2017 Predictive Analytics Symposium

Transcription

1 2017 Predictive Analytics Symposium Session 36, Artificial Intelligence--Science Fiction or Reality? Moderator: Satadru Sengupta Presenter: Boyi Xie,Ph.D. SOA Antitrust Compliance Guidelines SOA Presentation Disclaimer

2 Artificial Intelligence Science Fiction or Reality? Boyi Xie, SOA Predictive Analytics Symposium, 0 9 / 15 / 2 017

3 Artificial Intelligence When most of us try to picture an artificial intelligence, of course we think first of a robot. Human performance Rationality Thought processes and reasoning Behavior System that think like humans Systems that act like humans Systems that think rationally Systems that act rationally Foundations of Artificial Intelligence Philosophy Economics Psychology Control theory and Cybernetics Mathematics Neuroscience Computer engineering Linguistics Weak AI: Machines could possibly act intelligently, or act as if they were intelligent. Strong AI: Machines that do so are actually thinking, as opposed to simulating thinking. 2

4 Deep Blue A chess-playing computer developed by IBM. It is known for being the first computer chess-playing system to win both a chess game and a chess match against a reigning world champion under regular time controls. Deep Blue studies thousands of master games. It has an evaluation function that determines how important a position is compares to other alternatives. It search to a depth between six to eight moves to a maximum of twenty or even more moves in some situations. Interestingly, a bug in Deep Blue s software led to a seemingly random move which Kasparov misattributed to super intelligence. / wp-content/ uploads/ 2014/ 12/ kasparov-vs-ibm-deep-blue-640x414.jpg 3

5 AlphaGo A program developed by DeepMind (found in 2010 and acquired by Alphabet s Google in 2014) that plays the board game Go. In March 2016, it beat Lee Sedol in a five-game match, the first time a computer Go program has beaten a 9-dan professional. It was chosen by Science as one of the Breakthrough of the Year runners-up. Uses a combination of machine learning and tree search techniques Implemented using deep neural network technology It was initially trained to mimic human play by attempting to match the moves of expert players from recorded historical games, using a database of around 30 million moves. Later trained by being set to play large numbers of games against itself, using reinforcement learning to improve / e n.wikipe dia.org/ wiki/ AlphaGo 4

6 Turing Test Devised in 1950 by mathematician, computer scientist, and World War II code breaker Alan Turing to determine whether a machine can exhibit intelligence. The test is for a program to have a conversation (via online typed messages) with an interrogator for 5 minutes. The interrogator then has to guess if the conversation is with a program or a person. The program passes the test if it fools the interrogator 30% of the time. Few AI researchers pay attention to the Turing Test, preferring to concentrate on their systems performance on practical tasks, rather than the ability to imitate humans. 5

7 Chinese Room In 1980, John Searle described a hypothetical system that is clearly running a program and passes the Turing Test, but does not understand anything of its inputs and outputs. Assume you only speak English, and image yourself in a room with two slits, a rule book written in English, and some scratch paper. Someone slides you some Chinese characters through the first slit, you follow the instructions in the book, transcribing characters instructed onto the scratch paper, and slide the resulting sheet out the second slit. To people outside, it appears the room speaks Chinese, yet you do not understand a word of Chinese. His conclusion is that running the appropriate program (i.e., having the right outputs) is not a sufficient condition for being a mind. The aim of the Chinese Room argument is to refute strong AI the claim that running the right sort of program necessarily results in a mind. 6

8 ELIZA ELIZA A Computer Program for the Study of Natural Language Communication between Man and Machine ELIZA was intended to simulate or caricature the conversation between a psychoanalyst and a patient, with the machine in the role of analyst. ELIZA: Good afternoon. Please type in your name. [The visitor did so, and ELIZA continued with some polite preliminaries.] ELIZA: What brought you here to see me today? VISITOR: Oh, nothing much. I m feeling a bit tired, that s all. ELIZA: Why do you think you re feeling a bit tired? VISITOR: Well, I ve been traveling a lot, and away from home. ELIZA: Tell me about your family. [The conversation suddenly became intimate. The visitor began to disclose his worries about his wife, his children, his distance both geographical and emotional from them.] Partly ELIZA had to do with the illustrations of mutual understanding as Joseph Weizenbaum explained: What I mean here is the cocktail party conversation. Someone says something to you that you really don t fully understand, but because of the context and lost of other things, you are in fact able to give a response which appears appropriate, and in fact the conversation continues for quite a long time. 7

9 IBM Watson A question answering system developed by IBM s DeepQA project. It was specifically developed to answer questions on the Jeopardy! show. In 2011, Watson completed the Jeopardy! winning the first prize of $1 million. Watson had access to 200 million pages of structured and structured content of four terabytes, including the full text of Wikipedia / en.wikipedia.org/ wiki/ File:DeepQA.svg 8

10 NELL: Never Ending Language Learning Research Goal To build a never-ending machine learning system that acquires the ability to extract structured information from unstructured web pages. If successful, this will result in a knowledge base (i.e., a relational database) of structured information that mirrors the content of the Web. We call this system NELL (Never-Ending Language Learner). Never-Ending Learning. T. Mitchell, W. Cohen, E. Hruschka, P. Talukdar, J. Betteridge, A. Carlson, B. Dalvi, M. Gardner, B. Kisiel, J. Krishnamurthy, N. Lao, K. Mazaitis, T. Mohamed, N. Nakashole, E. Platanios, A. Ritter, M. Samadi, B. Settles, R. Wang, D. Wijaya, A. Gupta, X. Chen, A. Saparov, M. Greaves, J. Welling. In Proceedings of the Conference on Artificial Intelligence (AAAI),

