Research on Forward Neural Network for Corner Classification Based on Real Vectors

Instant classification is a hot area for researchers on online information retrieval and forward neural network is an important neural network for corner classification. This paper presents a new forward neural network consisting of two kinds of neurons. The topology, learning algorithm and behaviors of the network are given. Theoretical analysis shows that the time complexity of the network learning is linear. Experimental results indicate that ,compared with those consisting of binary neurons, the network can improve classification precision remarkably while satisfying the time requirement satisfied consisting of binary neurons.