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.



Published on

  • Be the first to comment


  1. 1. CS451/CS551/EE565 ARTIFICIAL INTELLIGENCE Learning & Connectionism 12-04-2006 Prof. Janice T. Searleman, jetsza
  2. 2. Outline  Learning Agents  Neural Nets Reading Assignment: AIMA Chapter 18, Learning from Observations Chapter 19, sections 19.1, Logical Formulation of Learning Chapter 20, section 20.5, Neural Networks Final Exam: Mon, 12/11/06, 8:00 am, SC342 HW#7 posted; due: Wed. 12/06/06
  3. 3. What is Learning?  memorizing something  learning facts through observation and exploration  improving motor and/or cognitive skills through practice  organizing new knowledge into general, effective representations
  4. 4. Some types of learning  Rote learning  Reinforcement learning – an agent interacting with the world makes observations, takes actions, and is rewarded or punished; it should learn to choose actions in such a way as to obtain a lot of reward  Supervised induction (with a teacher) – given a set of input/output pairs, find a rule that does a good job of predicting the output associated with a new input  Unsupervised induction  Analogy (case-based) learning  Evolutionary learning
  5. 5. Learning Rote learning Samuel‟s checker program … by taking advice FOO … from problem-solving STRIPS (Shakey) experience … from examples & Winston‟s concept learner counterexamples … by parameter adjustment Samuel‟s checker program … by chunking SOAR … by analogy … by discovery AM/EURISKO Genetic learning Connectionism (Neural Net)
  6. 6. Learning agents
  7. 7. Learning  The idea behind learning is that percepts should not only be used for acting now, but also for improving the agent‟s ability to act in the future.  In the psychology literature, learning is considered one of the keys to human intelligence.
  8. 8. Learning element  Design of a learning element is affected by  Which components of the performance element are to be learned  What feedback is available to learn these components  What representation is used for the components  Type of feedback:  Supervised learning: correct answers for each example  Unsupervised learning: correct answers not given  Reinforcement learning: occasional rewards
  9. 9. Rote Learning  This is the simplest form of learning.  It involves nothing more than memorizing experiences, e.g., sensor inputs, actions taken, rewards received.  It‟s surprisingly effective!
  10. 10. Machine learning programs for classification (concept learning)  Assume you have a goal concept that you are trying to learn, called the target concept. Your guesses or approximations of the target concept are called hypotheses.  Anexample target concept might be a description of diseased soybean plants.  An object (fact) which is used to help learn the goal concept is called an instance or an example. It can also be called a case.  An instance/example x is described by a vector of features, also called attributes , i.e., x = <x1,……,xn>.
  11. 11. Classification tasks  Many engineering and diagnosis tasks involve classification or prediction, e.g.,  Parts inspection: classify parts into: defective or OK.  Mammogram analysis: given a mammogram, estimate the probability that is normal, pre-cancerous or cancerous.  Document understanding: given a rectangular region from a scanned region, classify it as text or graphics.  Soybean plant analysis: classify plants into: diseased or not diseased.
  12. 12. Machine learning programs for classification (concept learning)  Given: A labeling function f that maps feature vectors into a discrete set of k classes. That is, f(x) in {0,1,……,k-1}. Often, there are only 2 classes, called “positive” (+) and “negative” (-).  Represent each training example as a pair (x,f(x)). These are the examples that will be used for learning the concept.  Problem: From a set of (x,f(x)) pairs, learn the target concept f.
  13. 13. The learning problem  Given <x,f(x)> pairs, infer f x f(x) Given a finite sample, it is often 1 1 impossible to guess the true function f. 2 4 3 9 Approach: Find some pattern (called a hypothesis) in 4 16 the training examples, and assume 5 ? that the pattern will hold for future examples too.
  14. 14. Hypothesis or model selection  What class of hypotheses (models) should we consider?  Assume f is a set of rules. Then the space of hypotheses consists of rule sets.  Assume f is a simple polynomial. Then the space of hypotheses consists of simple polynomials. Regression could be used to learn f.  Assume f can be expressed as a decision tree. Then the space of hypotheses consists of decision trees. Decision tree learning can be applied.  Assume f can be expressed as a neural network. Then the space of hypotheses consists of neural nets, and backprop can be used to learn the weights.  Let H be the space of all possible hypotheses H that a learning program considers. Then the learner seeks the H in H that “fits” the given data the “best.” This is a process of search through the space of possible hypotheses in H.
  15. 15. Learning decision trees Problem: decide whether to wait for a table at a restaurant, based on the following attributes: 1. Alternate: is there an alternative restaurant nearby? 2. Bar: is there a comfortable bar area to wait in? 3. Fri/Sat: is today Friday or Saturday? 4. Hungry: are we hungry? 5. Patrons: number of people in the restaurant (None, Some, Full) 6. Price: price range ($, $$, $$$) 7. Raining: is it raining outside? 8. Reservation: have we made a reservation? 9. Type: kind of restaurant (French, Italian, Thai, Burger) 10. WaitEstimate: estimated waiting time (0-10, 10-30, 30- 60, >60)
  16. 16. Attribute-based representations  Examples described by attribute values (Boolean, discrete, continuous)  E.g., situations where I will/won't wait for a table:  Classification of examples is positive (T) or negative (F)
