From Associative Memories to Deep Networks and From Associative Memories to Universal Machines
Explore a comprehensive panel discussion featuring Professors Christos Papadimitriou, Tomaso A. Poggio, and Santosh Vempala, moderated by Kenneth Blum. Delve into the evolution of associative memories to deep networks and universal machines over the past 50 years. Discover how holography inspired associative memory models and how Willshaw networks, similar to modern deep nets, were implemented using threshold neurons. Gain insights into the realistic potential of deep learning in understanding human intelligence. Examine topics such as deep networks as memories, kernels, and lookup tables, their limitations, and evolutionary discoveries. Explore concepts like the grandmother myth, Jennifer Aniston cells, recurrent nets, assembly operations, and the neural basis of language. Investigate working memory, robustness, plasticity rules, and representations. Engage with thought-provoking questions from attendees and learn about relevant experiments in the field.