Intelligent Systems CSCI 6501 Dr. D. Riordan Pradeep Monga  (B00342080) Satwant Sandhu (B00201045)
<ul><li>Structure for Credit-Apportionment Problem in Rule Based Systems </li></ul>FINAL REPORT
<ul><li>In this Project we have worked with Credit - Apportionment Problem in Rule Based Systems.  </li></ul><ul><li>We ha...
<ul><li>The Credit-Apportionment process provides a formal basis for the problem analysis and algorithm design.  </li></ul...
<ul><li>Credit-Apportionment problem can be formulated as:- </li></ul><ul><li>Local level problem </li></ul><ul><li>Estima...
<ul><li>This System is used for students who are seeking admission into an engineering institute after clearing the entran...
Implementation <ul><li>We have following information </li></ul><ul><li>No. of Seats available in each Branch, </li></ul><u...
Branches and Parameters Table showing sample threshold values of branches in different parameters  X CHAD   = 8 X MEAD   =...
Calculation of Branch Aptitude Total   <ul><li>BAT of each branch for the candidate is calculated as follows, </li></ul><u...
What GAMBLE does? <ul><li>GAMBLE shows BAT of each Branch and suggests the most suitable Branch for the candidate. </li></...
How are classifier strings generated? Rules  1 if input > Avg + Sd/2 0 if input < Avg – Sd/2 $ if Avg – Sd/2 < input < Avg...
More insight.. Performance
Comparison between two classifiers <ul><li>“ Two classifiers do not match if one classifier has 0 at a position and other ...
Learning Process <ul><li>Threshold classifier strings are compared with the classifier strings of all the students that ha...
Artificial Classifier generation <ul><li>In case, there is no match for threshold classifier in the records, the system is...
Conclusions   <ul><li>The learning mechanism is one of a clear candidate for a cognitive invariant in humans which involve...
References <ul><li>[1]  http://www.icce2001.org/cd/pdf/p14/IN002.pdf </li></ul><ul><li>GAMBLE expert system , Credit appor...
Contd.. <ul><li>[5] Goldberg, David E., “ Genetic Algorithms in Search Optimization, and  Machine Learning”, Addison Wesle...
<ul><li>Any Questions are welcome? </li></ul>
Upcoming SlideShare
Loading in …5
×

final report (ppt)

1,240 views

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
1,240
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
12
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

final report (ppt)