11 NEIL: Never Ending Image Learner Research Goal Build visual knowledge bases from Internet images, and enrich these knowledge bases by automatically discovering objects and their segmentations from noisy Internet images. Running since July 15, Analyzed 5 million Images, Labeled 0.5 million images and Learned 3000 Common sense relationships. Never-Ending Image Learner: Xinlei Chen, Abhinav Shrivastava, Abhinav Gupta. NEIL: Extracting Visual Knowledge from Web Data. In International Conference on Computer Vision (ICCV),

12 From Text to Scene Generation The Text2Scene project aims to explore how to automatically generate 3D scenes from a natural text description. This project attempts to learn such knowledge from a dataset of scenes and use the learned priors to infer missing constraints when generating a scene. Text2Scene ( /nlp.stanford.edu/projects/text2scene.shtml) Angel Chang, Will Monroe, Manolis Savva, Christopher Potts, and Christopher D. Manning. ACL

13 From Text to Scene Generation WordsEye ( / a image floor. a man is on the floor. the sun's azimuth is 10 degrees. a 1st 1.5 foot tall and 3 foot deep and 1.5 foot wide 40% dark tan cube is -1.5 foot above and -1 foot in front of the man. it leans to the front. a woman is 1 foot to the right of and in front of the man. a 2nd 1.5 foot tall and 3 foot deep and 1.5 foot wide 40% dark tan cube is -1.5 foot above and -1 foot in front of the woman. it leans to the front. a 1st small 80% yellow moon is 1 foot above and 2 feet behind the woman. a 2nd 1 foot tall flat moon is in front of and -.8 foot to the left of the 1st moon. a small tree is behind the floor. the ground is clear. Watching the eclipse at work today. 12

14 Boston Dynamics Boston Dynamics began as a spin-off from MIT, where they developed the robots that ran and maneuvered like animals. They combine the principles of dynamic control and balance with sophisticated mechanical designs, cutting-edge electronics and software for perception, navigation, and intelligence. Boston Dynamics 13

15 Applications of Mode rn Artificial Inte llige nce Find patterns in data sets to predict future outcomes and trends. Are you interested in buying the book Harry Potter? Will Roger Federer win Rafael Nadal in the French Open final? Are you looking for fruits or electronics when you search with keyword apple? Are you interested in a vacation to Hawaii? Do you want to connect on Facebook?

16 Branches of Artificial Inte llige nce Natural Language Processing Analyze textual data written in human language, from words, syntactic structure, to semantic meaning. Machine Learning Research on fundamental methods in statistical analysis on how to reason, learn and make inference. Computer Vision Process images and videos. Recognize objects, identifying depth of objects, direction of light, and the inte rpretation of image s. Speech Recognition Process signals and transform waveforms into words and sentences. Robotics Artificial Inte llige nce Affective Computing Studies the computing that relates to, arises from, or deliberately influences emotion or other affective phenomena. Design and build machineries by using computer systems for their control, sensory feedback, and information processing. 15

17 Reference Stuart Russell and Peter Norvig. Artificial Intelligence: A Modern Approach. Pamela McCorduck Machines who Think: A Personal Inquiry into the History and Prospects of Artificial Intelligence. AK Peters Ltd. wp-content/ uploads/ 2014/ 12/ kasparov-vs-ibm-deep-blue-640x414.jpg / en.wikipedia.org/ wiki/ AlphaGo T. Mitchell, W. Cohen, E. Hruschka, P. Talukdar, J. Betteridge, A. Carlson, B. Dalvi, M. Gardner, B. Kisiel, J. Krishnamurthy, N. Lao, K. Mazaitis, T. Mohamed, N. Nakashole, E. Platanios, A. Ritter, M. Samadi, B. Settles, R. Wang, D. Wijaya, A. Gupta, X. Chen, A. Saparov, M. Greaves, J. Welling. In Proceedings of the Conference on Artificial Intelligence (AAAI), Xinlei Chen, Abhinav Shrivastava, Abhinav Gupta. NEIL: Extracting Visual Knowledge from Web Data. In International Conference on Computer Vision (ICCV), Angel Chang, Will Monroe, Manolis Savva, Christopher Potts, and Christopher D. Manning. ACL WordsEye ( / ) Boston Dynamics / Wikipedia pages and images of machine learning topics Other papers published in academic journals and conferences 16

18 17

19 Legal notice 2017 Swiss Re. All rights reserved. You are not permitted to create any modifications or derivative works of this presentation or to use it for commercial or other public purposes without the prior written permission of Swiss Re. The information and opinions contained in the presentation are provided as at the date of the presentation and are subject to change without notice. Although the information used was taken from reliable sources, Swiss Re does not accept any responsibility for the accuracy or comprehensiveness of the details given. All liability for the accuracy and completeness thereof or for any damage or loss resulting from the use of the information contained in this presentation is expressly excluded. Under no circumstances shall Swiss Re or its Group companies be liable for any financial or consequential loss relating to this presentation. 18