Yuxuan Han

Yuxuan Han, Ruicheng Wang, Jiaolong Yang

This paper deals with the challenging task of synthesizing novel views for in-the-wild photographs. Existing methods have shown promising results leveraging monocular depth estimation and color inpainting with layered depth representations. However, these methods still have limited capability to handle scenes with complex 3D geometry. We propose a new method based on the multiplane image (MPI) representation. To accommodate diverse scene layouts in the wild and tackle the difficulty in producing high-dimensional MPI contents, we design a network structure that consists of two novel modules, one for plane depth adjustment and another for depth-aware color prediction. The former adjusts the initial plane positions using the RGBD context feature and an attention mechanism. Given adjusted depth values, the latter predicts the color and density for each plane separately with proper inter-plane interactions achieved via a feature masking strategy. To train our method, we construct large-scale stereo training data using only unconstrained single-view image collections by a simple yet effective warp-back strategy. The experiments on both synthetic and real datasets demonstrate that our trained model works remarkably well and achieves state-of-the-art results.

Bingzhe Wu, Zhipeng Liang, Yuxuan Han et al.

Recently, federated learning has emerged as a promising approach for training a global model using data from multiple organizations without leaking their raw data. Nevertheless, directly applying federated learning to real-world tasks faces two challenges: (1) heterogeneity in the data among different organizations; and (2) data noises inside individual organizations. In this paper, we propose a general framework to solve the above two challenges simultaneously. Specifically, we propose using distributionally robust optimization to mitigate the negative effects caused by data heterogeneity paradigm to sample clients based on a learnable distribution at each iteration. Additionally, we observe that this optimization paradigm is easily affected by data noises inside local clients, which has a significant performance degradation in terms of global model prediction accuracy. To solve this problem, we propose to incorporate mixup techniques into the local training process of federated learning. We further provide comprehensive theoretical analysis including robustness analysis, convergence analysis, and generalization ability. Furthermore, we conduct empirical studies across different drug discovery tasks, such as ADMET property prediction and drug-target affinity prediction.

Xiuzhong Hu, Guangming Xiong, Zheng Zang et al.

Reconstructing large-scale 3D scenes is essential for autonomous vehicles, especially when partial sensor data is lost. Although the recently developed neural radiance fields (NeRF) have shown compelling results in implicit representations, the large-scale 3D scene reconstruction using partially lost LiDAR point cloud data still needs to be explored. To bridge this gap, we propose a novel 3D scene reconstruction framework called parent-child neural radiance field (PC-NeRF). The framework comprises two modules, the parent NeRF and the child NeRF, to simultaneously optimize scene-level, segment-level, and point-level scene representations. Sensor data can be utilized more efficiently by leveraging the segment-level representation capabilities of child NeRFs, and an approximate volumetric representation of the scene can be quickly obtained even with limited observations. With extensive experiments, our proposed PC-NeRF is proven to achieve high-precision 3D reconstruction in large-scale scenes. Moreover, PC-NeRF can effectively tackle situations where partial sensor data is lost and has high deployment efficiency with limited training time. Our approach implementation and the pre-trained models will be available at https://github.com/biter0088/pc-nerf.

Libo Zhang, Yuxuan Han, Wenbin Lin et al.

We present PRTGaussian, a realtime relightable novel-view synthesis method made possible by combining 3D Gaussians and Precomputed Radiance Transfer (PRT). By fitting relightable Gaussians to multi-view OLAT data, our method enables real-time, free-viewpoint relighting. By estimating the radiance transfer based on high-order spherical harmonics, we achieve a balance between capturing detailed relighting effects and maintaining computational efficiency. We utilize a two-stage process: in the first stage, we reconstruct a coarse geometry of the object from multi-view images. In the second stage, we initialize 3D Gaussians with the obtained point cloud, then simultaneously refine the coarse geometry and learn the light transport for each Gaussian. Extensive experiments on synthetic datasets show that our approach can achieve fast and high-quality relighting for general objects. Code and data are available at https://github.com/zhanglbthu/PRTGaussian.

Yuxuan Han, Jialin Zeng, Yang Wang et al.