  17. 17. Connectionism
  18. 18. Recall: Physical Symbol System  A physical symbol system is a machine that produces through time an evolving collection of symbol structures. Such a system exists in a world of objects wider than just these symbolic expressions themselves. The Physical Symbol System Hypothesis  A physical symbol system has the necessary and sufficient means for general intelligent action. PSS => Intelligence Intelligence => PSS Well…maybe ???
  19. 19. Symbols & The Main Goals of AI  Engineering: Build intelligent systems  Lots of fantastic symbolic AI systems for a multitude of specialized tasks….and many more to come!  But general intelligent systems are a major problem, since common sense is hard to represent and reason with symbolically.  Science: Understand natural intelligence via computers  Cognitive Science founded by symbolic AI researchers.  But they took the metaphor too far.  Organisms clearly compute, but not necessarily:  as Von Neumann computers (i.e. serially)  as Logic theorem provers (i.e. mathematically complete & consistent)  with symbols!! …We can interpret the reasoning process as symbolic, but the underlying mechanism may not be. The ends don‟t explain the means!
  20. 20. The Intelligence Spectrum  Robot (Moravec ,1999) Calculate Reason Sense & Act Humans Computers ■ On the fringes:  Humans are slow, error-prone calculators  Robots sense and act no better (and much slower) than frogs ■ The battle for middle ground:  Deep Blue beat the best human chess player  But minimax search ≠ “reasoning”.
  21. 21. GOFAI -vs- The New AI Calc Reason Sense & Act GOFAI New AI  GOFAI (“Good Old Fashioned AI) ■ Disembodied reasoning systems can‟t plug-and-play on robots. ■ Lack of common sense => no general human reasoning abilities.  New AI ■ Embodied S&A gives basis for common sense but has not yet scaled up to sophisticated human-like abstract reasoning.
  22. 22. Evolutionary Progressions along the Intelligence Spectrum Living organisms Computers Sense & Act: 10,000,000+ years. 15+ years Reason: 100,000+ 30+ years Calculate 1,000+ 50+ years  Evolution of reasoning was tightly constrained and influenced by sensorimotor capabilities. Else extinction!  GOFAI systems are often in their own little worlds, making unreasonable assumptions about independent sensorimotor apparatus.  To achieve AI‟s scientific goal of understanding human intelligence, the road from sense-and-act to reasoning via simulated evolution may be the only way.  But, to achieve AI‟s engineering goals, both approaches seem important.
  23. 23. The Situated & Embodied AI Hypothesis  Complex intelligence is better understood and more successfully embodied in artifacts by working up from low-level sensory-motor agents than from abstract cognitive mechanisms of rationality (e.g. logic, means-ends analysis, etc.).  Cognitive Incrementalism: Cognition (and hence common sense) is an extension of sensorimotor behavior. Brooks, Steels, Pfeifer, Scheier, Beer, Nolfi, Floreano…
  24. 24. The Artificial Life Approach to AI Simple Robots Cellular Automata 22222222 2170140142 2022222202 272 212 Simulated Real Worlds 212 212 202 212 272 212 21222222122222 20710710711111 2222222222222 Langton Loop ■ Synthetic: Bottom-up, multiple interacting agents ■ Self-Organizing: Global structure is emergent. ■ Self-Regulating: No global/centralized control. ■ Adaptive: Learning and/or evolving ■ Complex: On the edge of chaos; dissipative
  25. 25. Adaptation  Key focus of Situated & Embodied AI (i.e., Alife AI)  But now, often at level of simple organisms (ants, flies, frogs, etc.)  Machine Learning (ML) is also a key part of GOFAI.  Alife AI is very interested in subsymbolic ML techniques:  Artificial Neural Networks (ANNs)  Evolutionary Algorithms (EAs)  Learning: agents modify their own behavior (normally to improve performance) in their lifetime.  Evolution: populations of agents change their behavior over the course of many generations.  Both: Evolving populations of learning agents
  26. 26. Neural Networks  Also „connectionism‟, „Parallel Distributed Processing‟, „subsymbolic AI‟  AI technique  Analogous to processes in the brain  “Intelligence emerges from the interactions of large numbers of simple processing units” (Rumelhart et al., 1986)  Roughly based on brains – some simplification is made
  27. 27. Real Neuron from Searleman & Searleman, Introduction to Cognition
  28. 28. Two interacting neurons  Excitatory (E) and Inhibitory (I) impulses (from Searleman & Searleman, Introduction to Cognition)
  29. 29. Neurons NeuroPhysiology Synapses Axon Dendrites • Dense: Human brain has 1011 neurons, 1014 synapses • Highly Interconnected: Human neurons have 104 fan-in. • Neurons firing: send action potentials (APs) down the axons when sufficiently stimulated by SUM of incoming APs along the dendrites. • Neurons can either stimulate or inhibit other neurons. • Synapses vary in transmission efficiency
  30. 30. Features of the human brain  Robust – fault tolerant and degrade gracefully  Flexible -- can learn without being explicitly programmed  Can deal with fuzzy, probabilistic information  Is highly parallel
  31. 31. Connectionist models  Key intuition: Much of intelligence is in the connections between the 10 billion neurons in the human brain.  Neuron switching time is roughly 0.001 second; scene recognition time is about 0.1 second. This suggests that the brain is massively parallel because 100 computational steps are simply not sufficient to accomplish scene recognition.  Development: Formation of basic connection topology  Learning: Fine-tuning of topology + Major synaptic- efficiency changes. The matrix IS the intelligence!