Mainak Deb

Unlike the brain, most ANNs lack any kind of organization of units🧠 In our paper (ICLR 2025) with @mayukh091 @apurvaratan, we present TopoLoss, a simple, scalable way to add brain-like topography to AI models with minimal performance loss🧵 toponets.github.io

What if different parts of a neural network were forced to learn different aspects of the input? They start showing specialized regions like the brain Listen to @MainakDeb19 talk about an architecture that got him an ICLR 2024 spotlight! New talk from @lossfunk!

My work got featured in @TEDAIVienna👀 AtMan (NeurIPS 2023) explains LLM outputs, telling you which part of the context was critical. ted.com/talks/jonas_an… neurips.cc/virtual/2023/p…

🚀At @NeurIPSConf tomorrow? Don't miss @nikolasmcneal and @MainakDeb19 's poster at the @unireps workshop on the adversarial sensitivity of vision encoding models of fMRI responses! A brief teaser about what they find.

I'll be joining @artflow_ai as an intern 🚀 @artflow_ai lets users generate visual content using #AI, exciting stuff! 🖌️🤖🎨 give it a try 👀 -> artflow.ai