We study the stochastic contextual bandit with knapsacks (CBwK) problem, where each action, taken upon a context, not only leads to a random reward but also costs a random resource consumption in a vector form. The challenge is to maximize the total reward without violating the budget for each resource. We study this problem under a general realizability setting where the expected reward and expected cost are functions of contexts and actions in some given general function classes $\mathcal{F}$ and $\mathcal{G}$, respectively. Existing works on CBwK are restricted to the linear function class since they use UCB-type algorithms, which heavily rely on the linear form and thus are difficult to extend to general function classes. Motivated by online regression oracles that have been successfully applied to contextual bandits, we propose the first universal and optimal algorithmic framework for CBwK by reducing it to online regression. We also establish the lower regret bound to show the optimality of our algorithm for a variety of function classes.

Yuxuan Han, Zhibo Wang, Feng Xu

Modeling non-Lambertian effects such as facial specularity leads to a more realistic 3D Morphable Face Model. Existing works build parametric models for diffuse and specular albedo using Light Stage data. However, only diffuse and specular albedo cannot determine the full BRDF. In addition, the requirement of Light Stage data is hard to fulfill for the research communities. This paper proposes the first 3D morphable face reflectance model with spatially varying BRDF using only low-cost publicly-available data. We apply linear shiness weighting into parametric modeling to represent spatially varying specular intensity and shiness. Then an inverse rendering algorithm is developed to reconstruct the reflectance parameters from non-Light Stage data, which are used to train an initial morphable reflectance model. To enhance the model's generalization capability and expressive power, we further propose an update-by-reconstruction strategy to finetune it on an in-the-wild dataset. Experimental results show that our method obtains decent rendering results with plausible facial specularities. Our code is released \href{https://yxuhan.github.io/ReflectanceMM/index.html}{\textcolor{magenta}{here}}.

Yuxuan Han, Zhicong Liang, Zhipeng Liang et al.

Differentially private (DP) stochastic convex optimization (SCO) is ubiquitous in trustworthy machine learning algorithm design. This paper studies the DP-SCO problem with streaming data sampled from a distribution and arrives sequentially. We also consider the continual release model where parameters related to private information are updated and released upon each new data, often known as the online algorithms. Despite that numerous algorithms have been developed to achieve the optimal excess risks in different $\ell_p$ norm geometries, yet none of the existing ones can be adapted to the streaming and continual release setting. To address such a challenge as the online convex optimization with privacy protection, we propose a private variant of online Frank-Wolfe algorithm with recursive gradients for variance reduction to update and reveal the parameters upon each data. Combined with the adaptive differential privacy analysis, our online algorithm achieves in linear time the optimal excess risk when $1<p\leq 2$ and the state-of-the-art excess risk meeting the non-private lower ones when $2<p\leq\infty$. Our algorithm can also be extended to the case $p=1$ to achieve nearly dimension-independent excess risk. While previous variance reduction results on recursive gradient have theoretical guarantee only in the independent and identically distributed sample setting, we establish such a guarantee in a non-stationary setting. To demonstrate the virtues of our method, we design the first DP algorithm for high-dimensional generalized linear bandits with logarithmic regret. Comparative experiments with a variety of DP-SCO and DP-Bandit algorithms exhibit the efficacy and utility of the proposed algorithms.

Shiyin Lu, Yang Li, Yu Xia et al.

We present Ovis2.5, a successor to Ovis2 designed for native-resolution visual perception and strong multimodal reasoning. Ovis2.5 integrates a native-resolution vision transformer that processes images at their native, variable resolutions, avoiding the degradation from fixed-resolution tiling and preserving both fine detail and global layout -- crucial for visually dense content like complex charts. To strengthen reasoning, we train the model to move beyond linear chain-of-thought and perform reflection -- including self-checking and revision. This advanced capability is exposed as an optional "thinking mode" at inference time, allowing users to trade latency for enhanced accuracy on difficult inputs. The model is trained via a comprehensive five-phase curriculum that progressively builds its skills. The process begins with foundational visual and multimodal pretraining, advances through large-scale instruction tuning, and culminates in alignment and reasoning enhancement using DPO and GRPO. To scale these upgrades efficiently, we employ multimodal data packing and hybrid parallelism, yielding a significant end-to-end speedup. We release two open-source models: Ovis2.5-9B and Ovis2.5-2B. The latter continues the "small model, big performance" philosophy of Ovis2, making it ideal for resource-constrained, on-device scenarios. On the OpenCompass multimodal leaderboard, Ovis2.5-9B averages 78.3, marking a substantial improvement over its predecessor, Ovis2-8B, and achieving state-of-the-art results among open-source MLLMs in the sub-40B parameter range; Ovis2.5-2B scores 73.9, establishing SOTA for its size. Beyond aggregate scores, Ovis2.5 achieves leading results on STEM benchmarks, exhibits strong capabilities on grounding and video tasks, and achieves open-source SOTA at its scale for complex chart analysis.

