Dear Colleagues,
I'm sending you this friendly reminder for the upcoming webinar. You are cordially invited to join 2023 Dartmouth Summer Data Science and AI Webinar Series - Week 8. Please take a look at the details of the event below.
Title: Large Language Model Tomography
Date/Time: August 24, Thursday, 11:00 AM – 12:00 PM EST.
Abstract:
Large Language Models (LLM), such as ChatGPT, have made a big splash recently. A question many potential users are asking is whether such systems can be trusted when inserted into business processes. What do they "know"? What can they do? What is in the architecture? What were they trained on? Are they "ethical" and "fair"? How good are the programs they write? Overall, can they be "trusted"?
This talk will introduce the concept of LLM Tomography. As in classical tomography, the idea is to send a variety of signals into an opaque object with the goal of estimating some properties of the interior of the object. This involves crafting the input signals (prompts in the case of LLMs) and analyzing the responses. We will show some preliminary results about estimating various properties of an LLM using this paradigm.
Bio of the speaker:
George Cybenko is the Dorothy and Walter Gramm Professor of Engineering at Dartmouth. Professor Cybenko has made key research contributions in signal processing, neural computing, parallel processing and computational behavioral analysis. He was the Founding Editor-in-Chief of IEEE/AIP Computing in Science and Engineering, IEEE Security & Privacy and IEEE Transactions on Computational Social Systems. He has served on the Defense Science Board and the Air Force Scientific Advisory Board in the past and is an advisor to the Army Cyber Institute at West Point. Professor Cybenko is a Fellow of the IEEE and SIAM. He received his BS (University of Toronto) and PhD Princeton) degrees in Mathematics. Cybenko was a co-founder of Flowtraq Inc, which was acquired by Riverbed Technology in 2017.
This webinar will be offered online via Zoom. Please register to receive the Zoom link (one day before the webinar).
Registration link: libcal.dartmouth.edu/calendar/itc/2023DSAIW8
Look forward to seeing you at the webinar!
------------------------------
Jianjun Hua
Dartmouth College
------------------------------