Extended Geometric Jensen-Shannon Divergence for Positive Measures

New research extends the geometric Jensen-Shannon divergence to any positive measure, removing normalization. The paper, submitted 7 Aug 2025 and revised 18 Sep 2025, is 28 pages. Read more: getnews.me/extended-geometric-jense... #infotheory #ml

Protect Before Generate: Error Correcting Codes within Discrete Deep Generative Models Despite significant advancements in deep probabilistic models, learning low-dimensional discrete latent representations remains a challenging task. In this paper, we introduce a novel method that enha...

🚨 Paper accepted at UAI (2025)!
"Improved Variational Inference in Discrete VAEs using Error Correcting Codes". We explore the intersection of Information Theory and Generative AI—presenting this work in Brazil this July 🇧🇷

📄 arxiv.org/abs/2410.07840

#UAI2025 #GenAI #InfoTheory #MachineLearning

Join us next Tue, 27 May (17:00 CET) for Beyond The Edge seminar#5!
🎙️ Dr Thomas Varley (U Vermont) on dark information—the hidden structure networks miss.
📍 Live on Zoom →beyondtheedge.network/event/seminar-thomas-f-v...
#ComplexSystems #InfoTheory

Also, check out our ICLR’25 Oral Paper that extends this framework to larger networks and shows how to systematically identify and improve these per-neuron objectives: openreview.net/forum?id=CLE... #neuralnetworks #infotheory #AI #ML #MachineLearning #locallearning #collectiveintelligence 2/2