CS 104 Ar)ficial Intelligence Spring 2011 Acknowledgement: The course slides are based on materials used in AI classes
taught at other institutions.
Goals of this Course •  AI is a very broad field with many subareas –  Focus on the probabilis)c side of AI •  Since this is a graduate class –  You will read some recent research papers appearing in AI conferences –  Complete a class project where you explore some aspect of AI in some depth Class Overview •  Class Web page –  hHp://www.cs.dartmouth.edu/~CS104/ •  Review –  Organiza)onal details –  Textbook –  Schedule and syllabus –  Grading –  Academic honesty Class Overview •  Instructor: Tanzeem Choudhury –  office hours by appointment (210 Sudikoff) •  Lectures: MWF 10:00-­‐11:05 –  Loca)on: Sudikoff 214 •  Ar)ficial Intelligence: A Modern Approach (3rd Edi)on) by
Stuart Russell and Peter Norvig Textbook •  Some of our textbooks have a digital
version available, and it turns out
that ARTIFICIAL INTELLIGENCE: A
MODERN APPROACH is one such
)tle. •  This )tle retails new, in book form, at
$140.00. The digital download (good
for 180 days) retails @ $63.00. The
labels for the digital books are
created in store, and I have some on
the shelf now. Once ac)vated these
digital books are NOT RETURNABLE. Class Overview •  Paper presenta)on and cri)ques (30%) •  Project (60%) •  Class Par)cipa)on (10%) Tenta)ve Schedule •  Week 1 –  Introduc)on to AI (R&N Chapter 1) –  Agents (R&N Chapter 2) •  Week 2 –  Uncertainty (R&N Chapter 13) –  Uncertainty and Bayesian Networks (R&N Chapter 14) •  Week 3 –  Inference in Bayesian Networks (R&N Chapter 14) –  Hidden Markov Models R&N Chapter 15 •  Week 4 –  Project proposal –  Different Types of Learning (R&N Chapter 18) –  Learning with Incomplete Data (R&N Chapter 18) Tenta)ve Schedule •  Weeks 5 & 6 –  Papers •  Week 7 –  Project milestone –  Papers •  Week 8 –  Paper •  Week 9 –  Papers –  Project presenta)on •  Week 10 –  Project presenta)on Today’s Lecture •  What is intelligence? •  What is ar)ficial intelligence? •  A very brief history of AI •  An AI scorecard •  AI in prac)ce What is Intelligence? •  Intelligence: –  the capacity to learn, reason, understand and solve problems (Websters dic)onary) •  the ability to solve novel problems •  the ability to act ra2onally •  the ability to act like humans •  Ar)ficial Intelligence –  build and understand intelligent en))es or agents What’s involved in Intelligence? •  Ability to interact with the real world – to perceive, understand, and act –  e.g., speech recogni)on and understanding and synthesis –  e.g., image understanding –  e.g., ability to take ac)ons, have an effect What’s involved in Intelligence? •  Reasoning and Planning –  modeling the external world, given input –  solving new problems, planning, and making decisions –  ability to deal with unexpected problems, uncertain)es What’s involved in Intelligence? •  Learning and Adapta)on –  we are con)nuously learning and adap)ng –  our internal models are always being updated •  e.g., a baby learning to categorize and recognize animals Can Computers Talk? •  This is known as speech synthesis –  translate text to phone)c form •  e.g., fic))ous -­‐> fik-­‐)sh-­‐es –  use pronuncia)on rules to map phonemes to actual sound •  e.g., )sh -­‐> sequence of basic audio sounds Can Computers Talk? •  Difficul)es –  sounds made by this lookup approach sound unnatural –  sounds are not independent •  e.g., act and ac)on •  modern systems (e.g., at AT&T) can handle this preHy well –  a harder problem is emphasis, emo)on, etc •  humans understand what they are saying •  machines don’t: so they sound unnatural Can Computers Talk? •  Conclusion: –  NO, for complete sentences –  YES, for individual words Can Computers Recognize Speech? •  Speech Recogni)on: –  mapping sounds from a microphone into a list of words –  classic problem in AI, very difficult Can Computers Recognize Speech? •  Speech Recogni)on: –  mapping sounds from a microphone into a list of words –  classic problem in AI, very difficult •  Recognizing single words from a small vocabulary •  systems can do this with high accuracy (order of 99%) •  e.g., directory inquiries –  limited vocabulary (area codes, city names) –  computer tries to recognize you first, if unsuccessful hands you over to a human operator –  saves millions of dollars a year for the phone companies Recognizing human speech •  Recognizing normal speech is much more difficult –  speech is con)nuous: where are the boundaries between words? •  e.g., John’s car has a flat )re –  large vocabularies •  can be many thousands of possible words •  we can use context to help figure out what someone said –  background noise, other speakers, accents, colds, etc –  on normal speech, modern systems are only about 60-­‐70% accurate Recognizing human speech •  Conclusion: –  NO, normal speech is too complex to accurately recognize –  YES, for restricted problems (small vocabulary, single speaker) Can Computers Understand speech? •  Understanding is different to recogni)on: –  Time flies like an arrow •  assume the computer can recognize all the words •  how many different interpreta)ons are there? Can Computers Understand speech? •  Understanding is different to recogni)on: –  Time flies like an arrow •  assume the computer can recognize all the words •  how many different interpreta)ons are there? –  1. )me passes quickly like an arrow? –  2. command: )me the flies the way an arrow )mes the flies –  3. command: only )me those flies which are like an arrow –  4. )me-­‐flies are fond of arrows Can Computers Understand speech? •  Understanding is different to recogni)on: –  Time flies like an arrow •  assume the computer can recognize all the words •  how many different interpreta)ons are there? –  1. )me passes quickly like an arrow? –  2. command: )me the flies the way an arrow )mes the flies –  3. command: only )me those flies which are like an arrow –  4. )me-­‐flies are fond of arrows •  only 1. makes any sense, –  but how could a computer figure this out? –  clearly humans use a lot of implicit commonsense knowledge in communica)on Can Computers Understand speech? •  Conclusion: NO, much of what we say is beyond the capabili)es of a computer to understand at present Can Computers Learn and Adapt ? •  Learning and Adapta)on –  consider a computer learning to drive on the freeway –  we could teach it lots of rules about what to do –  or we could let it drive and steer it back on course when it heads for the embankment Can Computers Learn and Adapt ? •  Machine learning allows computers to learn to do things without explicit programming –  requires some set-­‐up •  Conclusion: YES, computers can learn and adapt, when presented with informa)on in the appropriate way Can Computers see ? •  Recogni)on v. Understanding (like Speech) –  Recogni)on and Understanding of Objects in a scene •  look around this room •  you can effortlessly recognize objects •  human brain can map 2d visual image to 3d map •  Why is visual recogni)on a hard problem? Can Computers see ? •  Conclusion: –  mostly NO: computers can only see certain types of objects under limited circumstances –  YES for certain constrained problems (e.g., face recogni)on) Can computers plan and make op)mal decisions? •  Intelligence –  involves solving problems and making decisions and plans –  e.g., you want to take a holiday in Brazil •  you need to decide on dates, flights •  you need to get to the airport, etc •  involves a sequence of decisions, plans, and ac)ons •  What makes planning hard? –  the world is not predictable: •  your flight is canceled or there’s a backup on the 405 –  there are a poten)ally huge number of details •  do you consider all flights? all dates? –  no: commonsense constrains your solu)ons –  AI systems are only successful in constrained planning problems Can computers plan and make op)mal decisions? •  Conclusion: NO, real-­‐world planning and decision-­‐making is s)ll beyond the capabili)es of modern computers –  excep)on: very well-­‐defined, constrained problems