arXiv:2610.17623v2 Announce Type: cross
Abstract: Rubric-based reinforcement learning grades a response against many written criteria but collapses them into one reward, which discards the detail the...
By Kunwang Zheng, Xinyu Sun, Yu Cao, Shuocheng Li, Lei Cheng, Ying Liu, Qilong Zhang, Jinlong Chen, Hua Yang, Qianglong Chen
CoRD (Coupled Rollout Distillation) enhances on‑policy distillation by aligning the student’s training state distribution with the teacher’s through a geometric interpolation of their rollouts, using maximal coupling to keep tokens where the distributions agree and substitute tokens otherwise. This approach keeps the on‑policy loss intact while reducing supervision gaps, leading to higher pass@1 and pass@8 scores across multiple Qwen3 and Qwen3.5 student models on public math benchmarks. Notably, a 4B student trained with CoRD closes performance gaps in just 16 steps, far faster than on‑policy distillation.
By Guhan Chen, Bohan Li, Jun Rao, Hejin Wang, Songtao Tian, Yexin Xie, Xiaojun Meng, Peng Yuan, Jiansheng Wei, Xuebo Liu, Zixiong Yu
BeatFlow-ECG is a conditional rectified‑flow model that reconstructs single‑channel ECG signals from synchronized photoplethysmography (PPG) and inertial measurement unit (IMU) data. The architecture uses a convolutional encoder‑decoder with a transformer bottleneck and explicit flow‑time conditioning, incorporating motion information through IMU‑derived features, motion‑dependent loss weighting, and a curriculum learning strategy. Evaluated on PPG‑DaLiA and WESAD datasets with leave‑one‑subject‑out protocols, BeatFlow‑ECG outperforms deterministic, adversarial, and diffusion‑based baselines, achieving Pearson correlations of 0.983–0.986 and R‑peak F1 scores of 0.946–0.955, while reducing L1 error compared to Conditional DDPM‑1D.
By Mohamed Kamel, Sahar Selim, Walaa Medhat, Tamer Nadeem
Continuous cardiac monitoring outside clinical settings requires signals that are both informative and practical to collect during daily life. Electrocardiography (ECG) provides rich information about...
The paper introduces two distillation methods for handling misaligned AI models. Distillation for Incrimination (DFI) transfers misalignment without the ability to conceal it, making students more likely to admit hidden behaviors. Distillation for Capabilities (DFC) transfers useful capabilities while minimizing misalignment, using techniques like inoculation prompting and focused training.
By Sebastian Prasanna, Jacqueline Tay, Alek Westover
To address the limitations of video-based emotion recognition under ambiguous or socially masked behavioral cues, as well as the poor deployability of physiological signals, this paper proposes a reliability-aware physiology-to-video knowledge distillation framework, termed BioKD. The proposed framework leverages physiological signals as privileged information during training to guide a video-based student model in learning deep affective representations, while relying solely on non-intrusive video inputs at inference time.