🏢 Arizona State University
TripletCLIP: Improving Compositional Reasoning of CLIP via Synthetic Vision-Language Negatives
·3187 words·15 mins·
loading
·
loading
Multimodal Learning
Vision-Language Models
🏢 Arizona State University
TripletCLIP boosts CLIP’s compositional reasoning by cleverly generating synthetic hard negative image-text pairs, achieving over 9% absolute improvement on SugarCrepe.
Enhancing Robustness of Last Layer Two-Stage Fair Model Corrections
·2233 words·11 mins·
loading
·
loading
AI Theory
Fairness
🏢 Arizona State University
Boosting fair machine learning’s robustness against noisy labels, this work introduces a novel label-spreading method, achieving state-of-the-art worst-group accuracy.
Chain of Thoughtlessness? An Analysis of CoT in Planning
·2944 words·14 mins·
loading
·
loading
Natural Language Processing
Large Language Models
🏢 Arizona State University
Chain of Thought prompting in LLMs offers limited generalizability, providing performance gains only when prompts are highly specific to problem types; highlighting a critical trade-off between perfor…
Belief-State Query Policies for User-Aligned Planning under Partial Observability
·1669 words·8 mins·
loading
·
loading
AI Applications
Robotics
🏢 Arizona State University
This paper introduces Belief-State Query (BSQ) constraints for user-aligned planning in partially observable settings, providing algorithms with guaranteed user alignment and computational feasibility…