Congratulations to the Livermore researchers whose work has been accepted to the 43rd International Conference on Machine Learning (ICML) on July 6–11 in Seoul, South Korea. According to the ICML website, the event showcases “research on all aspects of machine learning used in closely related areas like artificial intelligence, statistics and data science, as well as important application areas such as machine vision, computational biology, speech recognition, and robotics.” Follow @Livermore_Comp on X with the #ICML26 hashtag. Websites, abstracts, and/or preprints are linked below.
Posters
- GUI-Spotlight: Adaptive Iterative Focus Refinement for Enhanced GUI Visual Grounding | Chunhua Liao
- LongCoT: Benchmarking Long-Horizon Chain-of-Thought Reasoning | Daniel Nichols, Peggy Li, Tavish McDonald, Ruben Glatt, Tal Ben-Nun, Brian Bartoldson, Bhavya Kailkhura
- Position: Zeroth-Order Optimization in Deep Learning Is Underexplored, Not Underpowered | James Diffenderfer, Bhavya Kailkhura
Workshop Contributions
- End-to-End Context Compression at Scale | Bhavya Kailkhura, Harshitha Menon, Brian Bartoldson
- Multi-Token Prediction via Self-Distillation | Brian Bartoldson
- On the Role of Mechanistic Interpretability for Vision-Language Prompt Learning | Kowshik Thopalli
- Pre-Training on Noncovalent Interactions from Synthetic Protein-Ligand Structures to Better Predict Binding Affinity | Garrett Stevenson, Jonathan Allen, Hyojin Kim
- Revisiting NeRN: Optimizing Training Strategies for Weight Reconstruction | Hongjun Choi, Ruben Glatt, Shusen Liu
- Watermarking for Proprietary Dataset Protection | Brian Bartoldson, Bhavya Kailkhura