CGSC170 Lecture Notes - Lecture 9: Physical Symbol System, Artificial Neural Network
Document Summary
Can a language be represented in an artificial neural network. Aside from general learning algorithm not part of artificial neural networks. Learning a language as information processing model may not necessarily be based on rules. Understand rules by which words are combined. Maybe language users don"t have to represent rules, all they have to do is be able to distinguish applications of words that fit the rule from applications that don"t. Maybe language users represent the rules for combining words. Those two maybes are related but separate. Weak sense language users understand rules but do not explicitly represent them. Strong sense linguistic rules are represented explicitly by language users. How cog scientists understand linguistic rules will impact how they view learning. Can use them to form complex symbol structure. In out set they contain processes for manipulating those symbol structures. A physical symbol system is sufficient for general intelligence.