BCS 111 Lecture Notes - Lecture 18: Perceptron, Linear Regression, Eye Tracking
17 Jul 2018
School
Department
Course
Professor
Document Summary
Takes input if input is enough to push some critical threshold, then it fires. Perceptron is mathematical, and is the computer version of this. Cell body will be pushed past threshold and send signal to other neurons or it wont (all or nothing) linear weights some signals are multiplied by weights. Each value of gray may have a diff strength associated with it. Weight is = to 1, not (o. 1 on the slide) If threshold is about 0, then fire (this is a hypothetical instruction) This particular one likes the pattern where darker values are on the outside. Low weights on the outside do not let it fire, value becomes less than 0. This is a particular cell with different weights. You multiple weight by the red thin line. The input times weight (red line) gives you the grand sum. Center value is multiplied by green weight (weight =1) and is added to the whole thing.
