COGS 100 Study Guide - Final Guide: Connectionism, Spreading Activation, Semantic Network
16 Dec 2017
School
Department
Course
Professor
Document Summary
Information is stored as nodes on a graph. Links between nodes show dependencies of info pieces and how it affects overall behavior. Used to model the distributed nature of mental processes that occur through the firing of neurons. In ai, this representation is called artificial neural networking (ann) Key differences from other representations: execution is parallel, nodes contain limited information, can be organized into different layers. Signals coming from excitatory links act positively towards generating effect: inhibitory. Signals coming from inhibitory links act negatively towards generating effect. Activation of a node is affected by the signals from the connected nodes and the types of links. Two types of connectionist representations: local representation. One way excitatory connections make activation flow from observed behavior to inferred traits. Easy to tell how each node value affects the linked node. Reflects how it"s hard for both to be true: distributed representation. Networks use training data to strengthen excitatory links.
