We will discuss how we might design a computer system to implement memory representations (engrams) using neural networks, with the main aim of exploring new ideas using machine learning techniques, guided by challenges in neuroscience. We will analyze the use of latent neural spaces as indexes for storing and retrieving information in a compressed format based in autoencoders, an unsupervised method supporting predictive learning. We then consider how different states in latent neural spaces corresponding to different types of sensory input could be linked by synchronous activation, providing the basis for a sparse implementation of memory using concept neurons. Finally, we will comment on some of the challenges and questions that link neuroscience and data science and that could have implications for both fields.
Zoom link: https://zoom.us/j/98286706234?pwd=bm1JUFVYcTJkaVl1VU55L0FiWDRIUT09