Yuxuan Han, Xin Ming, Tianxiao Li et al.

Existing methods achieve high-quality facial appearance capture under controllable lighting, which increases capture cost and limits usability. We propose WildCap, a novel method for high-quality facial appearance capture from a smartphone video recorded in the wild. To disentangle high-quality reflectance from complex lighting effects in in-the-wild captures, we propose a novel hybrid inverse rendering framework. Specifically, we first apply a data-driven method, i.e., SwitchLight, to convert the captured images into more constrained conditions and then adopt model-based inverse rendering. However, unavoidable local artifacts in network predictions, such as shadow-baking, are non-physical and thus hinder accurate inverse rendering of lighting and material. To address this, we propose a novel texel grid lighting model to explain non-physical effects as clean albedo illuminated by local physical lighting. During optimization, we jointly sample a diffusion prior for reflectance maps and optimize the lighting, effectively resolving scale ambiguity between local lights and albedo. Our method achieves significantly better results than prior arts in the same capture setup, closing the quality gap between in-the-wild and controllable recordings by a large margin. Our code will be released \href{https://yxuhan.github.io/WildCap/index.html}{\textcolor{magenta}{here}}.

Xiuzhong Hu, Guangming Xiong, Zheng Zang et al.

Large-scale 3D scene reconstruction and novel view synthesis are vital for autonomous vehicles, especially utilizing temporally sparse LiDAR frames. However, conventional explicit representations remain a significant bottleneck towards representing the reconstructed and synthetic scenes at unlimited resolution. Although the recently developed neural radiance fields (NeRF) have shown compelling results in implicit representations, the problem of large-scale 3D scene reconstruction and novel view synthesis using sparse LiDAR frames remains unexplored. To bridge this gap, we propose a 3D scene reconstruction and novel view synthesis framework called parent-child neural radiance field (PC-NeRF). Based on its two modules, parent NeRF and child NeRF, the framework implements hierarchical spatial partitioning and multi-level scene representation, including scene, segment, and point levels. The multi-level scene representation enhances the efficient utilization of sparse LiDAR point cloud data and enables the rapid acquisition of an approximate volumetric scene representation. With extensive experiments, PC-NeRF is proven to achieve high-precision novel LiDAR view synthesis and 3D reconstruction in large-scale scenes. Moreover, PC-NeRF can effectively handle situations with sparse LiDAR frames and demonstrate high deployment efficiency with limited training epochs. Our approach implementation and the pre-trained models are available at https://github.com/biter0088/pc-nerf.

Zichuan Lin, Xiaokai Huang, Jiate Liu et al.

The estimation of individual treatment effects (ITE) focuses on predicting the outcome changes that result from a change in treatment. A fundamental challenge in observational data is that while we need to infer outcome differences under alternative treatments, we can only observe each individual's outcome under a single treatment. Existing approaches address this limitation either by training with inferred pseudo-outcomes or by creating matched instance pairs. However, recent work has largely overlooked the potential impact of post-treatment variables on the outcome. This oversight prevents existing methods from fully capturing outcome variability, resulting in increased variance in counterfactual predictions. This paper introduces Pseudo-outcome Imputation with Post-treatment Variables for Counterfactual Regression (PIPCFR), a novel approach that incorporates post-treatment variables to improve pseudo-outcome imputation. We analyze the challenges inherent in utilizing post-treatment variables and establish a novel theoretical bound for ITE risk that explicitly connects post-treatment variables to ITE estimation accuracy. Unlike existing methods that ignore these variables or impose restrictive assumptions, PIPCFR learns effective representations that preserve informative components while mitigating bias. Empirical evaluations on both real-world and simulated datasets demonstrate that PIPCFR achieves significantly lower ITE errors compared to existing methods.

Yuxuan Han, Kunyuan Wu, Qianyi Shao et al.

