The stability analysis for a novel feedback neural network with partial connection

摘要: Abstract This paper develops a new Partially Feedback Neural Network with partial connection, which is so-called “Partially Connected Network” (PCFNN). The information capacity improves and there more hidden for partially connected systems because the connections between neurons are random can be than one layer. proving of convergence PCFNN provided. Owing to complexities in systems, two theorems its stability proved theoretically by constructing novel energy function expectation. Three examples provided simulate various conditions different activation functions weight matrixes. simulation results show that this neural network stable under conditions. expressive space architecture also much larger original Hopfield architecture.