Ben Hoover
AI Researcher with IBM Research and GA Tech
Mechanistic interpretability from the perspective of Associative Memory.
My research focuses on understanding AI foundation models from the perspective of dynamical systems. When we look under the hood of foundation models, do we expect them to look like a cold computer or a living brain?
I'm a self-branded AI re-interpretability and visualization engineer seeking a coherent theory around the intelligence we create.
Publications
Dec
2022
A Universal Abstraction for Hierarchical Hopfield Networks
🏆Spotlight Poster at DLDE Workshop, Poster at MemARI Workshop, NeurIPS 2022
Oct
2022
Interactive and Visual Prompt Engineering for Ad-hoc Task Adaptation with Large Language Models
Albert Webson
;
Victor Sanh
;
Johanna Beyer
;
Hanspeter Pfister
;
Alexander M Rush
Published at IEEE Vis 2022
Apr
2022
Shared Interest: Measuring Human-AI Alignment to Identify Recurring Patterns in Model Behavior
🏅Best Paper Honorable Mention CHI 2022; IEE Vis: VisXAI Workshop; Demo at NeurIPS 2020
Dec
2021
Interactive Exploration for 60 Years of AI Research
Demo at NeurIPS 2021
Jul
2021
Fairytailor: A multimodal generative framework for storytelling
An AI co-authoring tool to interactively write text and images into a story resembling a child's fairytale.
May
2021
Can a Fruit Fly Learn Word Embeddings?
Poster at ICLR 2021
Apr
2021
LMdiff: A Visual Diff Tool to Compare Language Models
System Demo at EMNLP 2021. Best Demo at Neurips 2020
Apr
2021
The Design and Development of a Game to Study Backdoor Poisoning Attacks: The Backdoor Game
Zahra Ashktorab
;
Casey Dugan
;
James Johnson
;
Aabhas Sharma
;
Dustin Ramsey Torres
;
Ingrid Lange
;
Heiko Ludwig
;
Bryant Chen
;
Nathalie Baracaldo
;
Werner Geyer
;
Qian Pan
Jan
2021
Operational response simulation tool for epidemics within refugee and IDP settlements
An operational response simulation tool for epidemics within settlements
Dec
2020
RXNMapper: Unsupervised Attention Guided Atom Mapping
Published in Science Advances 2021; Best Poster at #IOPPoster, Poster at ICML 2020 Workshop: AI for Scientific Discovery & ACS Fall 2020
Dec
2020
CogMol: Target-Specific and Selective Drug Design for COVID-19 Using Deep Generative Models
🏆Spotlight Poster at Neurips 2020
Jul
2020
exBERT: A Visual Analysis of Transformer Models
System Demo at ACL 2020. Nominated for Best Demo at Neurips 2019.
Sep
2017
Managing the SOS Response for Enhanced CRISPR-Cas-Based Recombineering in E. coli through Transient Inhibition of Host RecA Activity