Join us for #AI for science workshop at @Caltech on Feb. 9th for a hybrid workshop (both in-person and on Zoom). Register before Feb. 3rd here: https://
ai4science.caltech.edu/events/worksho
p.html
…
@animaanandkumar
-
AI for Science Workshop at Caltech February 9th Registration Open
By
–
-
Accessible Explanation of Diffusion Models Through Physical Ink Process
By
–
I love the accessible explanation of diffusion models through the physical process of ink diffusing through water and following that trajectory and reversing it.
-
Generative Models Potential for Improving Predictive Tasks
By
–
I talk about the enormous potential of generative models to improve downstream predictive tasks in this excellent article by @anilananth
-
Genome-Scale Language Model Predicts Coronavirus Variants of Concern
By
–
It is a privilege to collaborate with @arvindr_ at @argonne_lcf on genome-scale language model to predict new variants of concern for coronavirus
-
Advances in Surgical AI: Skill Assessment and Patient Outcome Prediction
By
–
We made strides in surgical #AI which involves assessing the skill of surgeons, predicting patient outcomes, and discovering novel surgeon biomarkers based on multi-modal data and deep learning algorithms. @AjhungMD gives an excellent overview here
-
NVIDIA Releases Extended Model Cards for Trustworthy AI
By
–
We made further headway towards enabling trustworthy AI at scale. We released extended model cards at @nvidia with model-specific information concerning bias, explainability, privacy, safety, and security.
-
ZerO: Deterministic Weight Initialization with Binary Values
By
–
We developed the first fully deterministic weight initialization procedure for training any neural network. Our scheme, ZerO, initializes networks’ weights with only zeros and ones based on identity, and Hadamard transforms. ZerO achieves SOTA https://
arxiv.org/abs/2110.12661 -
Robust Vision Transformer Architecture Wins Semantic Segmentation Challenge
By
–
We also developed robust vision transformer architecture, fully attention networks (FAN), with channel-based attention for robustness. We won the Semantic Segmentation Tracking of Robust Vision Challenge at ECCV. https://
arxiv.org/abs/2210.12852 -
Neural Operators Enable Single-Step Diffusion Model Image Generation
By
–
2022 saw diffusion models for generative AI. But one downside is very slow sampling. We used neural operators to replace ODE solvers and showed that just one function evaluation is sufficient to achieve the best image generation quality. https://
arxiv.org/abs/2211.13449 -
MineDojo: Generalist AI Agents with Text Prompts in Minecraft
By
–
Decision-Making with Generalist AI Agents: We built Minedojo framework in Minecraft with text prompts for open-ended tasks using internet-scale knowledge of videos, wiki, and text information. It was an outstanding paper at NeurIPS 2022.