Multilayer feedforward networks with adaptive spline activation function

in this paper, a new adaptive spline activation function neural network (AS NN) is presented. Due to the ASNN's high representation capabilities, netwo rks with a small number of interconnections can be trained to solve both pa ttern recognition and data processing real-time problems. The main idea is to use a Catmull-Rom cubic spline as the neuron's activation function, whic h ensures a simple structure suitable for both software and hardware implem entation. Experimental results demonstrate improvements in terms of general ization capability and of learning speed in both pattern recognition and da ta processing tasks.