Cambridge ELLIS Seminar Series- Dr Roger Grosse- Studying Neural Net Generalization through Influence Functions

Add to your list(s) Download to your calendar using vCal

• Speaker to be confirmed
• Monday 06 March 2023, 16:15-17:15
• https://eng-cam.zoom.us/j/82557742057?pwd=Sk1sRWRxL09pV2FPYzNpK0FrU096Zz09.

If you have a question about this talk, please contact Kimberly Cole.

The Cambridge ELLIS Unit Seminar Series holds talks by leading researchers in the area of machine learning and AI. Next first speaker of 2023 will be Dr. Roger Grosse. Details of his talk can be found below.

Title: “Studying Neural Net Generalization through Influence Functions”

Abstract: How can we trace surprising behaviors of machine learning models back to their training data?  Influence functions aim to predict how the trained model would change if a specific training example were added to the training set. I’ll address two issues that have blocked their applicability to large-scale neural nets: apparent inaccuracy of the results, and the difficulty of computing inverse-Hessian-vector products. Towards the former issue, I’ll reformulate the goals of influence estimation in a way that applies to overparameterized, incompletely trained models, and argue that the apparent inaccuracy was largely illusory. I’ll then discuss an approach to scaling influence estimation to large language models and show some resulting insights into their patterns of generalization..  https://eng-cam.zoom.us/j/82557742057?pwd=Sk1sRWRxL09pV2FPYzNpK0FrU096Zz09

This talk is part of the Cambridge Ellis Unit series.

This talk is included in these lists:

• Cambridge Ellis Unit
• Hanchen DaDaDash
• Interested Talks
• https://eng-cam.zoom.us/j/82557742057?pwd=Sk1sRWRxL09pV2FPYzNpK0FrU096Zz09

Note that ex-directory lists are not shown.