Qian Shao

Qiyuan Chen, Hongsen Huang, Jiahe Chen et al.

Vision-language reward modeling faces a dilemma: generative approaches are interpretable but slow, while discriminative ones are efficient but act as opaque "black boxes." To bridge this gap, we propose VL-MDR (Vision-Language Multi-Dimensional Reward), a framework that dynamically decomposes evaluation into granular, interpretable dimensions. Instead of outputting a monolithic scalar, VL-MDR employs a visual-aware gating mechanism to identify relevant dimensions and adaptively weight them (e.g., Hallucination, Reasoning) for each specific input. To support this, we curate a dataset of 321k vision-language preference pairs annotated across 21 fine-grained dimensions. Extensive experiments show that VL-MDR consistently outperforms existing open-source reward models on benchmarks like VL-RewardBench. Furthermore, we show that VL-MDR-constructed preference pairs effectively enable DPO alignment to mitigate visual hallucinations and improve reliability, providing a scalable solution for VLM alignment.

Yixuan Wu, Kaiyuan Hu, Qian Shao et al.

The advent of telemedicine represents a transformative development in leveraging technology to extend the reach of specialized medical expertise to remote surgeries, a field where the immediacy of expert guidance is paramount. However, the intricate dynamics of Operating Room (OR) scene pose unique challenges for telemedicine, particularly in achieving high-fidelity, real-time scene reconstruction and transmission amidst obstructions and bandwidth limitations. This paper introduces TeleOR, a pioneering system designed to address these challenges through real-time OR scene reconstruction for Tele-intervention. TeleOR distinguishes itself with three innovative approaches: dynamic self-calibration, which leverages inherent scene features for calibration without the need for preset markers, allowing for obstacle avoidance and real-time camera adjustment; selective OR reconstruction, focusing on dynamically changing scene segments to reduce reconstruction complexity; and viewport-adaptive transmission, optimizing data transmission based on real-time client feedback to efficiently deliver high-quality 3D reconstructions within bandwidth constraints. Comprehensive experiments on the 4D-OR surgical scene dataset demostrate the superiority and applicability of TeleOR, illuminating the potential to revolutionize tele-interventions by overcoming the spatial and technical barriers inherent in remote surgical guidance.

Qian Shao, Shih-Fen Cheng

Optimizing delivery routes for last-mile logistics service is challenging and has attracted the attention of many researchers. These problems are usually modeled and solved as variants of vehicle routing problems (VRPs) with challenging real-world constraints (e.g., time windows, precedence). However, despite many decades of solid research on solving these VRP instances, we still see significant gaps between optimized routes and the routes that are actually preferred by the practitioners. Most of these gaps are due to the difference between what's being optimized, and what the practitioners actually care about, which is hard to be defined exactly in many instances. In this paper, we propose a novel hierarchical route optimizer with learnable parameters that combines the strength of both the optimization and machine learning approaches. Our hierarchical router first solves a zone-level Traveling Salesman Problem with learnable weights on various zone-level features; with the zone visit sequence fixed, we then solve the stop-level vehicle routing problem as a Shortest Hamiltonian Path problem. The Bayesian optimization approach is then introduced to allow us to adjust the weights to be assigned to different zone features used in solving the zone-level Traveling Salesman Problem. By using a real-world delivery dataset provided by the Amazon Last Mile Routing Research Challenge, we demonstrate the importance of having both the optimization and the machine learning components. We also demonstrate how we can use route-related features to identify instances that we might have difficulty with. This paves ways to further research on how we can tackle these difficult instances.

Qian Shao, Jiangrui Kang, Qiyuan Chen et al.

