Something

Just another WordPress site

About

This is currently my personal site with some random stuff, maybe it will become a portfolio site when I have enough skills in website programming, I hope.

Not any solutions that I have posted here are correct, they may contain my naive guess and even my fallacies. Take them at your own risk. Anyway, various posts here serve as my personal notes rather than complete articles.

8 Comments

  • Jacob Dasuqi

    November 26, 2018 at 9:48 pm → Reply

    Hi Vy, I was wondering if you have any solutions / pointers with exercise 8.13 from Learning From Data?
    The question is as follows:
    KT complementary slackness gives that if αn*>0, then (xn,yn) is on the boundary of the optimal fat-hyperplane and yn(w*Txn+b*) = 1. Show that the reverse is not true. Namely, it is possible that αn*= 0 and yet (xn,yn) is on the boundary satisfying yn(w*Txn+b*) = 1.
    Hint: Consider a toy data set with two positive examples at (0,0) and (1,0), and one negative example at (0,1).

    • Vy Nguyen

      January 20, 2019 at 11:00 am → Reply

      Sorry, it’s been a long time since my last touch on SVM so I don’t remember much. And I don’t think I had the solution for Exercise 8.13.

  • Jeff V

    December 2, 2018 at 3:37 am → Reply

    Hey, was just looking at your Prolog program for the Jealous Husbands problem. Very helpful; thanks so much!!
    I am trying to train my machine using your program to find simpler and simpler solutions (fewer and fewer crosses of the river). As I understand it, the simplest solutions have 11 steps, whereas the results I am currently getting are in the 20’s.
    Any idea how I might query this or modify your program to achieve this?
    Thanks again!!!

Leave a Comment

Your email address will not be published. Required fields are marked *