The importance of Supercomputing in Artificial Intelligence

169 views

Published on

Presentation at #27 CO-SESSION ARTIFICIAL INTELLIGENCE IN BUSINESS event

Published in: Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
169
On SlideShare
0
From Embeds
0
Number of Embeds
107
Actions
Shares
0
Downloads
4
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

The importance of Supercomputing in Artificial Intelligence

  1. 1. 1.  Everyday Tech and AI 2.  AI are not a new concept 3.  Why during this decade? 4.  Why Deep Learning now? 5.  Computation Democratization 6.  What about the software? 7.  AI will transform everything 8.  Computers can now teach themselves 9.  The new AI market 10.  What to do now?
  2. 2. Quantum leaps in the quality of a wide range of everyday technologies thanks to the Artificial Intelligence
  3. 3. Credits: h+ps://www.yahoo.com/tech/ba+le-of-the-voice-assistants-siri-cortana-211625975.html we are increasingly interacting with “our” computers by just talking to them Speech Recognition
  4. 4. Google Translate now renders spoken sentences in one language into spoken sentences in another, for 32 pairs of languages and offers text translation for 100+ languages. Natural Language Processing
  5. 5. Google Translate now renders spoken sentences in one language into spoken sentences in another, for 32 pairs of languages and offers text translation for 100+ languages. Natural Language Processing
  6. 6. Google Translate now renders spoken sentences in one language into spoken sentences in another, for 32 pairs of languages and offers text translation for 100+ languages. Natural Language Processing
  7. 7. Now our computers can recognize images and generate descriptions for photos in seconds.Computer Vision
  8. 8. All these three areas are crucial to unleashing improvements in robotics, drones, self-driving cars, etc. Source: h+p://ediCon.cnn.com/2013/05/16/tech/innovaCon/robot-bartender-mit-google-makr-shakr/
  9. 9. All these three areas are crucial to unleashing improvements in robotics, drones, self-driving cars, etc. Source: h+p://axisphilly.org/arCcle/military-drones-philadelphia-base-control/
  10. 10. All these three areas are crucial to unleashing improvements in robotics, drones, self-driving cars, etc. Source: h+p://fortune.com/2016/04/23/china-self-driving-cars/
  11. 11. Many of these breakthroughs have been made possible by a family of Artificial Intelligence techniques popularly known as DEEP LEARNING Although the greatest impacts of deep learning may be obtained when it is integrated into the whole toolbox of other AI techniques
  12. 12. Artificial Intelligence, Neural Networks, are not a new concepts!
  13. 13. John McCarthy coined the term Artificial Intelligence in the 1950s h+p://www.independent.co.uk/news/obituaries/john-mccarthy-computer-scienCst-known-as-the-father-of-ai-6255307.html
  14. 14. In 1958 Frank Rosenblatt built a prototype neural net, which he called the Perceptron Source: h+p://www.enzyklopaedie-der-wirtschaLsinformaCk.de/wi-enzyklopaedie/Members/wilex4/Rosen-2.jpg/image_preview
  15. 15. Even the FIB in Barcelona, was teaching AI in 1982
  16. 16. Why, Artificial intelligence has, all of a sudden, become the next big thing again during this decade?
  17. 17. Source: Economist , Feb 25th, 2010 h+p://www.economist.com/node/15579717 now AI algorithms can be “trained” by exposing them to large data sets that were previously unavailable.The data deluge
  18. 18. and the Computing Power necessary to implement AI algorithms is now available
  19. 19. Do you know what “my” computer was like in 1982?
  20. 20. Credits:http://www.ithistory.org/sites/default/files/hardware/facom230-50.jpg
  21. 21. Credits:http://www.ithistory.org/sites/default/files/hardware/facom230-50.jpg FACOM 230 – Fujitsu Instructions per second: few Mips * (M = 1.000.000) Processors : 1
  22. 22. Source:https://upload.wikimedia.org/wikipedia/ commons/7/76/BSC-Convex-240.JPG
  23. 23. Convex Computer C3480 Instructions per second: 800 Mips (400 Flops) Processors : 8 Source:https://upload.wikimedia.org/wikipedia/ commons/7/76/BSC-Convex-240.JPG
  24. 24. IBM RS6000 SP Instructions per second: 192.000 MFlops Processors : 128
  25. 25. MARENOSTRUM III - IBM Instructions per second: 1.000.000.000 MFlops Processors : 6046 (48448 cores)
  26. 26. Until then, the increase in computational power every decade of “my” computer, was mainly thanks to CPU improvements!
  27. 27. Why is Deep Learning so popular and in demand these days?
  28. 28. SOURCE: https://www.hpcwire.com/2016/11/23/nvidia-sees-bright-future-ai-supercomputing/?eid=330373742&bid=1597894 Since then, the increase in computational power for deep learning has not only been from CPU improvements … but also from the realization that GPUs (NVIDIA) were 20 to 50 times more efficient than traditional CPUs.
  29. 29. And Intel … (*) Intel spent more than $400 million to buy this deep-learning startup.
  30. 30. And Google revealed in May that for over a year it had been secretly using its own tailor-made chips, called tensor processing units, or TPUs, to implement applications trained by deep learning. SOURCE: Google
  31. 31. Marenostrum 4 will have more than 3,400 new generation Intel Xeon processors nodes & emerging technologies as Power + NVIDIA GPUs, Intel Knights Landing and Intel Knights Hill ARMv8, …
  32. 32. COMPUTING POWER is the real enabler!
  33. 33. However, now we are entering into an era of computation democratization for companies !
  34. 34. And what is “my/your” computer like now?
  35. 35. Source: h+p://www.google.com/about/datacenters/gallery/images And what is “my/your” computer like now?
  36. 36. 28.000 m2 Credits: h+p://datacenterfronCer.com/server-farms-writ-large-super-sizing-the-cloud-campus/ Huge data centers!
  37. 37. Foto: Google 28.000 m2
  38. 38. Foto: Google 28.000 m2
  39. 39. Foto: Google 28.000 m2
  40. 40. For those (experts) who want to develop their own software, cloud services like Amazon Web Services provide GPU-driven deep-learning computation services
  41. 41. And what about the software that we require for AI?
  42. 42. Plentiful open-source software have greased the innovation process
  43. 43. as has an open-publication ethic, whereby many researchers publish their results immediately on one database without awaiting peer-review approval.
  44. 44. And for “less expert” people, various companies are providing a working scalable implementation of ML/AI algorithms as a Service (AI-as-a-Service) Source: https://twitter.com/smolix/status/804005781381128192Source: http://www.kdnuggets.com/2015/11/machine-learning-apis-data-science.html
  45. 45. Artificial intelligence will transform everything
  46. 46. Even the food we eat or the beer we drink will be affected!
  47. 47. Even the food we eat or the beer we drink will be affected!
  48. 48. Source: http://edition.cnn.com/2013/05/16/tech/innovation/robot-bartender-mit-google-makr-shakr/
  49. 49. Source: http://edition.cnn.com/2013/05/16/tech/innovation/robot-bartender-mit-google-makr-shakr/
  50. 50. Robot bartender creates crowd-sourced cocktails
  51. 51. Such computers can now teach themselves
  52. 52. No human being has programmed a computer to perform any of the stunts described above. Expose a learning algorithm to terabytes of data to train it, and then allow the computer to figure out for itself how to proceed. Source: https://cs.byu.edu/artificial-intelligence-and-machine-learning
  53. 53. AlphaGo wasn’t designed to play Go, it learnt it by playing! Source: http://fortune.com/2016/04/23/china-self-driving-cars/ Source: https://gogameguru.com/alphago-defeats-lee-sedol-game-1/
  54. 54. And this can be applied to many sectors, not just for playing!
  55. 55. IBM has bought a handful of companies with vast stores of medical data databases and is using Artificial Intelligence to try to help doctors spot diseases more rapidly. http://www.techradar.com/news/calling-dr-watson-ibms-ai-helps-to-diagnose-diseases
  56. 56. The new Artificial Intelligence Market
  57. 57. Source: https://www.oreilly.com/ideas/the-new-artificial-intelligence-market Aman Naimat provides the results of a data- driven analysis into the U.S. industries and companies using or building AI products right now. September, 2016
  58. 58. Source: ‪@cosminnegruseri NIPS 2016 – Barcelona 5-10 December Thirtieth Annual Conference on Neural Information Processing Systems
  59. 59. NIPS 2016: many people 6000+ attendance
  60. 60. NIPS 2016: a lot of knowledge
  61. 61. NIPS 2016: many companies
  62. 62. Some demos @ NIPS2016 – Barcelona 5-10 December The new kind of store featuring the world’s most advanced shopping technology Spot robot by Boston Dynamics
  63. 63. Some demos @ NIPS2016 – Barcelona 5-10 December
  64. 64. Some demos @ NIPS2016 – Barcelona 5-10 December
  65. 65. What to do now?
  66. 66. “The Fourth Industrial Revolution, which includes developments in previously disjointed fields such as artificial intelligence … …, will cause widespread disruption not only to business models but also to labour markets … … 65% of children entering primary schools today will ultimately work in new job types and functions that currently don’t yet exist. ”
  67. 67. In the past, a lot of companies wished they had started thinking earlier about their Internet strategy.
  68. 68. In the past, a lot of companies wished they had started thinking earlier about their Internet strategy. I think in a few years from now there will be a number of companies that wish they had started thinking earlier about their AI strategy.
  69. 69. In the past, a lot of companies wished they had started thinking earlier about their Internet strategy. I think in a few years from now there will be a number of companies that wish they had started thinking earlier about their AI strategy.
  70. 70. http://www.JordiTorres.Barcelona JordiTorres@bsc.es - @JordiTorresBCN

×