|Title||From Associative Memories to Powerful Machines|
|Publication Type||CBMM Memos|
|Year of Publication||2021|
Associative memories were implemented as simple networks of threshold neurons by Willshaw and Longuet-Higgins in the '60s. Today's deep networks are quite similar: they can be regarded as approximating look-up tables, similar to Gaussian RBF networks. Thinking about deep networks as large associative memories provides a more realistic and sober perspective on the promises of deep learning.
- CBMM Funded