Semi-Supervised Learning (SSL) has become a preferred paradigm in many deep learning tasks, which reduces the need for human labor. Previous studies primarily focus on effectively utilising the labelled and unlabeled data to improve performance. However, we observe that how to select samples for labelling also significantly impacts performance, particularly under extremely low-budget settings. The sample selection task in SSL has been under-explored for a long time. To fill in this gap, we propose a Representative and Diverse Sample Selection approach (RDSS). By adopting a modified Frank-Wolfe algorithm to minimise a novel criterion $α$-Maximum Mean Discrepancy ($α$-MMD), RDSS samples a representative and diverse subset for annotation from the unlabeled data. We demonstrate that minimizing $α$-MMD enhances the generalization ability of low-budget learning. Experimental results show that RDSS consistently improves the performance of several popular SSL frameworks and outperforms the state-of-the-art sample selection approaches used in Active Learning (AL) and Semi-Supervised Active Learning (SSAL), even with constrained annotation budgets.

Jiahe Chen, Jiaying He, Qiyuan Chen et al.

Large Vision-Language Models (LVLMs) face a tug-of-war between powerful linguistic priors and visual evidence, often leading to ``semantic drift'' -- the progressive detachment from visual input that we identify as the root cause of hallucination. While several existing training-free decoding strategies have achieved considerable success, they still suffer from inherent limitations. Many are computationally prohibitive, requiring multiple forward passes through the entire LVLM, while others rely on indirect, heuristic-based proxies that are unreliable correlates for a direct semantic conflict. We propose \textbf{D}ynamic \textbf{L}ogits \textbf{C}alibration (DLC), a novel training-free framework that is the first to cure semantic drift in a direct, dynamic, and efficient manner. At each decoding step, DLC introduces a real-time visual referee that performs a dual-aspect visual alignment check: (1) it assesses the intrinsic visual relevance of a candidate token and (2) its contextual visual coherence. By dynamically balancing these two checks and evaluating them against an adaptive baseline, DLC surgically modulates the output logits to favor grounded tokens. Extensive experiments show DLC significantly outperforms existing methods in mitigating hallucinations while, crucially, maintaining high inference efficiency by avoiding costly multiple LVLM forward passes. Our work presents a powerful and practical solution for building more reliable and visually-grounded LVLMs. Code will be released on https://github.com/JiaheChen2002/DLC.

Qiyuan Chen, Jiahe Chen, Hongsen Huang et al.

The paradigm shift from traditional ranked-based search to Generative Search Engines has rendered conventional SEO metrics obsolete, creating an urgent need to understand, measure, and optimize for content influence on synthesized answers. This paper introduces a comprehensive, end-to-end framework for Generative Search Engine Optimization (GSEO) to address this challenge. We make two primary contributions. First, we construct CC-GSEO-Bench, a large-scale, content-centric benchmark, and propose a multi-dimensional evaluation framework that systematically quantifies influence, moving beyond surface-level attribution to assess substantive semantic impact. Second, we design a novel multi-agent system that operationalizes this framework, automating the strategic refinement of content through a collaborative analyze-revise-evaluate workflow. Our empirical analysis using this framework reveals novel insights into the dynamics of content influence, offering actionable strategies for creators and establishing a principled foundation for future GSEO research.

Qian Shao, Pradeep Varakantham, Shih-Fen Cheng

Complex planning and scheduling problems have long been solved using various optimization or heuristic approaches. In recent years, imitation learning that aims to learn from expert demonstrations has been proposed as a viable alternative to solving these problems. Generally speaking, imitation learning is designed to learn either the reward (or preference) model or directly the behavioral policy by observing the behavior of an expert. Existing work in imitation learning and inverse reinforcement learning has focused on imitation primarily in unconstrained settings (e.g., no limit on fuel consumed by the vehicle). However, in many real-world domains, the behavior of an expert is governed not only by reward (or preference) but also by constraints. For instance, decisions on self-driving delivery vehicles are dependent not only on the route preferences/rewards (depending on past demand data) but also on the fuel in the vehicle and the time available. In such problems, imitation learning is challenging as decisions are not only dictated by the reward model but are also dependent on a cost-constrained model. In this paper, we provide multiple methods that match expert distributions in the presence of trajectory cost constraints through (a) Lagrangian-based method; (b) Meta-gradients to find a good trade-off between expected return and minimizing constraint violation; and (c) Cost-violation-based alternating gradient. We empirically show that leading imitation learning approaches imitate cost-constrained behaviors poorly and our meta-gradient-based approach achieves the best performance.

