Hello, my name is Meet Barot. I'm interested in machine learning, proteins, networks, collective intelligence and decentralization.
I did my PhD at New York University in the Center for Data Science; I graduated in 2023. Afterward I did some consulting work for companies that need AI expertise, focusing on language models and network analysis.
Currently, I'm working on an AI/ALife research company I started called Mythos Scientific. I'm mainly developing meta-learning and open-ended learning methods for neural networks. Specifically, I'm modeling neural network learning as neural cellular automata processes.
If you'd like to chat, email me at meet@mythos.science.
Late Breaking Abstract Poster: Meta-Neural Cellular Automata
Late Breaking Abstract Document: Meta-Neural Cellular Automata
Demo video showing a neural network being updated to solve the iris classification task (the "task neural network"), using a local update rule which is defined by another neural network (the "local rule network"). In this video, I am setting weights to zero (in red) to see if the local rule updates afterward can recover performance of the network:
Tymor Hamamsy, Meet Barot, James T Morton, Martin Steinegger, Richard Bonneau, and Kyunghyun Cho, Learning sequence, structure, and function representations of proteins with language models, bioRxiv 2023.11.26.568742; doi: https://doi.org/10.1101/2023.11.26.568742
Meet Barot, Vladimir Gligorijević, Richard Bonneau, Kyunghyun Cho, Automated Protein Function Description for Novel Class Discovery, bioRxiv 2022.10.13.512154; doi: https://doi.org/10.1101/2022.10.13.512154 (A version of this was in the NeurIPS 2022 AI4Science workshop here.)
Meet Barot, Vladimir Gligorijević, Kyunghyun Cho, Richard Bonneau, NetQuilt: deep multispecies network-based protein function prediction using homology-informed network similarity, Bioinformatics, 2021;, btab098, https://doi.org/10.1093/bioinformatics/btab098
NetQuilt Simons Foundation News Article
NetQuilt CDS Blog Post
Vladimir Gligorijević, Meet Barot, Richard Bonneau, deepNF: deep network fusion for protein function prediction, Bioinformatics, Volume 34, Issue 22, 15 November 2018, Pages 3873-3881, https://doi.org/10.1093/bioinformatics/bty440
As a function prediction method submitter to the 3rd Critical Assessment of Function Annotation (CAFA) challenge:
Zhou, N., Jiang, Y., Bergquist, T.R. et al. The CAFA challenge reports improved protein function prediction and new functional annotations for hundreds of genes through experimental screens. Genome Biol 20, 244 (2019). https://doi.org/10.1186/s13059-019-1835-8
Meet Barot, Daniel Berenberg, James Morton, Vladimir Gligorijević, Kyunghyun Cho, Richard Bonneau, (2020). Learning sequence, structure and network features for protein function prediction. 20-minute talk delivered virtually at the ISMB Conference. Recording
Meet Barot, Vladimir Gligorijević, Kyunghyun Cho, Richard Bonneau, (2019). Graph-Regularized Autoencoders for Protein Feature Learning. 10-minute talk delivered at the ISMB/ECCB Joint Conference, Basel, Switzerland. Recording
Function SIG Poster: Automated Protein Function Description for Novel Class Discovery
Function SIG Talk Presentation Slides: seqSCAN: Unsupervised Classification of Proteins for New Function Discovery
Zhuolin's jewelry store! (Etsy alternative)