Input Feedback Networks: Classification and Inference Based on Network Structure

摘要: We present a mathematical model of interacting neuron-like units that we call Input Feedback Networks (IFN). Our is motivated by new approach to biological neural networks, which contrasts with current approaches (e.g. Layered Neural Networks, Perceptron etc.). Classification reasoning in IFN are accomplished an iterative algorithm, and learning changes only structure. Feature relevance determined during classification. Thus it emphasizes network structure over edge weights. IFNs more flexible than previous approaches. In particular, integration node can affect the outcome existing nodes without modifying their prior produce informative responses partial inputs or when networks extended other tasks. It also enables recognition complex entities images) from parts. This promising for future contributions integrated human-level intelligent applications due its flexibility, dynamics structural similarity natural neuronal networks.