Abstract This application discloses a system that optimizes a neural network by generating all of the discrete weights for a given neural node by creating a normalized weight vector for each possible weight combination. The normalized vectors for each node define the weight space for that node. This complete set of weight vectors for each node is searched using a direct search method during the learning phase to optimize the network. The search evaluates a node cost function to determine a base point from which a pattern more within the weight space is made. Around the pattern mode point exploratory moves are made which are cost function evaluated. The pattern move is performed by eliminating from the search vectors with lower commonality.
An optimizing adaptive reasoning system, comprising: a discrete point digital neural network having digital neural nodes; and digital training menas for determinign discrete points in a weight space for each digital neural node, applying a training pattern, searching the discrete points for a rule as the training pattern is applied and setting the rule in the corresponding digital neural node when a match occurs.
Claim 2
Dependent Claim
64 words
An optimizing adaptive reasoning system, comprising: a discrete point digital neural network having digital neural nodes; and digital training menas for determinign discrete points in a weight space for each digital neural node, applying a training pattern, searching the discrete points for a rule as the training pattern is applied and setting the rule in the corresponding digital neural node when a match occurs.
Claim 3
Dependent Claim
39 words
An optimized neural network, comprising: a discrete point digital neural network having digital neural nodes with discrete point rules; and digital rule means for obtianing the discrete point rules by a direct search of a discrete point rule space.
Claim 4
Dependent Claim
39 words
An optimized neural network, comprising: a discrete point digital neural network having digital neural nodes with discrete point rules; and digital rule means for obtianing the discrete point rules by a direct search of a discrete point rule space.