  1. 1. Intelligent Systems CSCI 6501 Dr. D. Riordan Pradeep Monga (B00342080) Satwant Sandhu (B00201045)
  2. 2. <ul><li>Structure for Credit-Apportionment Problem in Rule Based Systems </li></ul>FINAL REPORT
  3. 3. <ul><li>In this Project we have worked with Credit - Apportionment Problem in Rule Based Systems. </li></ul><ul><li>We have implemented a hybrid expert system called GAMBLE (Genetic Algorithm Based Machine Learning Expert). </li></ul>Overview
  4. 4. <ul><li>The Credit-Apportionment process provides a formal basis for the problem analysis and algorithm design. </li></ul><ul><li>It includes </li></ul><ul><li>System Environment sub - model which provides integrated </li></ul><ul><li>view about the payoff to, as well as the external and internal </li></ul><ul><li>aspects of rule based system. </li></ul><ul><li>Principles of Usefulness , which define the usefulness of rule </li></ul><ul><li>actions. </li></ul><ul><li>Definitions of the Credit-Apportionment problem which guides </li></ul><ul><li>the algorithm synthesis. </li></ul>Contd…
  5. 5. <ul><li>Credit-Apportionment problem can be formulated as:- </li></ul><ul><li>Local level problem </li></ul><ul><li>Estimation of the inherent usefulness values in a particular context. </li></ul><ul><li>Global level problem </li></ul><ul><li>As approximation to the inherent usefulness functions ( ) from the payoffs. </li></ul>Problem Formulation
  6. 6. <ul><li>This System is used for students who are seeking admission into an engineering institute after clearing the entrance examination. The student is advised by this system as to which branch would be most suited for him, with the help of an algorithm used for branch selection. </li></ul>GAMBLE
  7. 7. Implementation <ul><li>We have following information </li></ul><ul><li>No. of Seats available in each Branch, </li></ul><ul><li>Thresholds of parameters (like logic, memory, aesthetic-sense, adaptability) of each Branch, </li></ul><ul><li>Records of students already admitted (includes scores, averages, standard deductions, classifier strings and their weights. </li></ul><ul><li>On starting execution, system prompts for </li></ul><ul><li>Student Name </li></ul><ul><li>Logic score </li></ul><ul><li>Memory score </li></ul><ul><li>Aesthetic-sense score </li></ul><ul><li>Adaptability score </li></ul>
  8. 8. Branches and Parameters Table showing sample threshold values of branches in different parameters X CHAD = 8 X MEAD = 5 X ELAD = 4 X CSAD = 7 ADAPTABILITY X CHAS = 5 X MEAS = 8 X ELAS = 7 X CSAS = 4 AESTHETIC SENSE X CHMM = 8 X MEMM = 2 X ELMM =10 X CSMM = 3 MEMORY X CHLG = 2 X MELG = 8 X ELLG = 3 X CSLG =10 LOGIC CHEM. MECH. ELECT. COMP.
  9. 9. Calculation of Branch Aptitude Total <ul><li>BAT of each branch for the candidate is calculated as follows, </li></ul><ul><li>BAT = ∑ p = LG,IM… (X BP * S BT ) / X BT </li></ul><ul><li>Where, </li></ul><ul><li>B = Subscript for a particular branch, </li></ul><ul><li>P = Subscript for a particular parameter, </li></ul><ul><li>X BP = Weight of parameter ‘P’ for branch ‘B’, </li></ul><ul><li>S BT = Score obtained by student in a particular parameter, </li></ul><ul><li>X BT = Total of the weighted parameters. </li></ul>
  10. 10. What GAMBLE does? <ul><li>GAMBLE shows BAT of each Branch and suggests the most suitable Branch for the candidate. </li></ul><ul><li>It shows the branches in which student is ineligible to seek admission and prompts to choose among the branches in which the candidate is eligible (i.e. in which he/she has more than minimum marks and seats are available). </li></ul><ul><li>Student is granted admission, if seats available in the chosen branch. </li></ul><ul><li>Following information is updated in the database – </li></ul><ul><ul><li>No. of seats available, </li></ul></ul><ul><ul><li>Record of new student is added, </li></ul></ul><ul><ul><li>and Learning process is started. </li></ul></ul>
  11. 11. How are classifier strings generated? Rules 1 if input > Avg + Sd/2 0 if input < Avg – Sd/2 $ if Avg – Sd/2 < input < Avg # if Avg > input > Avg + Sd/2 In this case the generated string is “1$0$” where 1 is for logic, $ is for memory, 0 is for Aesthetic-sense $ is for Adaptability 0.54 0.65 1.14 0.99 Standard deviation / 2 5.2 2.8 5.0 6.4 Average 5 1 4 10 Input Scores adaptability Aesthetic - sense Memory Logic
  12. 12. More insight.. Performance
