1 / 16

Analytical Learning

Analytical Learning. Introduction Learning with Perfect Domain Theories Explanation-Based Learning Search Control Knowledge Summary. Discovering New Features. The Prolog-EBG system we described before can formulate new features that do not show up in the examples.

Download Presentation

Analytical Learning

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Analytical Learning • Introduction • Learning with Perfect Domain Theories • Explanation-Based Learning • Search Control Knowledge • Summary

  2. Discovering New Features The Prolog-EBG system we described before can formulate new features that do not show up in the examples. Example: Volume * Density > 5 (derived from the domain theory) This is different from neural networks using hidden nodes. why?

  3. Discovering New Features This is different from neural networks using hidden nodes. why? Output nodes Internal nodes Input nodes

  4. Inductive Bias in Explanation-Based Learning What is the inductive bias of explanation based learning? The hypothesis h follows deductively from D and B D: database B: Background knowledge Bias: Prefer small sets of maximally general Horn Clauses

  5. Search Control Knowledge Problem: learning to speed up search programs. This is also called “speedup learning” Final state Initial state How do we improve our search control strategy to find a solution quickly?

  6. Search Control Knowledge Examples include: playing chess scheduling and optimization problems. Problem formulation: S: set of possible search states O: set of legal operators (transform one state into another state) G: predicate over S indicating the goal states

  7. Search Control Knowledge Learn an algorithm to stack the blocks so that it reads “UNIVERSAL” Example: N E S V U L A I R

  8. Search Control Knowledge S: all possible configurations of blocks on the table O: {(MS x) move block x to stack if x is on table, (MT x) move block x to table if x is on the stack} G: G(si) = true if si is the configuration where the blocks read UNIVERSAL N E S R V U L A I

  9. Prodigy Prodigy is a planning system. Input: state space S and operators O. Output: A sequence of operators that lead from the initial state to the final state. Prodigy uses a means-end planner: we decompose goals into subgoals: Goal Subproblems

  10. Prodigy Example: Goal: accommodate blocks to read “UNIVERSAL” On(A,L) On(U,N) On(N,I) On(I,V) On(V,E) … Question: what subgoal should be attacked first? Answer is given by search control knowledge

  11. Prodigy and Explanation Based Learning Prodigy defines a set of target concepts to learn, e.g., which operator given the current state takes you to the goal state? An example of a rule learned by Prodigy in the block- stacking problem is: IF One subgoal to be solved is On(x,y) AND One subgoal to be solved is On(y,z) THEN Solve the subgoal On(y,z) before On(x,y)

  12. Prodigy and Explanation Based Learning The rationale behind the rule is that it would avoid a conflict when stacking blocks. Prodigy learns by first encountering a conflict, then explaining the reason for the conflict and creating a rule like the one above. Experiments show an improvement in efficiency by a factor of two to four.

  13. Problems with EBL • The number of control rules that must be learned is very large. • If the control rules are many, much time will be spent looking for the best rule. Utility analysis is used to determine what rules to keep and what rules to forget. Prodigy: 328 possible rules 69 pass test 19 were retained

  14. Problems with EBL • Another problem with EBL is that it is sometimes intractable to create an explanation for the target concept. For example, in chess, learning a concept like: “states for which operator A leads to a solution” The search here grows exponentially.

  15. Summary • Different from inductive learning, analytical learning looks for a hypothesis that fit the background knowledge and covers the training examples. • Explanation based learning is one kind of analytical learning that divides into three steps: • Explain the target value for the current example • Analyze the explanation (generalize) • Refine the hypothesis

  16. Summary • Prolog-EBG constructs intermediate features after analyzing examples. • Explanation based learning can be used to find search control rules. • In all cases we depend on a perfect domain theory.

More Related