1. Planning Chapter 7 article 7.4 Production Systems Chapter 5 article 5.3 RBSChapter 7 article 7.2

2. 2 RBS CS 331/531 Dr M M Awais Expert System: A SYSTEM that mimics a human expert Human experts always have in most case some vague (not very precise) ideas about the associations Handling uncertainties is a essential part of an expert system Expert systems are RBS with some level of uncertainty incorporated in the system

3. 3 RBS CS 331/531 Dr M M Awais Choosing a Problem Costs: Choose problems that justify the development cost of the expert systems Technical Problems: Choose a problem that is highly technical in nature problems with Well-formed knowledge are the best choice. Highly specialized expert requirements, solution time for the problem is not short time. Cooperation from an expert: Experts are willingly to participate in the activity.

4. 4 RBS CS 331/531 Dr M M Awais Choosing a Problem Problems that are not suitable Problems for which experts are not available at all, or they are not willingly to participate Problems in which high common sense knowledge is involved Problems which involve high physical skills

5. 5 RBS CS 331/531 Dr M M Awais ES Architecture interface user Explanation system Inference engine Knowledge Base editor Case specific Data Knowledge Base Expert System Shell

6. 6 RBS CS 331/531 Dr M M Awais ES Architecture interface user Explanation system Inference engine Knowledge Base editor Case specific Data Knowledge Base Expert System Shell Uses Menus, NLP, etc… Which is used to interact With the users

7. 7 RBS CS 331/531 Dr M M Awais ES Architecture interface user Explanation system Inference engine Knowledge Base editor Case specific Data Knowledge Base Expert System Shell Explains why a decision is taken, uses keywords Such as HOW, WHY etc… Implements the reasoning methods Generally backward chaining Updates the KB

8. 8 RBS CS 331/531 Dr M M Awais ES Architecture interface user Explanation system Inference engine Knowledge Base editor Case specific Data Knowledge Base Expert System Shell Pre-solved problems, Frequently referred cases Collection of facts And rules

9. 9 RBS CS 331/531 Dr M M Awais Shells General purpose toolkit/shell is problem independent Shells commercially available CLIPS is one such shell Freely available

10. 10 RBS CS 331/531 Dr M M Awais Reasoning with Uncertainty Case Studies: MYCIN Implements certainty factors approach INTERNIST: Modeling Human Problem Solving Implements Probability approach

11. 11 RBS CS 331/531 Dr M M Awais RBS: Handling Uncertainties How to handle vague concepts? Why vagueness occurs? All rules are not 100% deterministic Certain rules are often true but not always Headache may be caused in flu, but may not always occur An expert may not always be sure about certain relations and associations

12. 12 RBS CS 331/531 Dr M M Awais First Source of Uncertainty: The Representation Language  Possible States are large  Single representation may correspond to multiple states, which the agent can’t represent distinguishably  Languages are generally less expressive A BC A BC A BC on(A,B)  on(B,Table)  on(C,Table)  clear(A)  clear(C)

13. 13 RBS CS 331/531 Dr M M Awais Second source of Uncertainty: Imperfect Observation of the World Observation of the world can be:  Partial, e.g., a vision sensor can’t see through obstacles (lack of percepts) R1R1 R2R2 The robot may not know whether there is dust in room R2

14. 14 RBS CS 331/531 Dr M M Awais Second source of Uncertainty: Imperfect Observation of the World Observation of the world can be:  Partial, e.g., a vision sensor can’t see through obstacles  Ambiguous, e.g., percepts have multiple possible interpretations A B C on(a,b)  on(a,c)

15. 15 RBS CS 331/531 Dr M M Awais Second source of Uncertainty: Imperfect Observation of the World Observation of the world can be:  Partial, e.g., a vision sensor can’t see through obstacles  Ambiguous, e.g., percepts have multiple possible interpretations  Incorrect

16. 16 RBS CS 331/531 Dr M M Awais Third Source of Uncertainty: Ignorance, Laziness, Efficiency Laziness/Efficiency:  An action may have a long list of preconditions, e.g.:  Drive-Car: have(keys)   empty(gas-tank)  battery-Ok  ignition-Ok   flat-Tires   stolen(Car)...  Medical Treatment symptoms(p,toothache)  disease(p, cavity)  The writer may not list all the condition  Results in incorrect representation or several interpretations

