A biologically plausible neural network for local supervision in cortical microcircuits
This work addresses the problem of biological implausibility of backpropagation for modeling brain function, which is a foundational problem for computational neuroscience and biologically-inspired AI.
This paper proposes a biologically plausible neural network algorithm that avoids the weight sharing requirement of backpropagation. The algorithm performs comparably to backpropagation on several datasets.
The backpropagation algorithm is an invaluable tool for training artificial neural networks; however, because of a weight sharing requirement, it does not provide a plausible model of brain function. Here, in the context of a two-layer network, we derive an algorithm for training a neural network which avoids this problem by not requiring explicit error computation and backpropagation. Furthermore, our algorithm maps onto a neural network that bears a remarkable resemblance to the connectivity structure and learning rules of the cortex. We find that our algorithm empirically performs comparably to backprop on a number of datasets.