fa13-cs188-lecture-22-1PP

Similarity only linear or can they lets nd out

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ch category Best ­fit to image using min variance Cost for high distor)on of template Cost for image points being far from distorted template   Used in many commercial digit recognizers Examples from [Has)e 94] A Tale of Two Approaches…   Nearest neighbor ­like approaches   Can use fancy similarity func)ons   Don’t actually get to do explicit learning   Perceptron ­like approaches   Explicit training to reduce empirical error   Can’t use fancy similarity, only linear   Or can they? Let’s find out! Kerneliza)on Perceptron Weights   What is the final value of a weight wy of a perceptron?   Can it be any real vector?   No! It’s built by adding up inputs.   Can reconstruct weight vectors (the primal representa)on) from update counts (the dual representa)on) Dual Perceptron   How to classify a new example x?   If someone tells us the value of K for each pair of examples, never need to build the...
View Full Document

This note was uploaded on 12/22/2013 for the course CS 188 taught by Professor Staff during the Fall '08 term at Berkeley.

Ask a homework question - tutors are online