  13. 13. Comparison between two classifiers <ul><li>“ Two classifiers do not match if one classifier has 0 at a position and other has 1 at the same position or vice-versa, else in all other cases classifiers match”. </li></ul><ul><li>“‘ $’ or ‘#’ are fuzzy variables and match with any value in other classifier”. </li></ul><ul><li>For e.g. 001$ and ##10 – Match </li></ul><ul><ul><ul><ul><ul><li>1$0# and 1100 – Match </li></ul></ul></ul></ul></ul><ul><ul><ul><ul><ul><li>10$# and 00$# - Unmatched </li></ul></ul></ul></ul></ul>
  14. 14. Learning Process <ul><li>Threshold classifier strings are compared with the classifier strings of all the students that have been admitted in the past. </li></ul><ul><li>The strengths of classifiers that matched with threshold classifier are increased while the strengths of classifiers that didn’t matched are reduced. </li></ul><ul><li>One winning classifier is chosen randomly amongst the classifiers that matched and had comparatively higher strengths. </li></ul><ul><li>Strengths of those classifiers is again increased by some percentage that were equally competent but couldn’t win. They are rewarded so that they have better chance in future. </li></ul><ul><li>The parameter values of winning classifier are ascertained and they are made the new thresholds. </li></ul>
  15. 15. Artificial Classifier generation <ul><li>In case, there is no match for threshold classifier in the records, the system is robust enough to handle the situation by unleashing the power of ‘Genetic Algorithms’. System uses a mechanism that implements the tripartite process of reproduction, crossover and mutation to produce the temporary classifiers. </li></ul><ul><li>Fitness function: </li></ul><ul><li>If the incoming threshold message element is 1 then the corresponding classifier element should not be 0 . </li></ul><ul><li>if the incoming threshold message element is 0 then the corresponding classifier element should not be 1. </li></ul><ul><li>if the incoming threshold message element is 1 then the corresponding parameter value is AVG + δ/8. </li></ul><ul><li>if the incoming threshold message element is 0 then the corresponding parameter value is AVG - δ/8. </li></ul><ul><li>In all other cases the parameter value is AVG. </li></ul>
  16. 16. Conclusions <ul><li>The learning mechanism is one of a clear candidate for a cognitive invariant in humans which involves the ability to acquire facts, skills and more abstract concepts. </li></ul><ul><li>human learning aspects can be reproduced in a computer system by understanding the criteria by means of which humans learn. </li></ul><ul><li>In the coming days and also in present situations, learning would tend to be more efficient than programming. </li></ul><ul><li>An important aspect of student education has been covered in this report and the field is still open to make the system handle the effect of various other changes in the environment and its response towards them. </li></ul>
  17. 17. References <ul><li>[1] http://www.icce2001.org/cd/pdf/p14/IN002.pdf </li></ul><ul><li>GAMBLE expert system , Credit apportionment process and Bucket Brigade Algorithm Indian Institute of Technology, Roorkee </li></ul><ul><li>India 2001. </li></ul><ul><li>[2] IEEE Transactions On Systems And Cybernatics, “Framework for the Credit-Apportionment Process in Rule-Based Systems” Vol 19, No 3, May/June 1989. </li></ul><ul><li>[3] “Bucket Brigade Performance: 1 long sequences of classifiers” in Genetic algorithms and their application: proc 2nd int. conf on GA. </li></ul><ul><li>J.Grefenstette, Ed. July 1987. </li></ul><ul><li>[4] A study on apportionment of credits of fuzzy classifier system for knowledge acquisition of large scale systems Nakaoka, K.; Furuhashi, T.; Uchikawa, Y.; Fuzzy Systems, 1994. IEEE World Congress on Computational Intelligence, Proceedings of the Third IEEE Conference on, 26-29 June 1994 </li></ul>
  18. 18. Contd.. <ul><li>[5] Goldberg, David E., “ Genetic Algorithms in Search Optimization, and Machine Learning”, Addison Wesley Longman, International Student </li></ul><ul><li>Edition 1999. </li></ul><ul><li>[6] J. Holland. Escaping brittleness: the possibilities of general purpose </li></ul><ul><li>learning algorithms applied to parallel ruled based systems. In </li></ul><ul><li>R. Michalski, J Corbonell and T Mitchell, editors, Machine learning: </li></ul><ul><li>An Artificial intelligence approach, Morgan Kauffmann Publishers, Inc. </li></ul><ul><li>Los Altos, CA., 1986 </li></ul>
  19. 19. <ul><li>Any Questions are welcome? </li></ul>

×