Jiashu Pu, Yajing Wan, Yuru Zhang et al.

Previous in-context learning (ICL) research has focused on tasks such as classification, machine translation, text2table, etc., while studies on whether ICL can improve human-like dialogue generation are scarce. Our work fills this gap by systematically investigating the ICL capabilities of large language models (LLMs) in persona-based dialogue generation, conducting extensive experiments on high-quality real human Chinese dialogue datasets. From experimental results, we draw three conclusions: 1) adjusting prompt instructions is the most direct, effective, and economical way to improve generation quality; 2) randomly retrieving demonstrations (demos) achieves the best results, possibly due to the greater diversity and the amount of effective information; counter-intuitively, retrieving demos with a context identical to the query performs the worst; 3) even when we destroy the multi-turn associations and single-turn semantics in the demos, increasing the number of demos still improves dialogue performance, proving that LLMs can learn from corrupted dialogue demos. Previous explanations of the ICL mechanism, such as $n$-gram induction head, cannot fully account for this phenomenon.

Qiyuan Chen, Hongsen Huang, Qian Shao et al.

Large Language Models (LLMs) require high quality preference datasets to align with human preferences. However, conventional methods for constructing such datasets face significant challenges: reliance on pre-collected instructions often leads to distribution mismatches with target models, while the need for sampling multiple stochastic responses introduces substantial computational overhead. In this work, we explore a paradigm shift by leveraging inherent regulation of LLMs' representation space for efficient and tailored preference dataset construction, named Icon$^{2}$. Specifically, it first extracts layer-wise direction vectors to encode sophisticated human preferences and then uses these vectors to filter self-synthesized instructions based on their inherent consistency. During decoding, bidirectional inherent control is applied to steer token representations, enabling the precise generation of response pairs with clear alignment distinctions. Experimental results demonstrate significant improvements in both alignment and efficiency. Llama3-8B and Qwen2-7B achieve an average win rate improvement of 13.89% on AlpacaEval 2.0 and 13.45% on Arena-Hard, while reducing computational costs by up to 48.1%.

Ling Cheng, Jiashu Pu, Ruicheng Liang et al.

Semi-supervised community detection methods are widely used for identifying specific communities due to the label scarcity. Existing semi-supervised community detection methods typically involve two learning stages learning in both initial identification and subsequent adjustment, which often starts from an unreasonable community core candidate. Moreover, these methods encounter scalability issues because they depend on reinforcement learning and generative adversarial networks, leading to higher computational costs and restricting the selection of candidates. To address these limitations, we draw a parallel between crystallization kinetics and community detection to integrate the spontaneity of the annealing process into community detection. Specifically, we liken community detection to identifying a crystal subgrain (core) that expands into a complete grain (community) through a process similar to annealing. Based on this finding, we propose CLique ANNealing (CLANN), which applies kinetics concepts to community detection by integrating these principles into the optimization process to strengthen the consistency of the community core. Subsequently, a learning-free Transitive Annealer was employed to refine the first-stage candidates by merging neighboring cliques and repositioning the community core, enabling a spontaneous growth process that enhances scalability. Extensive experiments on \textbf{43} different network settings demonstrate that CLANN outperforms state-of-the-art methods across multiple real-world datasets, showcasing its exceptional efficacy and efficiency in community detection.

Qian Shao, Bang Du, Zepeng Li et al.

