Pinned
Chaitanya K. Joshi
3,733 posts
Chaitanya K. Joshi
@chaitjo
AI for biology 🧬
Stanford Data Science Fellow @RDasLab
PhD @Cambridge_Uni · prev @AIatMeta @PrescientDesign @MRC_LMB
Are you applying for a PhD in Machine Learning, Artificial Intelligence, and beyond? Here's a thread of high-quality resources that helped me understand the process + craft my application better. 👇- In one interview, Demis talked about how he likes to work at night b/c there's no disturbance and time feels 'infinite' in the sense that you can just stay awake/keep working till the next day if you are in a flow state. CS undergrads will probably relate...Nothing beats burning the midnight oil in a maximum state of flow...
- Excited to share a blog post on the connection between #Transformers for NLP and #GraphNeuralNetworks (GNNs or GCNs). graphdeeplearning.github.io/post/transform…
- ❓New to Geometric GNNs, GDL, PyTorch Geometric, etc.? Want to understand how theory/equations connect to real code? Try this practical notebook before diving into this exciting area! **Geometric GNNs 101** github.com/chaitjo/geomet…
🚀Excited+nervous to share our latest work on understanding geometric GNNs for biomolecules, materials, etc. "On the Expressive Power of Geometric GNNs" with @crisbodnar @SimMat20 @TacoCohen @pl219_Cambridge PDF: arxiv.org/abs/2301.09308 Code: github.com/chaitjo/geomet… Findings👇 - On the bottleneck introduced by Attention as a GNN aggregation function - Packing information into a single attention weight (as opposed to a d-dim message) regularises the model. I find this concept very beautiful and elegant!
- Our first attempts at mechanistic interpretability of Transformers from the perspective of network science and graph theory! A wonderful collaboration with superstars @elb4tu, Deepro Choudhury, @pl219_Cambridge as part of the Geometric Deep Learning class at @Cambridge_CL!
- Really thought-provoking new paper on representation learning and the notion of 'semantic compression' by @ChenShani2 @jurafsky @ylecun @ziv_ravid
- Needing a permanent doi was a forcing function to revisit this article on Transformers and Graph Neural Networks. This is (very) old news, so what's new?
Excited to share a blog post on the connection between #Transformers for NLP and #GraphNeuralNetworks (GNNs or GCNs). graphdeeplearning.github.io/post/transform… - Flow matching has gained popularity recently which is better, diffusion or flow matching? They are formally equivalent Our purpose is to help practitioners understand and use these frameworks interchangeably -- **regardless of what it’s called**
- Introducing All-atom Diffusion Transformers — towards Foundation Models for generative chemistry, from my internship with the FAIR Chemistry team @OpenCatalyst @AIatMeta There are a couple ML ideas which I think are new and exciting in here 👇
- Hi GNN friends 👋, I'm creating an awesome-list on efficient Graph Neural Networks and scalable Graph Representation Learning. I'm excited about real-world applications of GNNs, and am looking to learn more about deploying them! Please help improve:
- Looking forward to publicly talking about my PhD research for the first time, together with @SimMat20! 💙 "Graph Neural Networks for Geometric Graphs" When: this Tuesday, 8 November 2022, 1pm - 2pm GMT. Where: LT2, Computer Lab as well as Zoom. Link: talks.cam.ac.uk/talk/index/183…
- In case anyone is reading...I feel a bit broken by my NeurIPS rebuttal experience this year :(
