The videos from Brain Day 2023 are now available on line at our youtube channel. Hope you enjoy.
Blake Richards
Department of Cell & Systems Biology
University of Toronto
Deep Learning with Pyramidal Neurons
Deep learning has led to significant advances in artificial intelligence, in part, by adopting strategies motivated by neurophysiology. However, it is unclear how deep learning could occur in the real brain, due to the difficulty of performing credit assignment without backpropagation. Here, we show that deep learning can be achieved in a biologically feasible simulation by moving away from point neuron models and towards multi-compartment neurons. Like neocortical pyramidal neurons, neurons in our model receive feedforward sensory information and higher-order feedback in electrotonically segregated compartments. Thanks to this segregation, the neurons in different layers of the network can coordinate local synaptic weight updates to achieve global optimization. As a result, the network can take advantage of multilayer architectures---the hallmark of deep learning. This work demonstrates that deep learning can be achieved using segregated dendritic compartments for feedforward and feedback information, which may help to explain the dendritic morphology of neocortical pyramidal neurons.
200 University Avenue West
Waterloo, ON N2L 3G1
Canada