Course Webpage Introduction to Artificial Intelligence http://cs.nyu.edu/~fergus/teaching/ai/pmwiki.php V22.0472-001 Fall 2009 Lecture 1: Introduction Lecture 1: Introduction Rob Fergus – Dept of Computer Science, Courant Institute, NYU Many slides over the course adapted from either Dan Klein, Stuart Russell or Andrew Moore People Course Timing/Location • Monday: 3.30 – 4.45pm • Wednesday: 3.30 – 4.45pm • Prof. Rob Fergus • Room 1221, 715 Broadway • Office Hours: Wednesday 5-6pm, Room 1226, 715 Broadway • Teaching Assistant: None at present • Let me know if you need card access to the 12 th floor Course Details Related Course Book: Russell & Norvig, AI: A Modern Approach, 2 nd Ed (Green one). • • Course will follow structure of UC Berkeley • Prerequisites: AI Course (CS188), as taught by Prof. Dan Linear algebra and some programming experience • • There will be a lot of statistics and programming Klein Klein • Work and Grading: • Four assignments divided into checkpoints • Programming: Python, groups of 1-2 • http://inst.eecs.berkeley.edu/~cs188/fa08/ • Written: solve together, write-up alone • 5 late days • Mid-term and final • Fixed scale • Academic integrity policy 1
Please Fill Out the Signup Sheet Announcements • Important stuff: • Python lab: Next Tuesday, 7pm-8pm in this room • Please go through python tutorial beforehand • First assignment on web soon • Communication: • Announcements: Course webpage (http://cs.nyu.edu/~fergus/teaching/ai/pmwiki.php ) • Course email: v22_0472_001_fa09@cs.nyu.edu • Questions? Today Sci-Fi AI? • What is AI? • Brief history of AI y • What can AI do? • What is this course? What is AI? Acting Like Humans? • Turing (1950) “Computing machinery and intelligence” “Can machines think?” → “Can machines behave intelligently?” • The science of making machines that: Operational test for intelligent behavior: the Imitation Game • Think like humans Think like humans Think rationally Think rationally Act like humans Act rationally • Predicted by 2000, a 30% chance of fooling a lay person for 5 minutes • In 2008 Loebner competition, top program (Elbot) fooled 3 out of 12 human judges. • Anticipated all major arguments against AI in following 50 years • Suggested major components of AI: knowledge, reasoning, language understanding, learning • Problem: Turing test is not reproducible or amenable to mathematical analysis http://kschnee.xepher.net/loebner/lpc2009/log1.txt 2
Thinking Like Humans? Thinking Rationally? • The cognitive science approach: • The “Laws of Thought” approach • 1960s ``cognitive revolution'': information-processing • What does it mean to “think rationally”? psychology replaced prevailing orthodoxy of behaviorism • Normative / prescriptive rather than descriptive • Scientific theories of internal activities of the brain • Logicist tradition: • What level of abstraction? “Knowledge'' or “circuits”? • Logic: notation and rules of derivation for thoughts • Cognitive science: Predicting and testing behavior of Cognitive science: Predicting and testing behavior of • • Aristotle: what are correct arguments/thought processes? Aristotle: what are correct arguments/thought processes? human subjects (top-down) • Direct line through mathematics, philosophy, to modern AI • Cognitive neuroscience: Direct identification from neurological data (bottom-up) • Problems: • Both approaches now distinct from AI • Not all intelligent behavior is mediated by logical deliberation • Both share with AI the following characteristic: • What is the purpose of thinking? What thoughts should I (bother to) have? The available theories do not explain (or engender) anything resembling human-level general intelligence • Logical systems tend to do the wrong thing in the presence of uncertainty • Hence, all three fields share one principal direction! Images from Oxford fMRI center Acting Rationally Rational Agents • Rational behavior: doing the “right thing” • An agent is an entity that perceives and acts (more • The right thing: that which is expected to maximize goal achievement, given the available information examples later) • Doesn't necessarily involve thinking, e.g., blinking • This course is about designing • Thinking can be in the service of rational action rational agents • Entirely dependent on goals! • Abstractly, an agent is a function • Irrational ≠ insane, irrationality is sub-optimal action I l l b l from percept histories to actions: f h • Rational ≠ successful • Our focus here: rational agents • Systems which make the best possible decisions given goals, evidence, and constraints • For any given class of environments and tasks, we seek the agent (or • In the real world, usually lots of uncertainty class of agents) with the best performance (define some utility function) • … and lots of complexity • Can pose as maximizing the expected utility • Usually, we’re just approximating rationality • Computational limitations make perfect rationality unachievable • “Computational rationality” a better title for this course • So we want the best program for given machine resources AI Adjacent Fields Neuroscience • Philosophy: • Center for Neural Science at NYU • Logic, methods of reasoning • Mind as physical system • How do brains process information? • Foundations of learning, language, rationality • Mathematics • Neurons in brain: • Formal representation and proof • Algorithms, computation, (un)decidability, (in)tractability • • P Probability and statistics b bilit d t ti ti • Psychology • Adaptation • Phenomena of perception and motor control • Experimental techniques (psychophysics, etc.) • Economics: formal theory of rational decisions • Linguistics: knowledge representation, grammar • Neuroscience: physical substrate for mental activity • Control theory: • homeostatic systems, stability • simple optimal agent designs • Explore with fMRI and other techniques 3
Human Brain vs Computer Sub-Fields of AI • Many problems have split off to form their own sub-areas of research Computer Human Brain Computational units 1 CPU, 10^9 gates 10^11 neurons Storage Units 10^10 bits RAM 10^11 neurons • Classical AI assumed that sensing the real • Classical AI assumed that sensing the real 10^11 bits disk 10^14 synapses world would be straightforward Cycle time 10^-9 sec 10^-3 sec Bandwidth 10^10 bits/sec 10^14 bits/sec Memory updates/sec 10^9 10^14 • Not so in practice Computer Vision Natural Language • Speech technologies • Automatic speech recognition (ASR) • Text-to-speech synthesis (TTS) • Dialog systems • Language processing technologies • Machine translation: M hi t l ti Aux dires de son président, la commission serait en mesure de le faire . According to the president, the commission would be able to do so . Pascal VOC 2008 Il faut du sang dans les veines et du cran . We must blood in the veines and the courage . • Information extraction • Information retrieval, question answering • Text classification, spam filtering, etc… Jitendra Malik Robotics Logic • Robotics • Logical systems • Part mech. eng. • Theorem provers • Part AI • Reality much • NASA fault diagnosis harder than simulations! • Question answering • Technologies • Vehicles • Methods: • Rescue • Soccer! • Deduction systems • Lots of automation… • Constraint satisfaction • In this class: • Satisfiability solvers (huge • We ignore mechanical aspects advances here!) • Methods for planning • Methods for control Images from stanfordracing.org, CMU RoboCup, Honda ASIMO sites Image from Bart Selman 4
Recommend
More recommend
