Biological Inspiration for Artificial Neural Networks
description
Transcript of Biological Inspiration for Artificial Neural Networks
Biological Inspiration for Biological Inspiration for Artificial Neural NetworksArtificial Neural Networks
Nick MascolaNick Mascola
Artificial NeuronArtificial Neuron
Output=f(Σ(Weights*Inputs))
Basic Structure
Several Layered NetworkSeveral Layered Network
A Typical Network Organizes these Neurons into layers that feed into each other sequentially
Typical Transfer Typical Transfer FunctionsFunctions
Recall that Over Time:Recall that Over Time:
Finite Amount of ResourcesFinite Amount of Resources
ImplementationImplementation void distributeweightpoints(Connections con){void distributeweightpoints(Connections con){ vector<Weight> list = con.weights;vector<Weight> list = con.weights; int totalpoints=con.points;int totalpoints=con.points; double total=weightsummation(list);double total=weightsummation(list); double temp;double temp; for(unsigned int i=0; i<con.weights.size(); i++){for(unsigned int i=0; i<con.weights.size(); i++){ temp=list[i].value/total;temp=list[i].value/total; if(temp<1/totalpoints){if(temp<1/totalpoints){ con.weights[i]=0;}con.weights[i]=0;} else{else{ con.weights[i]=temp;}con.weights[i]=temp;} }} }}
Long Term PotentiationLong Term Potentiation
SpecificityCooperativity
Features Similar to ANN Functionality:
Distinct FeatureDistinct Feature
Associativity
Possible SolutionPossible Solution
……Or More GenerallyOr More Generally
ReferencesReferences
http://hagan.ecen.ceat.okstate.edu/nnd.htmlhttp://hagan.ecen.ceat.okstate.edu/nnd.html Matlab Neural Network ToolboxMatlab Neural Network Toolbox Pattern ClassificationPattern Classification (2nd ed) by Richard O. (2nd ed) by Richard O.
Duda, Peter E. Hart and David G. StorkDuda, Peter E. Hart and David G. Stork Pattern Recognition and Machine Learning. Pattern Recognition and Machine Learning.
Christopher M. BishopChristopher M. Bishop The long-term potential of LTP Robert C.Malenka