End-to-end autonomous driving has emerged as a promising paradigm integrating perception, decision-making, and control within a unified learning framework. Recently, Vision-Language Models (VLMs) have gained significant attention for their potential to enhance the robustness and generalization of end-to-end driving models in diverse and unseen scenarios. However, existing VLM-based approaches still face challenges, including suboptimal lane perception, language understanding biases, and difficulties in handling corner cases. To address these issues, we propose AppleVLM, an advanced perception and planning-enhanced VLM model for robust end-to-end driving. AppleVLM introduces a novel vision encoder and a planning strategy encoder to improve perception and decision-making. Firstly, the vision encoder fuses spatial-temporal information from multi-view images across multiple timesteps using a deformable transformer mechanism, enhancing robustness to camera variations and facilitating scalable deployment across different vehicle platforms. Secondly, unlike traditional VLM-based approaches, AppleVLM introduces a dedicated planning modality that encodes explicit Bird's-Eye-View spatial information, mitigating language biases in navigation instructions. Finally, a VLM decoder fine-tuned by a hierarchical Chain-of-Thought integrates vision, language, and planning features to output robust driving waypoints. We evaluate AppleVLM in closed-loop experiments on two CARLA benchmarks, achieving state-of-the-art driving performance. Furthermore, we deploy AppleVLM on an AGV platform and successfully showcase real-world end-to-end autonomous driving in complex outdoor environments.

Yingying Fan, Yuxuan Han, Jinchi Lv et al.

We study contextual dynamic pricing in a semiparametric scalar-index valuation model where the latent value is $v_t=μ_\ast(\mathsf c_t)+ξ_t$, with an unknown utility map $μ_\ast$ and an unknown additive noise distribution. The key decision object is the one-dimensional oracle price map $u\mapsto p^\ast(u)$ induced by the scalar index $u=μ_\ast(\mathsf c)$ and the noise tail. Under the $β$-Hölder smoothness of the tail function for $β\geq 2$ and a revenue-geometry condition that gives a unique, stable, interior maximizer, this oracle map is itself $(β-1)$-smooth. We exploit such structure through $\mathsf{ORBIT}$, a modular coarse-to-fine policy that takes a scalar pilot index as input, localizes a benchmark price in each active bin, and learns a local polynomial approximation of the oracle map inside a trust region via bandit convex optimization. For the baseline linear utility model $μ_\ast(\mathsf c)=\mathsf c^\topθ_\ast$, an adaptive elliptical exploration scheme constructs the required scalar pilot online without distributional assumptions on the contexts. The resulting policy achieves regret $\widetilde{O}\big(T^{\frac{2β-1}{4β-3}}+\sqrt{dT}\big)$. For fixed $d$, we establish a matching lower bound in the horizon dependence, unveiling that the nonparametric oracle-map learning term is minimax sharp. The same scalar-pilot interface also yields extensions to sparse high-dimensional linear utility and nonparametric Hölder utility.

Haijun Li, Tianqi Shi, Zifu Shang et al.

Machine translation (MT) has become indispensable for cross-border communication in globalized industries like e-commerce, finance, and legal services, with recent advancements in large language models (LLMs) significantly enhancing translation quality. However, applying general-purpose MT models to industrial scenarios reveals critical limitations due to domain-specific terminology, cultural nuances, and stylistic conventions absent in generic benchmarks. Existing evaluation frameworks inadequately assess performance in specialized contexts, creating a gap between academic benchmarks and real-world efficacy. To address this, we propose a three-level translation capability framework: (1) Basic Linguistic Competence, (2) Domain-Specific Proficiency, and (3) Cultural Adaptation, emphasizing the need for holistic evaluation across these dimensions. We introduce TransBench, a benchmark tailored for industrial MT, initially targeting international e-commerce with 17,000 professionally translated sentences spanning 4 main scenarios and 33 language pairs. TransBench integrates traditional metrics (BLEU, TER) with Marco-MOS, a domain-specific evaluation model, and provides guidelines for reproducible benchmark construction. Our contributions include: (1) a structured framework for industrial MT evaluation, (2) the first publicly available benchmark for e-commerce translation, (3) novel metrics probing multi-level translation quality, and (4) open-sourced evaluation tools. This work bridges the evaluation gap, enabling researchers and practitioners to systematically assess and enhance MT systems for industry-specific needs.

Miao Lu, Yuxuan Han, Han Zhong et al.

