Kai Chen

Zijian Wu, Lingkai Kong, Wenwei Zhang et al.

Large language models (LLMs) have achieved significant progress in solving complex reasoning tasks by Reinforcement Learning with Verifiable Rewards (RLVR). This advancement is also inseparable from the oversight automated by reliable verifiers. However, current outcome-based verifiers (OVs) are unable to inspect the unreliable intermediate steps in the long reasoning chains of thought (CoTs). Meanwhile, current process-based verifiers (PVs) have difficulties in reliably detecting errors in the complex long CoTs, limited by the scarcity of high-quality annotations due to the prohibitive costs of human annotations. Therefore, we propose the Outcome-based Process Verifier (OPV), which verifies the rationale process of summarized outcomes from long CoTs to achieve both accurate and efficient verification and enable large-scale annotation. To empower the proposed verifier, we adopt an iterative active learning framework with expert annotations to progressively improve the verification capability of OPV with fewer annotation costs. Specifically, in each iteration, the most uncertain cases of the current best OPV are annotated and then subsequently used to train a new OPV through Rejection Fine-Tuning (RFT) and RLVR for the next round. Extensive experiments demonstrate OPV's superior performance and broad applicability. It achieves new state-of-the-art results on our held-out OPV-Bench, outperforming much larger open-source models such as Qwen3-Max-Preview with an F1 score of 83.1 compared to 76.3. Furthermore, OPV effectively detects false positives within synthetic dataset, closely align with expert assessment. When collaborating with policy models, OPV consistently yields performance gains, e.g., raising the accuracy of DeepSeek-R1-Distill-Qwen-32B from 55.2% to 73.3% on AIME2025 as the compute budget scales.

Shihan Dou, Shichun Liu, Yuming Yang et al.

We offer a novel perspective on reward modeling by formulating it as a policy discriminator, which quantifies the difference between two policies to generate a reward signal, guiding the training policy towards a target policy with desired behaviors. Based on this conceptual insight, we propose a scalable pre-training method named Policy Discriminative Learning (POLAR), which trains a reward model (RM) to discern identical policies and discriminate different ones. Unlike traditional reward modeling methods relying on absolute preferences, POLAR captures the relative difference between one policy and an arbitrary target policy, which is a scalable, high-level optimization objective suitable for modeling generic ranking relationships. Leveraging the POLAR pre-training paradigm, we present a series of RMs with parameter scales from 1.8B to 7B. Empirical results show that POLAR substantially outperforms traditional non-pre-trained methods, significantly enhancing RM performance. For instance, POLAR-7B could improve preference accuracy from 54.8% to 81.0% on STEM tasks and from 57.9% to 85.5% on creative writing tasks compared to SOTA baselines. POLAR also shows robust generalization capabilities in RLHF using Reinforcement Fine-tuning (RFT), providing reliable reward signals and markedly enhancing policy performance--improving LLaMa3.1-8B from an average of 47.36% to 56.33% and Qwen2.5-32B from 64.49% to 70.47% on 20 benchmarks. Moreover, scaling experiments reveal a clear power-law relationship between computation and performance, supported by linear correlation coefficients approaching 0.99. The impressive performance, strong generalization, and scaling properties suggest that POLAR is a promising direction for developing general and strong reward models.

Chiu-Han Hsiao, Kai Chen, Tsung-Yu Peng et al.

The traditional interpretation of Intravascular Ultrasound (IVUS) images during Percutaneous Coronary Intervention (PCI) is time-intensive and inconsistent, relying heavily on physician expertise. Regulatory restrictions and privacy concerns further hinder data integration across hospital systems, complicating collaborative analysis. To address these challenges, a parallel 2D U-Net model with a multi-stage segmentation architecture has been developed, utilizing federated learning to enable secure data analysis across institutions while preserving privacy. The model segments plaques by identifying and subtracting the External Elastic Membrane (EEM) and lumen areas, with preprocessing converting Cartesian to polar coordinates for improved computational efficiency. Achieving a Dice Similarity Coefficient (DSC) of 0.706, the model effectively identifies plaques and detects circular boundaries in real-time. Collaborative efforts with domain experts enhance plaque burden interpretation through precise quantitative measurements. Future advancements may involve integrating advanced federated learning techniques and expanding datasets to further improve performance and applicability. This adaptable technology holds promise for environments handling sensitive, distributed data, offering potential to optimize outcomes in medical imaging and intervention.

Yinuo Wang, Yue Zeng, Kai Chen et al.

Introduction: Timely identification of intracranial hemorrhage (ICH) subtypes on non-contrast computed tomography is critical for prognosis prediction and therapeutic decision-making, yet remains challenging due to low contrast and blurring boundaries. This study evaluates the performance of zero-shot multi-modal large language models (MLLMs) compared to traditional deep learning methods in ICH binary classification and subtyping. Methods: We utilized a dataset provided by RSNA, comprising 192 NCCT volumes. The study compares various MLLMs, including GPT-4o, Gemini 2.0 Flash, and Claude 3.5 Sonnet V2, with conventional deep learning models, including ResNet50 and Vision Transformer. Carefully crafted prompts were used to guide MLLMs in tasks such as ICH presence, subtype classification, localization, and volume estimation. Results: The results indicate that in the ICH binary classification task, traditional deep learning models outperform MLLMs comprehensively. For subtype classification, MLLMs also exhibit inferior performance compared to traditional deep learning models, with Gemini 2.0 Flash achieving an macro-averaged precision of 0.41 and a macro-averaged F1 score of 0.31. Conclusion: While MLLMs excel in interactive capabilities, their overall accuracy in ICH subtyping is inferior to deep networks. However, MLLMs enhance interpretability through language interactions, indicating potential in medical imaging analysis. Future efforts will focus on model refinement and developing more precise MLLMs to improve performance in three-dimensional medical image processing.

Jiaqi Liu, Yonghao Long, Kai Chen et al.

Accurate segmentation of surgical instrument tip is an important task for enabling downstream applications in robotic surgery, such as surgical skill assessment, tool-tissue interaction and deformation modeling, as well as surgical autonomy. However, this task is very challenging due to the small sizes of surgical instrument tips, and significant variance of surgical scenes across different procedures. Although much effort has been made on visual-based methods, existing segmentation models still suffer from low robustness thus not usable in practice. Fortunately, kinematics data from the robotic system can provide reliable prior for instrument location, which is consistent regardless of different surgery types. To make use of such multi-modal information, we propose a novel visual-kinematics graph learning framework to accurately segment the instrument tip given various surgical procedures. Specifically, a graph learning framework is proposed to encode relational features of instrument parts from both image and kinematics. Next, a cross-modal contrastive loss is designed to incorporate robust geometric prior from kinematics to image for tip segmentation. We have conducted experiments on a private paired visual-kinematics dataset including multiple procedures, i.e., prostatectomy, total mesorectal excision, fundoplication and distal gastrectomy on cadaver, and distal gastrectomy on porcine. The leave-one-procedure-out cross validation demonstrated that our proposed multi-modal segmentation method significantly outperformed current image-based state-of-the-art approaches, exceeding averagely 11.2% on Dice.