{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/backpropagation-for-implicit-spectral","title":"Backpropagation for Implicit Spectral Densities","arxiv_id":"1806.00499","date":"2018-06-01","proceeding":null,"authors":["Aditya Ramesh","Yann Lecun"],"abstract":"Most successful machine intelligence systems rely on gradient-based learning,\nwhich is made possible by backpropagation. Some systems are designed to aid us\nin interpreting data when explicit goals cannot be provided. These unsupervised\nsystems are commonly trained by backpropagating through a likelihood function.\nWe introduce a tool that allows us to do this even when the likelihood is not\nexplicitly set, by instead using the implicit likelihood of the model.\nExplicitly defining the likelihood often entails making heavy-handed\nassumptions that impede our ability to solve challenging tasks. On the other\nhand, the implicit likelihood of the model is accessible without the need for\nsuch assumptions. Our tool, which we call spectral backpropagation, allows us\nto optimize it in much greater generality than what has been attempted before.\nGANs can also be viewed as a technique for optimizing implicit likelihoods. We\nstudy them using spectral backpropagation in order to demonstrate robustness\nfor high-dimensional problems, and identify two novel properties of the\ngenerator G: (1) there exist aberrant, nonsensical outputs to which G assigns\nvery high likelihood, and (2) the eigenvectors of the metric induced by G over\nlatent space correspond to quasi-disentangled explanatory factors.","url_abs":"http://arxiv.org/abs/1806.00499v1","url_pdf":"http://arxiv.org/pdf/1806.00499v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"backpropagation-for-implicit-spectral","repo_url":"https://github.com/EiffL/SpectralFlow","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}