{"conference":{"location":"Mumbai, India","name":"ITW: Information Theory Workshop","start_date":"2022-11-01","end_date":"2022-11-09"},"date_published":"2022-11-16T00:00:00Z","publication":"IEEE Information Theory Workshop","publication_status":"published","day":"16","external_id":{"arxiv":["2205.08199"]},"doi":"10.1109/ITW54588.2022.9965870","author":[{"full_name":"Amani, Mohammad Hossein","last_name":"Amani","first_name":"Mohammad Hossein"},{"id":"ca726dda-de17-11ea-bc14-f9da834f63aa","last_name":"Bombari","full_name":"Bombari, Simone","first_name":"Simone"},{"first_name":"Marco","full_name":"Mondelli, Marco","last_name":"Mondelli","id":"27EB676C-8706-11E9-9510-7717E6697425","orcid":"0000-0002-3242-7020"},{"last_name":"Pukdee","full_name":"Pukdee, Rattana","first_name":"Rattana"},{"first_name":"Stefano","full_name":"Rini, Stefano","last_name":"Rini"}],"_id":"12538","oa":1,"main_file_link":[{"url":" https://doi.org/10.48550/arXiv.2205.08199","open_access":"1"}],"quality_controlled":"1","language":[{"iso":"eng"}],"page":"588-593","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","citation":{"mla":"Amani, Mohammad Hossein, et al. “Sharp Asymptotics on the Compression of Two-Layer Neural Networks.” <i>IEEE Information Theory Workshop</i>, IEEE, 2022, pp. 588–93, doi:<a href=\"https://doi.org/10.1109/ITW54588.2022.9965870\">10.1109/ITW54588.2022.9965870</a>.","short":"M.H. Amani, S. Bombari, M. Mondelli, R. Pukdee, S. Rini, IEEE Information Theory Workshop (2022) 588–593.","ista":"Amani MH, Bombari S, Mondelli M, Pukdee R, Rini S. 2022. Sharp asymptotics on the compression of two-layer neural networks. IEEE Information Theory Workshop., 588–593.","ama":"Amani MH, Bombari S, Mondelli M, Pukdee R, Rini S. Sharp asymptotics on the compression of two-layer neural networks. <i>IEEE Information Theory Workshop</i>. 2022:588-593. doi:<a href=\"https://doi.org/10.1109/ITW54588.2022.9965870\">10.1109/ITW54588.2022.9965870</a>","ieee":"M. H. Amani, S. Bombari, M. Mondelli, R. Pukdee, and S. Rini, “Sharp asymptotics on the compression of two-layer neural networks,” <i>IEEE Information Theory Workshop</i>. IEEE, pp. 588–593, 2022.","apa":"Amani, M. H., Bombari, S., Mondelli, M., Pukdee, R., &#38; Rini, S. (2022). Sharp asymptotics on the compression of two-layer neural networks. <i>IEEE Information Theory Workshop</i>. Mumbai, India: IEEE. <a href=\"https://doi.org/10.1109/ITW54588.2022.9965870\">https://doi.org/10.1109/ITW54588.2022.9965870</a>","chicago":"Amani, Mohammad Hossein, Simone Bombari, Marco Mondelli, Rattana Pukdee, and Stefano Rini. “Sharp Asymptotics on the Compression of Two-Layer Neural Networks.” <i>IEEE Information Theory Workshop</i>. IEEE, 2022. <a href=\"https://doi.org/10.1109/ITW54588.2022.9965870\">https://doi.org/10.1109/ITW54588.2022.9965870</a>."},"type":"journal_article","date_created":"2023-02-10T13:47:56Z","article_processing_charge":"No","scopus_import":"1","title":"Sharp asymptotics on the compression of two-layer neural networks","date_updated":"2023-12-18T11:31:47Z","month":"11","year":"2022","abstract":[{"text":"In this paper, we study the compression of a target two-layer neural network with N nodes into a compressed network with M<N nodes. More precisely, we consider the setting in which the weights of the target network are i.i.d. sub-Gaussian, and we minimize the population L_2 loss between the outputs of the target and of the compressed network, under the assumption of Gaussian inputs. By using tools from high-dimensional probability, we show that this non-convex problem can be simplified when the target network is sufficiently over-parameterized, and provide the error rate of this approximation as a function of the input dimension and N. In this mean-field limit, the simplified objective, as well as the optimal weights of the compressed network, does not depend on the realization of the target network, but only on expected scaling factors. Furthermore, for networks with ReLU activation, we conjecture that the optimum of the simplified optimization problem is achieved by taking weights on the Equiangular Tight Frame (ETF), while the scaling of the weights and the orientation of the ETF depend on the parameters of the target network. Numerical evidence is provided to support this conjecture.","lang":"eng"}],"publication_identifier":{"isbn":["9781665483414"]},"publisher":"IEEE","article_type":"original","status":"public","oa_version":"Preprint","department":[{"_id":"MaMo"}]}