IEEE Data Storage Best Paper Award for Hesam Salavati and team
© 2015 EPFL
The Data Storage Technical Committee of the IEEE Communications Society has awarded the Best Student Paper Award to Amin Karbasi, Amir Hesam Salavati, Professor Amin Shokrollahi and Lav R. Varshney for their paper "Noise-Enhanced Associative Memories".
The Data Storage Technical Committee of the IEEE Communications Society has awarded the Best Student Paper Award to Amin Karbasi, Amir Hesam Salavati, Professor Amin Shokrollahi (School of Computer and Communication Sciences) and Lav R. Varshney for their paper "Noise-Enhanced Associative Memories" published in Proceedings of Neural Information Processing Systems in 2013 NIPS Proceedings.
Abstract
Recent advances in associative memory design through structured pattern sets and graph-based inference algorithms have enabled reliable learning and recall of an exponential number of patterns. Although these designs correct external errors in recall, they assume neurons that compute noiselessly, in contrast to the highly variable neurons found within the hippocampus and olfactory cortex. Here we consider associative memories with noisy internal computations and analytically characterize performance. As long as the internal noise level is below a specified threshold, the error probability in the recall phase can be made exceedingly small. More surprisingly, we show that internal noise actually improves the performance of the recall phase. Computational experiments lend additional support to our theoretical analysis. This work suggests a functional benefit to noisy neurons in biological neuronal networks.
About the IEEE Data Storage paper awards
Each year, the Data Storage Technical Committee of the IEEE Communications Society gives a Best Paper Award and Best Student Paper Award for papers published in the previous year. The recognized papers have substantial and meaningful impact on both theory and practice for existing or future data storage systems or for emerging storage technologies that are gaining momentum. By giving these awards, the DSTC motivates researchers to continue to make great original contributions to the field of data storage.
More details can be found on the awards page of the DSTC website: http://committees.comsoc.org/dstc/index.php?page=award.
Further information on the paper can be found here: http://papers.nips.cc/paper/4875-noise-enhanced-associative-memories.