1. Learning from Observations Copyright, 1996 © Dale Carnegie & Associates, Inc. Chapter 18 Spring 2004

2. CS 471/598 by H. Liu2 Learning agents Improve their behavior through diligent study of their own experiences. Acting -> Experience -> Better Acting We’ll study how to make a learning agent to learn; what is needed for learning; and some representative methods of learning from observations

3. CS 471/598 by H. Liu3 A general model What are the components of a learning agent? Learning element - learn and improve (Fig 2.15) Performance element - an agent itself to perceive & act Problem generator - suggest some exploratory actions Critic - provide feedback how the agent is doing The design of an LA is affected by four issues: prior info, feedback, representation, performance

4. CS 471/598 by H. Liu4 What do we need Components of the performance element Each component should be learnable given feedback Representation of the components Propositional Logic, FOL, or others Available feedback Supervised, Reinforcement, Unsupervised Prior knowledge Nil, some, (Why not all?) Put it all together as learning some functions

5. CS 471/598 by H. Liu5 Inductive Learning Data described by examples an example is a pair (x, f(x)) Induction - given a collection of examples of f, return a function h that approximates f. Data in Fig 18.3 Concepts about learning (explained using Fig 18.1)  Hypothesis  Bias Learning incrementally or in batch

6. CS 471/598 by H. Liu6 Some questions about inductive learning Are there many forms of inductive learning? We’ll learn some Can we achieve both expressiveness and efficiency? How can one possibly know that one’s learning algorithm has produced a theory that will correctly predict the future? If one does not, how can one say that the algorithm is any good?

7. CS 471/598 by H. Liu7 Learning decision trees A decision tree takes as input an object described by a set of properties and outputs yes/no “decision”. One of the simplest and yet most successful forms of learning To make a decision “wait” or “not wait”, we need information such as … (page 654 for 10 attributes for the data set in Fig 18.3) Patrons(Full)^WaitEstimate(0-10)^Hungry(N)=>WillWait

8. CS 471/598 by H. Liu8 Let’s make a decision Where to start?

9. CS 471/598 by H. Liu9 Expressiveness of a DT Continued from page 7 - A possible DT (e.g., Fig 18.2 ) The decision tree language is essentially propositional, with each attribute test being a proposition. Any Boolean functions can be written as a decision tree (truth tables DTs) DTs can represent many functions with much smaller trees, but not for all Boolean functions (parity, majority)

10. CS 471/598 by H. Liu10 How many different functions are in the set of all Boolean functions on n attributes? How to find consistent hypotheses in the space of all possible ones? And which one is most likely the best?

11. CS 471/598 by H. Liu11 Inducing DTs from examples Extracting a pattern (DTs) means being able to describe a large number of cases in a concise way - a consistent & concise tree. Applying Occam’s razor: the most likely hypothesis is the simplest one that is consistent with all observations. How to find the smallest DT? Examine the most important attribute first (Fig 18.4) Algorithm (Fig 18.5, page 658) Another DT (Fig 18.6)

12. CS 471/598 by H. Liu12 Choosing the best attribute A computational method - information theory Information - informally, the more surprise you have, the more information you have; mathematically, I(P(v1),…,P(vn)) = sum[-P(vi)logP(vi)]  I(1/2,1/2) = 1  I(0,1) = (1,0) = 0 Information alone can’t help much to answer “what is the correct classification?”.

13. CS 471/598 by H. Liu13 Information gain - the difference between the original and the new info requirement: Remainder(A) = p1*I(B1)+…+pn*I(Bn) where p1+…+pn = 1 Gain(A) = I(A) - Remainder(A)

14. CS 471/598 by H. Liu14 Which attribute? Revisit the example of “Wait” or “Not Wait” using your favorite 2 attributes.

15. CS 471/598 by H. Liu15 Assessing the performance A fair assessment: the one the learner has not seen. Errors Training and test sets: Divide the data into two sets Learn on the training set Test on the test set If necessary, shuffle the data and repeat Learning curve - “happy graph” (Fig 18.7)

16. CS 471/598 by H. Liu16 Practical use of DT learning BP’s use of GASOIL Learning to fly on a flight simulator An industrial strength system - Quinlan’s C4.5 Who’s the next hero?

17. CS 471/598 by H. Liu17 Some issues of DT applications Missing values Multivalued attributes Continuous-valued attributes

18. CS 471/598 by H. Liu18 Why learning works? How can one possibly know that his/her learning algorithm will correctly predict the future? How do we know that h is close enough to f without knowing f? Computational learning theory has provided some answers. The basic idea is that because any wrong h will make an incorrect prediction, it will be found out with high probability after a small number of examples. So, if h is consistent with a sufficient number of examples, it is unlikely to to seriously wrong - probably approximately correct (PAC). Stationarity assumption – Tr and Te have the same probability distribution

19. CS 471/598 by H. Liu19 Summary Learning is essential for intelligent agents dealing with the unknowns improving its capability over time All types of learning can be considered as learning an accurate representation h of f. Inductive learning - f from data to h Decision trees - deterministic Boolean functions