Learning without feedback: Fixed random learning signals allow for feedforward training of deep neural networks
This addresses the problem of enabling low-cost adaptive smart sensors at the edge by offering a biologically plausible and computationally efficient alternative to backpropagation, though it is incremental as it builds on existing feedforward training concepts.
The paper tackled the biological implausibility and computational inefficiency of backpropagation in deep neural networks by proposing a feedforward training method using fixed random projections of labels, which reduces memory and processing requirements. The result was the DRTP algorithm, achieving a tradeoff suitable for edge computing devices, with competitive accuracy on benchmarks like CIFAR-10 and ImageNet.
While the backpropagation of error algorithm enables deep neural network training, it implies (i) bidirectional synaptic weight transport and (ii) update locking until the forward and backward passes are completed. Not only do these constraints preclude biological plausibility, but they also hinder the development of low-cost adaptive smart sensors at the edge, as they severely constrain memory accesses and entail buffering overhead. In this work, we show that the one-hot-encoded labels provided in supervised classification problems, denoted as targets, can be viewed as a proxy for the error sign. Therefore, their fixed random projections enable a layerwise feedforward training of the hidden layers, thus solving the weight transport and update locking problems while relaxing the computational and memory requirements. Based on these observations, we propose the direct random target projection (DRTP) algorithm and demonstrate that it provides a tradeoff between accuracy and computational cost that is suitable for adaptive edge computing devices.