Clinical trials remain critical in cardiac drug development but face high failure rates due to efficacy limitations and safety risks, incurring substantial costs. In-silico trial methodologies, particularly generative models simulating drug-induced electrocardiogram (ECG) alterations, offer a potential solution to mitigate these challenges. While existing models show progress in ECG synthesis, their constrained fidelity and inability to characterize individual-specific pharmacological response patterns fundamentally limit clinical translatability. To address these issues, we propose a novel Drug-Aware Diffusion Model (DADM). Specifically, we construct a set of ordinary differential equations to provide external physical knowledge (EPK) of the realistic ECG morphology. The EPK is used to adaptively constrain the morphology of the generated ECGs through a dynamic cross-attention (DCA) mechanism. Furthermore, we propose an extension of ControlNet to incorporate demographic and drug data, simulating individual drug reactions. Compared to the other eight state-of-the-art (SOTA) ECG generative models: 1) Quantitative and expert evaluation demonstrate that DADM generates ECGs with superior fidelity; 2) Comparative results on two real-world databases covering 8 types of drug regimens verify that DADM can more accurately simulate drug-induced changes in ECGs, improving the accuracy by at least 5.79% and recall by 8%. In addition, the ECGs generated by DADM can also enhance model performance in downstream drug-effect classification tasks.

Qian Shao, Bang Du, Kai Zhang et al.

Deep learning models for pulmonary disease screening from Computed Tomography (CT) scans promise to alleviate the immense workload on radiologists. Still, their high computational cost, stemming from processing entire 3D volumes, remains a major barrier to widespread clinical adoption. Current sub-sampling techniques often compromise diagnostic integrity by introducing artifacts or discarding critical information. To overcome these limitations, we propose an Efficient and Reliable Framework (ERF) that fundamentally improves the practicality of automated CT analysis. Our framework introduces two core innovations: (1) A Cluster-based Sub-Sampling (CSS) method that efficiently selects a compact yet comprehensive subset of CT slices by optimizing for both representativeness and diversity. By integrating an efficient k-Nearest Neighbor (k-NN) search with an iterative refinement process, CSS bypasses the computational bottlenecks of previous methods while preserving vital diagnostic features. (2) A lightweight Hybrid Uncertainty Quantification (HUQ) mechanism, which uniquely assesses both Aleatoric Uncertainty (AU) and Epistemic Uncertainty (EU) with minimal computational overhead. By maximizing the discrepancy between auxiliary classifiers, HUQ provides a robust reliability score, which is crucial for building trust in automated systems operating on partial data. Validated on two public datasets with 2,654 CT volumes across diagnostic tasks for 3 pulmonary diseases, our proposed ERF achieves diagnostic performance comparable to the full-volume analysis (over 90% accuracy and recall) while reducing processing time by more than 60%. This work represents a significant step towards deploying fast, accurate, and trustworthy AI-powered screening tools in time-sensitive clinical settings.

Qian Shao, Ye Dai, Haochao Ying et al.

Uveitis demands the precise diagnosis of anterior chamber inflammation (ACI) for optimal treatment. However, current diagnostic methods only rely on a limited single-modal disease perspective, which leads to poor performance. In this paper, we investigate a promising yet challenging way to fuse multimodal data for ACI diagnosis. Notably, existing fusion paradigms focus on empowering implicit modality interactions (i.e., self-attention and its variants), but neglect to inject explicit modality interactions, especially from clinical knowledge and imaging property. To this end, we propose a jointly Explicit and implicit Cross-Modal Interaction Network (EiCI-Net) for Anterior Chamber Inflammation Diagnosis that uses anterior segment optical coherence tomography (AS-OCT) images, slit-lamp images, and clinical data jointly. Specifically, we first develop CNN-Based Encoders and Tabular Processing Module (TPM) to extract efficient feature representations in different modalities. Then, we devise an Explicit Cross-Modal Interaction Module (ECIM) to generate attention maps as a kind of explicit clinical knowledge based on the tabular feature maps, then integrated them into the slit-lamp feature maps, allowing the CNN-Based Encoder to focus on more effective informativeness of the slit-lamp images. After that, the Implicit Cross-Modal Interaction Module (ICIM), a transformer-based network, further implicitly enhances modality interactions. Finally, we construct a considerable real-world dataset from our collaborative hospital and conduct sufficient experiments to demonstrate the superior performance of our proposed EiCI-Net compared with the state-of-the-art classification methods in various metrics.