COG250Y1 Lecture Notes - Lecture 14: Perceptron, Neurogenesis, Unsupervised Learning
11 Apr 2017
School
Department
Course
Professor
Document Summary
The idea was that you"d give up the variables proposition. Instead, we"ll use variable that are more brain-like . We can look at neurons" level of activity they can alter the signal strength, by altering how frequently they fire. We can look at excitation vs inhibition (signal valence) The weighting is a variable = how much influence one var has one another. Two problems/issues they need to solve and face: the criteria/problems descartes had, need to convince hobbs that computation isn"t needed for strong ai. Typically, people will have 500 or 600 nodes in the nn which can subsequently play a game, speak. But that would mean the system would only have 600 neurons which doesn"t make sense (even bees have > 1 million neurons). The webware is missing here, which is present in our brain: the neurochemicals which. Endocrine chemicals that diffuse and impact your performance.
