Machine Learning, Fall 2008, Homework 2

Problem Descriptions

• HW2.pdf

Data Sets

• hw2_3_1_train.dat
• hw2_3_1_test.dat
• hw2_3_2_train.dat
• hw2_3_2_test.dat
• hw2_4_train.dat
• hw2_4_test.dat

Information

• guidelines about submission and source code uploading: here
• CEIBA:
  • login: http://ceiba.ntu.edu.tw
  • discussion: http://tinyurl.com/ml08fall-forum-english
  • source uploading: http://tinyurl.com/ml08fall-hw or http://tinyurl.com/ml08fall-hw-ch
• TA session: 10/15/2008, noon to 2pm, R106
• other help from TAs: ml08fall AT csie DOT ntu DOT edu DOT tw

FAQ

• What does "nabla^2 E(u, v)" mean in Problem 2.2 (4)?
  The Hessian matrix. We clarified that in the latest version of HW2.pdf.
• Is "nabla E(u, v)" a row vector or a column one?
  A column one. We clarified that in the latest version of HW2.pdf.
• What do you want for Problem 2.2 (2) and (4)?
  A mathematical proof. You only need to prove it for the case of (u, v) = (0, 0).
• What do you want for Problem 2.2 (5)?
  Something like: (Delta u, Delta v) = (0.3, 0.4), and the resulting E = 2.1729 .
• What are the (u, v) values for Problem 2.2 (5)?
  Please use (0, 0).
• What is the base of "log" in Problem 2.4?
  We suggest (and intend to ask) you to use log_e. But log_10 is also fine.
• The data sets of Problem 2.3 (1) and 2.4 are identical. Is that right?
  Yes. The data sets are actually the same as the ones for Problem 1.4

Last updated at CST 09:10, October 15, 2008 Please feel free to contact me: