CS1674: Homework 8 - Written

Due: 4/3/2018, 11:59pm

This assignment is worth 10 points.
  1. [3 pts] Why do we say that neural networks are not linear classifiers?
  2. [4 pts] Imagine a vision system predicts the following scores for a cat image: 10 for the category "cat," 5 for the category "dog," and 3 for the category "cow." Another system predicts scores 8, 6 and 1 for the same categories. Is the hinge loss the same or different for these two systems? What about the cross-entropy loss? Why?
  3. [3 pts] Briefly, how does gradient descent work?