Mitigating Temporal Attacks in Deepfake Detection
Discover how SpInShield enhances deepfake detection by defending against temporal attacks and improving model robustness with spectral-invariant techniques...
Rubric-Based On-Policy Distillation for AI Model Alignment
Discover Rubric-based On-policy Distillation (ROPD), a scalable AI model alignment method that outperforms traditional logit-based techniques with 10x effi...
Reducing Unsolvability in Multi-LLM Routing: Key Insights
Explore how evaluation artifacts impact multi-LLM routing and learn methods to improve accuracy and cost-efficiency in AI model selection.
BalCapRL: Balanced RL Framework for MLLM Image Captioning
Discover BalCapRL, a balanced RL-based framework enhancing MLLM image captioning with improved accuracy, fluency, and utility across models.
Anchor-Centric Adaptation to Overcome Diversity Trap in Robotics
Discover how Anchor-Centric Adaptation improves robotic manipulation by balancing coverage and density for better real-world VLA model performance.