Assortment optimization is a fundamental challenge in modern retail and recommendation systems, where the goal is to select a subset of products that maximizes expected revenue under complex customer choice behaviors. While recent advances in data-driven methods have leveraged historical data to learn and optimize assortments, these approaches typically rely on strong assumptions -- namely, the stability of customer preferences and the correctness of the underlying choice models. However, such assumptions frequently break in real-world scenarios due to preference shifts and model misspecification, leading to poor generalization and revenue loss. Motivated by this limitation, we propose a robust framework for data-driven assortment optimization that accounts for potential distributional shifts in customer choice behavior. Our approach models potential preference shift from a nominal choice model that generates data and seeks to maximize worst-case expected revenue. We first establish the computational tractability of robust assortment planning when the nominal model is known, then advance to the data-driven setting, where we design statistically optimal algorithms that minimize the data requirements while maintaining robustness. Our theoretical analysis provides both upper bounds and matching lower bounds on the sample complexity, offering theoretical guarantees for robust generalization. Notably, we uncover and identify the notion of ``robust item-wise coverage'' as the minimal data requirement to enable sample-efficient robust assortment learning. Our work bridges the gap between robustness and statistical efficiency in assortment learning, contributing new insights and tools for reliable assortment optimization under uncertainty.

Xin Ming, Yuxuan Han, Tianyu Huang et al.

Reconstructing topologically consistent facial geometry is crucial for the digital avatar creation pipelines. Existing methods either require tedious manual efforts, lack generalization to in-the-wild data, or are constrained by the limited expressiveness of 3D Morphable Models. To address these limitations, we propose VGGTFace, an automatic approach that innovatively applies the 3D foundation model, i.e. VGGT, for topologically consistent facial geometry reconstruction from in-the-wild multi-view images captured by everyday users. Our key insight is that, by leveraging VGGT, our method naturally inherits strong generalization ability and expressive power from its large-scale training and point map representation. However, it is unclear how to reconstruct a topologically consistent mesh from VGGT, as the topology information is missing in its prediction. To this end, we augment VGGT with Pixel3DMM for injecting topology information via pixel-aligned UV values. In this manner, we convert the pixel-aligned point map of VGGT to a point cloud with topology. Tailored to this point cloud with known topology, we propose a novel Topology-Aware Bundle Adjustment strategy to fuse them, where we construct a Laplacian energy for the Bundle Adjustment objective. Our method achieves high-quality reconstruction in 10 seconds for 16 views on a single NVIDIA RTX 4090. Experiments demonstrate state-of-the-art results on benchmarks and impressive generalization to in-the-wild data. Code is available at https://github.com/grignarder/vggtface.

Geng Zhang, Yuxuan Han, Yuxuan Lou et al.

Mixture-of-Experts (MoE) enables efficient scaling of large language models by activating only a subset of experts per input token. However, deploying MoE-based models incurs significant memory overhead due to the need to retain all experts in memory. While structured pruning is promising to reduce memory costs, existing methods often show suboptimal performance and unstable degradation in three dimensions: model architectures, calibration data sources, and calibration sample sizes. This paper proposes Mixture-of-Novices-and-Experts (MoNE), a novel expert pruning method that replaces redundant experts with lightweight novices to achieve effective and robust model compression. MoNE evaluates expert redundancy based on two metrics: access frequency and output variance. Experts exhibiting low usage and stable outputs are pruned and replaced with lightweight novices-unbiased estimations of their original outputs-minimizing performance degradation. Extensive experiments demonstrate that MoNE consistently outperforms baseline methods with minimal accuracy degradation across the three dimensions, confirming its effectiveness and robustness. Notably, it improves the average zero shot accuracy across nine downstream tasks by up to 2.71 under 25\% pruning ratio and 3.61 under 50\% pruning. The code is available at https://github.com/zxgx/mode-pd.

Yuxuan Han, Junfeng Lyu, Kuan Sheng et al.

Existing facial appearance capture methods can reconstruct plausible facial reflectance from smartphone-recorded videos. However, the reconstruction quality is still far behind the ones based on studio recordings. This paper fills the gap by developing a novel daily-used solution with a co-located smartphone and flashlight video capture setting in a dim room. To enhance the quality, our key observation is to solve facial reflectance maps within the data distribution of studio-scanned ones. Specifically, we first learn a diffusion prior over the Light Stage scans and then steer it to produce the reflectance map that best matches the captured images. We propose to train the diffusion prior at the patch level to improve generalization ability and training stability, as current Light Stage datasets are in ultra-high resolution but limited in data size. Tailored to this prior, we propose a patch-level posterior sampling technique to sample seamless full-resolution reflectance maps from this patch-level diffusion model. Experiments demonstrate our method closes the quality gap between low-cost and studio recordings by a large margin, opening the door for everyday users to clone themselves to the digital world. Our code will be released at https://github.com/yxuhan/DoRA.