17. 17 RBS CS 331/531 Dr M M Awais Third Source of Uncertainty: Ignorance, Laziness, Efficiency Ignorance:  Theoretical: The domain knowledge in itself may not be complete. The domain knowledge may not have a complete theory e.g. many instances in Medical science are unexplainable  Practical Ignorance: The domain knowledge is complete but the implementing it in an real/artificial environment may be difficult. e.g., some tests may yield poor results due to low instrument precision

18. 18 RBS CS 331/531 Dr M M Awais Modelling Uncertainty  Non-deterministic model: Uncertainty is represented by a set of possible values, e.g., a set of possible worlds, a set of possible effects,  Probabilistic model: Uncertainty is represented by a probabilistic distribution over a set of possible values  Case specific models: Certainty factors used in MYCIN  Fuzzy models

19. 19 RBS CS 331/531 Dr M M Awais Example:  Non-deterministic: list all possible states, belief state represents all the states of the world that are possible at a given time or at a given stage of reasoning  Probabilistic: probability is attached to each state to measure its likelihood to be the actual state 0.20.30.40.1

20. 20 RBS CS 331/531 Dr M M Awais Probabilities ?  Probabilities: frequency interpretation  Relative occurrence of a particular state defined by the probabilistic distribution 0.20.30.40.1 This state would occur 20% of the times

21. 21 RBS CS 331/531 Dr M M Awais Example  We have a Dentist D who meets a new patient  D is interested in only one thing: whether Patient has a cavity, which D models using the proposition Cavity  Before making any observation, D’s belief state is:  This means that if D believes that a fraction p of patients have cavities Cavity  Cavity p 1-p

22. 22 RBS CS 331/531 Dr M M Awais Example  Now an observation is made ‘toothache’  D’s belief state wrt toothache is:  Can the observation and the effect be related  i.e. cavity and toothache (YES), How? toothache  toothache p 1-p

23. 23 RBS CS 331/531 Dr M M Awais Example  Lets relate the two: The patient has a cavity if he / she has toothache  This sentence suffers from laziness and/or ignorance.  Why? It may not be necessary that every patient that suffers from toothache may also have cavity. Then in order to capture real situation we may keep on increasing the reasons of toothache I.e, The patient has cavity or gum problem or … if he /she suffers from toothache.  Is there a simple way to solve this problem. YES  Attach probability to initial rule and that would summarize the uncertainty caused because of laziness and ignorance

24. 24 RBS CS 331/531 Dr M M Awais Example  Lets relate the two: The patient has a cavity if he / she has toothache  Probability of 0.7 (70% chance)  70% summarizes:  Cases in which all the factors needed for cavity to cause toothache are present  And cases in which the patient has both cavity and toothache but the two are unconnected  30% summarizes  all the other possible causes of toothache that we are too lazy/ignorant to confirm or deny

25. 25 RBS CS 331/531 Dr M M Awais Making decisions under uncertainty P(A1 gets me to goal | …) = 0.04 P(A2 gets me to goal | …) = 0.70 P(A3 gets me to goal | …) = 0.95 P(A4 gets me to goal | …) = 0.99 Which action to choose? Depends on my preferences for missing flight vs. time spent waiting, etc. Probability theory: summarizes uncertainties Utility theory: represents and infers preferences Decision theory = probability theory + utility theory

26. 26 RBS CS 331/531 Dr M M Awais Probability Degree of believe in a fact ‘x’, P(x) P(H): degree of believe in H, when supporting evidence is NOT given, H is the hypothesis Joint Probabilities P(H|E): degree of believe in H, when supporting evidence is given, E is the evidence supporting hypothesis P(H|E): conditional probability

27. 27 RBS CS 331/531 Dr M M Awais Prior probability Prior or unconditional probabilities of propositions e.g., P(Cavity = true) = 0.1 and P(Weather = sunny) = 0.72 Probability distribution gives values for all possible assignments: P(Weather) = (normalized, i.e., sums to 1) Joint probability distribution for a set of random variables gives the probability of every atomic event on those random variables P(Weather,Cavity) = 4 × 2 matrix of values: Weather =sunnyrainycloudysnow Cavity = true 0.1440.02 0.016 0.02 Cavity = false0.5760.08 0.064 0.08 Question about a domain can be answered by the joint distribution

28. 28 RBS CS 331/531 Dr M M Awais Conditional Probability P(H|E): conditional probability is given through a LAW (RULE)P(H|E)=P(H^E)/P(E) where P(H^E) is the probability of H and E occurring together (both are TRUE): joint

29. 29 RBS CS 331/531 Dr M M Awais Inference: Joint Prob.

30. 30 RBS CS 331/531 Dr M M Awais Reasoning P(toothache) = 0.108 + 0.012 + 0.016 + 0.064 = 0.2

31. 31 RBS CS 331/531 Dr M M Awais Reasoning Can also compute conditional probabilities: P(  cavity | toothache) = P(  cavity  toothache) P(toothache) = 0.016+0.064 0.108 + 0.012 + 0.016 + 0.064 = 0.4

32. 32 RBS CS 331/531 Dr M M Awais Without Joint distributions Conditional Probabilities are found, Single evidence: Simple If multiple evidences are available then BAYESIAN Updating is done through the use of conditional independence Find the conditional probabilities directly through the chain rule

33. 33 RBS CS 331/531 Dr M M Awais Evaluating: Conditional Probability P(H|E): P(Heart Attack|shooting arm pain) Two approaches can be adopted: Asking an expert about the frequency of it happening Finding the probability from the given data Second Approach Collect the data for all the patients complaining about the shooting arm pain. Find the proportion of the patients that had an heart attack from the data collected in step 1

34. 34 RBS CS 331/531 Dr M M Awais Evaluating: Conditional Probability P(H|E): P(Heart Attack|shooting arm pain) Probability of Disease given symptoms VS P(E|H): P(shooting arm pain|Heart Attack) Probability of symptoms given Disease Which is easier to find of the two? Perhaps P(E|H) is easier

35. 35 RBS CS 331/531 Dr M M Awais Evaluating: Conditional Probability P(H|E): P(Heart Attack|shooting arm pain) Probability of Disease given symptoms Headache is mostly experienced when a patient suffers from flu, fever, tumor, etc… Find out whether a patient suffers from tumor, given headache Collect the data for all the headache patients, and then find the proportion of patients that have tumor.

36. 36 RBS CS 331/531 Dr M M Awais Evaluating: Conditional Probability P(E|H): P(shooting arm pain|Heart Attack) Probability of symptoms given Disease Headache is mostly experienced when a patient suffers from flu, fever, tumor, etc… Find out whether a tumor patient suffers from headache Collect the data for all the tumor patients, and then find the proportion of patients that have headache

37. 37 RBS CS 331/531 Dr M M Awais Evaluating: Conditional Probability Generally speaking P(E|H): P(shooting arm pain|Heart Attack) is easier to find. Therefore the we need to convert P(H|E) in terms of P(E|H) P(H|E)=P(H^E)/P(E)P(H|E)=[P(E|H)*P(H)]/P(E)

38. 38 RBS CS 331/531 Dr M M Awais Evaluating: Conditional Probability More than one evidence Independence of events P(H|E1^E2)=P(H^E1^E2)/P(E1^E2) P(H|E1^E2)=[P(E1|H)* P(E2|H)* P(H)]/(P(E1)*P(E2))

39. 39 RBS CS 331/531 Dr M M Awais Other Approaches

40. 40 RBS CS 331/531 Dr M M Awais Certainty Factors (CF)-MYCIN CF for rules CF(R) From the experts CF for Pre-conditions CF(PC) From the end user CF for conclusions CF(cl) CF(cl)=CF(R)*CF(PC)

41. 41 RBS CS 331/531 Dr M M Awais Certainty Factors (CF) CF for rules CF(R) IF A then BCF(R) = 0.6 CF for Pre-conditions CF(PC) IF A (0.4) then BCF(A)= 0.4 CF for conclusions CF(cl) CF(B)=CF(R)*CF(A)= 0.6*0.4=0.24

42. 42 RBS CS 331/531 Dr M M Awais Finding Overall CF for PC If A(0.1) and B(0.4) and C(0.5) Then D Overall CF(PC)=min(CF(A),CF(B),CF(C)) CF(PC)=0.1 If A(0.1) or B(0.4) or C(0.5) Then D Overall CF(PC)=max(CF(A),CF(B),CF(C)) CF(PC)=0.5

43. 43 RBS CS 331/531 Dr M M Awais CFs for same conclusion rules When the conclusions are same and certainty factors are positive: CF(R1)+CF(R2) – CF(R1)*CF(R2) When the conclusions are same and the certainty factors are both negative CF(R1)+CF(R2) + CF(R1)*CF(R2) Otherwise: both conclusions are same but have different signs [CF(R1)+CF(R2)] / [1 – min ( | CF(R1) |, | CF(R1) |]

44. 44 RBS CS 331/531 Dr M M Awais Example Please see the class handouts

45. 45 RBS CS 331/531 Dr M M Awais