Raw pixels Same net all games Beat previous Ais Beat humans on half May 2015: 3D games TORCS racing Beat Ais from pixels May 2015: 100’s of games
  46. 46. Aerial Drones: $98 Billion by 2025 Delivery, Surveillance, Agriculture, Military, Police http://mint-tek.com/wp-content/uploads/2015/08/commercialdronesforhire.jpg http://www.businessinsider.com/the-market-for-commercial-drones-2014-2 http://www.flybestdrones.com/best-5-drones-with-camera-under-50-dollars/
  47. 47. Deep Learning Has Blindspots http://arxiv.org/abs/1412.1897
  48. 48. Other Issues • Typically have problems to solve rather than reinforcement signals • Want confidence that system solves problem • Want confidence in no unintended behaviors • Systems often have to obey legal, corporate, or design constraints http://78813809ba6486e732cd-642fac701798512a2848affc62d0ffb0.r60.cf2.rackcdn.com/465DAB1D-1F8E-4164-8D18-3BFD150E02F4.jpg
  49. 49. Technology Needs Semantics! • Analyzing camera, sensor, weather data • Better search, question answering, info • Analysis and optimization of business processes • Health monitoring, medical diagnosis • Financial markets trading, stabilization • Autonomous cars, trucks, boats, subs, planes • Pollution monitoring and cleanup • Improved robotic manufacturing • Software and Hardware design
  50. 50. Approaches to Semantics • Montague – map into Typed Lambda Calculus • Denotational – map into CS Domains • Mathematical – map into Set Theory • Categorical – map into Category Theory • Distributional – Statistics of Contexts http://engineering.missouri.edu/wp-content/uploads/australian-cloudy-sky.jpg Representation, Encoding, Learning, Communication, Reasoning
  51. 51. New Possibilities Coming Soon! PossibilityResearch.com http://flinttown.com/wp-content/uploads/2015/07/Fireworks.jpg

×