Yuxuan Han, Jiaolong Yang, Ying Fu

Recent works have shown that a rich set of semantic directions exist in the latent space of Generative Adversarial Networks (GANs), which enables various facial attribute editing applications. However, existing methods may suffer poor attribute variation disentanglement, leading to unwanted change of other attributes when altering the desired one. The semantic directions used by existing methods are at attribute level, which are difficult to model complex attribute correlations, especially in the presence of attribute distribution bias in GAN's training set. In this paper, we propose a novel framework (IALS) that performs Instance-Aware Latent-Space Search to find semantic directions for disentangled attribute editing. The instance information is injected by leveraging the supervision from a set of attribute classifiers evaluated on the input images. We further propose a Disentanglement-Transformation (DT) metric to quantify the attribute transformation and disentanglement efficacy and find the optimal control factor between attribute-level and instance-specific directions based on it. Experimental results on both GAN-generated and real-world images collectively show that our method outperforms state-of-the-art methods proposed recently by a wide margin. Code is available at https://github.com/yxuhan/IALS.

Yuxuan Han, Xin Ming, Tianxiao Li et al.

High-quality facial appearance capture has traditionally required costly studio recording. Recent works consider an in-the-wild smartphone-based setup; however, their model-based inverse rendering paradigm struggles with the complex disentanglement of reflectance from unknown illumination. To bridge this gap, we propose to shift the paradigm into training a powerful delighting network as a prior to constrain the optimization. We leverage the OLAT dataset and the rendered Light Stage scans for training, and propose Dataset Latent Modulation (DLM) to seamlessly integrate these heterogeneous data sources. Specifically, by conditioning the core network on learnable source-aware tokens, we decouple dataset-specific styles from physical delighting principles, enabling the emergence of a delighting prior that outperforms existing proprietary models. This powerful delighting prior enables a simple and automatic appearance capture pipeline that achieves high-quality reflectance estimation from casual video inputs, outperforming prior arts by a large margin. Furthermore, we leverage our appearance capture method to transform the multi-view NeRSemble dataset into NeRSemble-Scan, a large-scale collection of 4K-resolution relightable scans. By open-sourcing our model and the NeRSemble-Scan dataset, we democratize high-end facial capture and provide a new foundation for the research community to build photorealistic digital humans.

Renxiang Xiao, Yichen Chen, Yuanfan Zhang et al.

Long-term autonomy requires robust navigation in environments subject to dynamic and static changes, as well as adverse weather conditions. Teach-and-Repeat (T\&R) navigation offers a reliable and cost-effective solution by avoiding the need for consistent global mapping; however, existing T\&R systems lack a systematic solution to tackle various environmental variations such as weather degradation, ephemeral dynamics, and structural changes. This work proposes LTR$^2$, the first cross-modal, cross-platform LiDAR-Teach-and-Radar-Repeat system that systematically addresses these challenges. LTR$^2$ leverages LiDAR during the teaching phase to capture precise structural information under normal conditions and utilizes 4D millimeter-wave radar during the repeating phase for robust operation under environmental degradations. To align sparse and noisy forward-looking 4D radar with dense and accurate omnidirectional 3D LiDAR data, we introduce a Cross-Modal Registration (CMR) network that jointly exploits Doppler-based motion priors and the physical laws governing LiDAR intensity and radar power density. Furthermore, we propose an adaptive fine-tuning strategy that incrementally updates the CMR network based on localization errors, enabling long-term adaptability to static environmental changes without ground-truth labels. We demonstrate that the proposed CMR network achieves state-of-the-art cross-modal registration performance on the open-access dataset. Then we validate LTR$^2$ across three robot platforms over a large-scale, long-term deployment (40+ km over 6 months), including challenging conditions such as nighttime smoke. Experimental results and ablation studies demonstrate centimeter-level accuracy and strong robustness against diverse environmental disturbances, significantly outperforming existing approaches.

Yuxuan Han, Yuanxing Zhang, Yushuo Wang et al.

