# CSCI 6962-   Soft Computing   -   Fall 2001

Tuesday 6:00-9:00pm, Ricketts 212

 Instructors: Bill Cheetham, Kai Goebel Office: Ricketts 212 Office Hours: TBD Phone: 387-5222 (Bill),  387-4194 (Kai) Email: cheetham@cs.rpi.edu, goebel@cs.rpi.edu

### Calendar

The viewgraphs are in pdf format. You can use ghostview or adobe acrobat reader to view the pdf.

Note: viewgraphs are linked for past and current lectures/homework only

 DATE TOPIC READING OUT/DUE 8/28 Course Description - Kai Introduction: What is SC? Why is it useful? - Bill, Matlab tutorial(here also as postscript) - Kai  Fuzzy Sets; - Kai (slides in 6 slide/page format: CD, Intro, Matlab, FS,) Jang: Forward, Preface Jang: 1 HW 1, data out 9/4 Fuzzy Reasoning; Fuzzy Inference - Kai (6slide/page format: FR, FI) Jang: 2, 3, 4 HW 2 out Due: HW 1 9/11 Class Cancelled 9/18 Gradient descent optimization: least squares methods - Bill  Genetic Algorithms -Bill in-class exercise: class acts as GA - Bill Applications of GA's - Class (6slides/page format: Gradient, GA, CryptoQuote) Jang 5, 7.2 Due: HW 2 HW 3 (with iris2D.dat), out 9/25 Fuzzy Applications - Kai Applications of Fuzzy Logic  - Class (6slides/page format: DT, FC, FA, FIP, FDF, FD, IF) Jang: 14, 15, 16 Lecture notes HW 4, out Due: HW 3 10/2 Fuzzy Image Processing; Fuzzy Data Fusion; Fuzzy Diagnosis; Neural Networks: Supervised Learning: Hopfield Nets, Perceptrons, gradient descent, multilayer nets, backpropagation, overfitting - Kai Applications of Neural Networks – Class (6 slides/page format: NN). Jang: 9 HW 5 out; Due: HW 4 10/9 NO CLASS - 10/16 Applications of Neural Networks – Class Jang: 9, 10, 11, 15 HW 6 out (iris4D.dat) Due: HW 5 10/23 Case-Based Reasoning: nearest neighbor, explanation-based learning, case selection, case adaptation -Bill Applications of CBR: (Watson) - Bill Automated Collaborative Filtering - Bill (6 slides/page format:  CBR, CBR_Apps) Watson: 1, 2, 3, 4.1 HW 7 out Due: HW 6 10/30 Applications of CBR: Color Matching - Bill  in-class exercise: determining a salary for baseball players -Bill  (6 slides/page format: Color) Watson: 5,6,7,8 HW 8 out test_houses.dat) Due: HW 7 Including Project Plan 11/6 Applications of CBR: APVT, C4.5 - Bill Applications of CBR - Class CBR for Help Desks, ELSI – Bill (6 slides/page format: APVT, Help Desks) HW 9 out Due: HW 8 11/13 Hybrid Systems: ANFIS, Fuzzy Filtered NN & Neural Fuzzy Systems, GA tuned Fuzzy System, Adaptive Fuzzy Clustering, etc., Summary Remarks - Kai (6 slides/page format: ANFIS, NN, HS, SR ). Handouts HW 10 out Due: HW 9 11/20 BBN - demo - Bill Determining when to use a technique - Bill Knowledge Management - Bill  (6 slides/page format: BBN) HW 11 out Due: HW 10 11/27 one-on-one project discussions Due: HW 11 Due: Project Update 12/4 Presentations & Pizza Due: Final Project

### Policies

• Assignments are due at the beginning of class.
• The syllabus and all accompanying dates are subject to change at any time; all such changes will be announced in class and updated on this web page.
• Cooperative efforts at understanding the material conceptually are encouraged. However, all written work and programs you turn in must be your own. You many not share solutions. Submitting work that is not your own is considered cheating, as per RPI policy.
• If you can't come for scheduled office hours, please send us email; we would be happy to schedule a meeting at some other mutually agreed upon time.

August 27, 2001
goebel@cs.rpi.edu; cheetham@cs.rpi.edu