Implementing engrams from a machine learning perspective: the relevance of a latent space

J Marco de Lucas

In our previous work, we proposed that engrams in the brain could be biologically implemented as autoencoders over recurrent neural networks. These autoencoders would comprise basi…