The Information Bottleneck principle and neural networksPublicationsDeep learning and the Information Bottleneck principleNaftali Tishby, Noga Zaslavsky2015 ITWPDF Cite DOIEarly motion processing circuit uses gap junctions to achieve efficient stimuli encodingSiwei Wang, Noga Zaslavsky, Alexander Borst, Naftali Tishby, Idan Segev2017 COSYNEPDF CiteEfficient encoding of motion is mediated by gap junctions in the fly visual systemSiwei Wang, Alexander Borst, Noga Zaslavsky, Naftali Tishby, Idan Segev2017 PLoS Comput BiolPDF Cite DOIDeterministic annealing and the evolution of Information Bottleneck representationsNoga Zaslavsky, Naftali Tishby2019 Technical reportPDF CiteInformation-Theoretic Principles in the Evolution of Semantic SystemsNoga Zaslavsky2020 PhD Thesis, The Hebrew UniversityPDF CiteToward human-like object naming in artificial neural systemsTiwalayo Eisape, Roger Levy, Joshua Tenenbaum, Noga Zaslavsky2020 BAICS @ ICLRPDF Cite Video ICLR WorkshopTowards Human-Agent Communication via the Information Bottleneck PrincipleMycal Tucker, Julie Shah, Roger Levy, Noga Zaslavsky2022 RSS Workshop on Social Intelligence in Humans and RobotsPDF Cite RSS WorkshopTrading off Utility, Informativeness, and Complexity in Emergent CommunicationMycal Tucker, Julie Shah, Roger Levy, Noga Zaslavsky2022 NeurIPSPDF Cite Code Supplementary Material
