riley16 / hebbian_pretraining Goto Github PK
View Code? Open in Web Editor NEWCourse project with Jay McClelland on Hebbian pre-training for error-driven learning. Includes attempts at reproducing the works of some authors, such as Lillicrap, T. P. et al. Random synaptic feedback weights support error backpropagation for deep learning. Nat. Commun. 7, 13276 doi: 10.1038/ncomms13276 (2016). Also includes a simple neural network framework for personal experimentation.