Storing, learning and retrieving biased patterns

作者：

Highlights：

• The biased Hopfield model is shown to be formally equivalent to a Boltzmann Machine (BM) with intra-layer interactions and external field.

• The equivalence allows finding an effective initialisation for BM’s parameters under a contrastive divergence learning rule.

• Robustness is checked versus the training dataset quality.

摘要

•The biased Hopfield model is shown to be formally equivalent to a Boltzmann Machine (BM) with intra-layer interactions and external field.•The equivalence allows finding an effective initialisation for BM’s parameters under a contrastive divergence learning rule.•Robustness is checked versus the training dataset quality.

论文关键词：Neural networks,Disordered systems,Machine learning

论文评审过程：Received 10 August 2021, Revised 28 September 2021, Accepted 30 September 2021, Available online 31 October 2021, Version of Record 31 October 2021.

论文官网地址：https://doi.org/10.1016/j.amc.2021.126716