Supervised learning in neural networks without feedback network
Robert Brandt, Feng Lin · 2002
In this paper, we study the supervised learning in neural networks. Unlike the common practice of backpropagating error feedback by a separate feedback network that must have the same topology and connection strengths as the feedforward network, we propose a new adaptation algorithm by which the same supervised learning as accomplished by the backpropagation algorithm can be achieved without using a separate feedback network. The elimination of the feedback network makes it more likely for the neural systems to achieve the same adaptation by means of some retrograde regulatory mechanisms that may exist in biological neural systems. Other advantages of this new algorithm include: (1) it allows a phaseless adaptation by neurons; and (2) it simplifies (hardware) implementation of artificial neural networks.