Super excited to share our paper on efficient training of Neural Operators with mixed precision, recently accepted at #ICLR2024! We show more than 50% gain in throughput with almost no loss in accuracy. Neural operators are #AI methods for solving #PDE . Unlike traditional
@animaanandkumar
-
Distinguished Alumnus Award from IIT Madras Honors Academic Excellence
By
–
Honored to be awarded the Distinguished Alumnus Award by @iitmadras Feeling nostalgic about the time I spent there and how fast time has flown!
-
Tensors Multilinear Operations Universal Approximation Machine Learning
By
–
Tensors are back! I am so glad to see this! We have been advocating that tensor or multilinear operations are all you need. The benefits: universal approximation and expressivity. Higher order BLAS operations for parallelism without the need for any transposition or data
-
First Multimodal Text-Chemical Structure Model for Chemistry
By
–
Text understanding with #LLMs is useful but not enough for scientific understanding and discovery. In chemistry, in addition to text, chemical structure is essential to determine the properties of molecules.
— Prof. Anima Anandkumar (@AnimaAnandkumar) 29 janvier 2024
We have created the first multimodal text-chemical structure model:… pic.twitter.com/157QheisUwText understanding with #LLMs is useful but not enough for scientific understanding and discovery. In chemistry, in addition to text, chemical structure is essential to determine the properties of molecules. We have created the first multimodal text-chemical structure model:
-
Lean Copilot Adopted in Undergraduate AI Course
By
–
Great to see our Lean copilot and talk by @KaiyuYang4 featured and being adopted in the undergraduate course. https://
t.co/8rbiAzC0MS -
Clinical Studies Exploration for Design Implementation
By
–
Thank you for your input. We would love to explore next steps in terms of clinical studies to try out our designs. Would love to discuss more!
-
AI Designed Catheter Reduces Bacterial Contamination by 100x
By
–
Thank you @sciam for covering our recent work on using #AI to design catheter that cuts down bacterial contamination by 100 times. "with help from artificial intelligence, researchers have designed a new catheter that they say could reduce bacterial contamination by up to two
-
Synthetic Data Generation Reduces PDE Learning Data Requirements
By
–
A simple yet effective way to overcome the data requirements in learning PDEs. We can use the equations to generate synthetic data in the reverse direction and reduce the dependence on numerical solvers => meaning start with a potential solution and generate the input using PDE.
-
Neural Operators Enable Advanced Medical Catheter Design Innovation
By
–
Our work on using Neural Operators to design a medical catheter just got published in Science Advances. Below you can read the @caltech article giving details on how the project came about as an interdisciplinary collaboration.
You can read more about the AI-side of catheter