Loading...
Please wait, while we are loading the content...
Similar Documents
Using Features for the Storage of Patterns in a Fully Connected Net (1995)
| Content Provider | CiteSeerX |
|---|---|
| Author | Coombes, S. Taylor, J.g. |
| Abstract | . One of the many possible conditions for pattern storage in a Hopfield net is to demand that the local field vector be a pattern reconstruction. We use this criterion to derive a set of weights for the storage of correlated biased patterns in a fully connected net. The connections are built from the eigenvectors or principal componentsof the pattern correlation matrix. Since these are often identified with the features of a pattern set we have named this particular set of weights as the Feature Matrix. We present simulation results that show the Feature Matrix to be capable of storing up to N random patterns in a network of N spins. Basins of attraction are also investigated via simulation and we compare them with both our theoretical analysis and those of the Pseudo-Inverse rule. A statistical mechanical investigation using the replica trick confirms the result for storage capacity. Finally we discuss a biologically plausible learning rule capable of realising the Feature Matrix in ... |
| File Format | |
| Language | English |
| Publisher Date | 1995-01-01 |
| Access Restriction | Open |
| Subject Keyword | Feature Matrix Fully Connected Net Statistical Mechanical Investigation Pattern Reconstruction Replica Trick Pseudo-inverse Rule Principal Componentsof Biased Pattern Storage Capacity Pattern Correlation Matrix Local Field Vector Hopfield Net Pattern Storage Theoretical Analysis Particular Set Present Simulation Result Random Pattern Many Possible Condition |
| Content Type | Text |
| Resource Type | Technical Report |
