Human-Centered Evaluation of Shapley XAI in High-Stakes AI
Explore a human-focused audit of Shapley XAI benchmarks, revealing gaps in AI explanations and boosting decision confidence in risk settings.
AI-Assisted Verified Code Generation with Dafny Formal Verification
Explore AI-driven verified code generation using Dafny formal verification and advanced prompting to ensure reliable, error-free software development.
HiLight: Enhancing Evidence Selection in Frozen LLMs
Discover HiLight, a novel framework improving evidence highlighting in frozen LLMs to boost reasoning without altering original inputs.
Data-Free Client Contribution Estimation in Federated Learning
Discover data-free methods using gradient von Neumann entropy for accurate client contribution estimation in federated learning while ensuring privacy.
Improving Hierarchical Driving VQA with Cross-Stage Coherence
Explore cross-stage coherence in hierarchical driving VQA using explicit baselines and gated context projectors to boost autonomous driving AI accuracy.