Intel (United States)

3 related articles

Subscribe to updates

arXiv Nov 8, 2026
Video action explanations split into what the model sees and when it moves: lowest deletion AUC on most axes across three datasets, with masks about five times sparser
Explanations for video action recognition usually fold spatial appearance and temporal motion into a single saliency map, so a user cannot tell whether the model predicted from the object or from the movement. This work separates the explanation into a spatial appearance mask and a temporal motion mask, optimized jointly against a frozen classifier, with nearest-frame replacement used to remove temporal dynamics. On EPIC-Kitchens-55, EGTEA Gaze+, and Something-Something V2, the method attains the lowest deletion AUC on most datasets and axes, with masks roughly one-fifth the size of the strongest baseline.
Explanations for video action recognition usually fold spatial appearance and temporal motion into a single saliency map, so a user cannot tell whether the model predicted from the object or from the movement. This work separates the explanation into a spatial appearance mask and a temporal motion mask, optimized jointly against a frozen classifier, with nearest-frame replacement used to remove temporal dynamics. On EPIC-Kitchens-55, EGTEA Gaze+, and Something-Something V2, the method attains the lowest deletion AUC on most datasets and axes, with masks roughly one-fifth the size of the strongest baseline.
Oregon State University Intel (United States)
Bhavan Vasu（First author）
Factored Spatio-Temporal Explanations Action Recognition Separation Evaluation Protocol Straight-Through Estimator
Nature Computational Science Oct 12, 2026
LLMs learn which elements are present but not where atoms sit, so geometric models keep winning on materials properties
On coordinate–category data, language models learn which elements are present but not how atoms are arranged. Using synthetic datasets with tunable coordinate and category weights, plus the MatText benchmark across many text representations and scales, coordinate-dominated tasks show consistently higher error than category-dominated ones, a coordinate–category cliff. Scaling to about two million structures and 70 billion parameters does not remove it, while graph neural networks and other geometric architectures stay ahead on structure-dominated properties.
On coordinate–category data, language models learn which elements are present but not how atoms are arranged. Using synthetic datasets with tunable coordinate and category weights, plus the MatText benchmark across many text representations and scales, coordinate-dominated tasks show consistently higher error than category-dominated ones, a coordinate–category cliff. Scaling to about two million structures and 70 billion parameters does not remove it, while graph neural networks and other geometric architectures stay ahead on structure-dominated properties.
Intel (United States)Friedrich Schiller University Jena
Nawaf Alampara（First author, Corresponding author）Kevin Maik Jablonka（Corresponding author）
CC-cliff (coordinate–category cliff)MatText Material Property Prediction Large language models
arXiv Feb 11, 2026
Power-SMC targets the sequence-level power distribution with parallel candidates and importance-weight pruning, matching or exceeding MH sampling on MATH500, GSM8K, GPQA and HumanEval while accelerating inference by up to 17.6x
The work introduces Power-SMC, a training-free sampling method that maintains multiple candidate sequences in parallel, scores each with an importance weight measuring how well it matches the sequence-level power distribution, and periodically prunes low-scoring candidates in favor of high-scoring ones, matching or exceeding Metropolis-Hastings sampling in accuracy on MATH500, GSM8K, GPQA and HumanEval, preserving output diversity, and accelerating inference by up to 17.6x.
The work introduces Power-SMC, a training-free sampling method that maintains multiple candidate sequences in parallel, scores each with an importance weight measuring how well it matches the sequence-level power distribution, and periodically prunes low-scoring candidates in favor of high-scoring ones, matching or exceeding Metropolis-Hastings sampling in accuracy on MATH500, GSM8K, GPQA and HumanEval, preserving output diversity, and accelerating inference by up to 17.6x.
Intel (United States)University of Southern California
Seyedarmin Azizi（First author）
Power-SMC Power Sampling Training-Free LLM Reasoning

You’re all caught up

Video action explanations split into what the model sees and when it moves: lowest deletion AUC on most axes across three datasets, with masks about five times sparser

LLMs learn which elements are present but not where atoms sit, so geometric models keep winning on materials properties

Power-SMC targets the sequence-level power distribution with parallel candidates and importance-weight pruning, matching or exceeding MH sampling on MATH500, GSM8K, GPQA and HumanEval while accelerating inference by up to 17.6x