Structured information extraction from long, multilingual scanned financial documents is a core requirement in industrial KYC and compliance workflows. These documents are typically non machine readable, noisy, and visually heterogeneous. They usually span dozens of pages while containing only sparse task relevant information. Although recent vision-language models achieve strong benchmark performance, directly applying them end to end to full financial reports often leads to unreliable extraction under real world conditions. We present a multistage extraction framework that integrates image preprocessing, multilingual OCR, hybrid page-level retrieval, and compact VLM-based structured extraction. The design separates page localization from multimodal reasoning, enabling more accurate extraction from complex multipage documents. We evaluated the framework on 120 production KYC documents comprising about 3000 multilingual scanned pages. Across multiple OCR-VLM combinations, the proposed pipeline consistently outperforms direct PDF-to-VLM baselines, improving field-level accuracy by up to 31.9 percentage points. The best configuration, PaddleOCR with MiniCPM2.6, achieves 87.27 percent accuracy. Ablation studies show that page-level retrieval is the dominant factor in performance improvements, particularly for complex financial statements and non-English documents.

Yuxuan Han, Junfeng Lyu, Feng Xu

Facial geometry and appearance capture have demonstrated tremendous success in 3D scanning real humans in studios. Recent works propose to democratize this technique while keeping the results high quality. However, they are still inconvenient for daily usage. In addition, they focus on an easier problem of only capturing facial skin. This paper proposes a novel method for high-quality face capture, featuring an easy-to-use system and the capability to model the complete face with skin, mouth interior, hair, and eyes. We reconstruct facial geometry and appearance from a single co-located smartphone flashlight sequence captured in a dim room where the flashlight is the dominant light source (e.g. rooms with curtains or at night). To model the complete face, we propose a novel hybrid representation to effectively model both eyes and other facial regions, along with novel techniques to learn it from images. We apply a combined lighting model to compactly represent real illuminations and exploit a morphable face albedo model as a reflectance prior to disentangle diffuse and specular. Experiments show that our method can capture high-quality 3D relightable scans.

Mingxuan Cui, Duo Zhou, Yuxuan Han et al.

Deep reinforcement learning (RL) has achieved significant success, yet its application in real-world scenarios is often hindered by a lack of robustness to environmental uncertainties. To solve this challenge, some robust RL algorithms have been proposed, but most are limited to tabular settings. In this work, we propose Distributionally Robust Soft Actor-Critic (DR-SAC), a novel algorithm designed to enhance the robustness of the state-of-the-art Soft Actor-Critic (SAC) algorithm. DR-SAC aims to maximize the expected value with entropy against the worst possible transition model lying in an uncertainty set. A distributionally robust version of the soft policy iteration is derived with a convergence guarantee. For settings where nominal distributions are unknown, such as offline RL, a generative modeling approach is proposed to estimate the required nominal distributions from data. Furthermore, experimental results on a range of continuous control benchmark tasks demonstrate our algorithm achieves up to $9.8$ times the average reward of the SAC baseline under common perturbations. Additionally, compared with existing robust reinforcement learning algorithms, DR-SAC significantly improves computing efficiency and applicability to large-scale problems.

Yuxuan Han, Han Zhong, Miao Lu et al.

We study the fundamental problem of offline assortment optimization under the Multinomial Logit (MNL) model, where sellers must determine the optimal subset of the products to offer based solely on historical customer choice data. While most existing approaches to learning-based assortment optimization focus on the online learning of the optimal assortment through repeated interactions with customers, such exploration can be costly or even impractical in many real-world settings. In this paper, we consider the offline learning paradigm and investigate the minimal data requirements for efficient offline assortment optimization. To this end, we introduce Pessimistic Rank-Breaking (PRB), an algorithm that combines rank-breaking with pessimistic estimation. We prove that PRB is nearly minimax optimal by establishing the tight suboptimality upper bound and a nearly matching lower bound. This further shows that "optimal item coverage" - where each item in the optimal assortment appears sufficiently often in the historical data - is both sufficient and necessary for efficient offline learning. This significantly relaxes the previous requirement of observing the complete optimal assortment in the data. Our results provide fundamental insights into the data requirements for offline assortment optimization under the MNL model.

Junyi Fan, Yuxuan Han, Jialin Zeng et al.

Efficiently learning equilibria with large state and action spaces in general-sum Markov games while overcoming the curse of multi-agency is a challenging problem. Recent works have attempted to solve this problem by employing independent linear function classes to approximate the marginal $Q$-value for each agent. However, existing sample complexity bounds under such a framework have a suboptimal dependency on the desired accuracy $\varepsilon$ or the action space. In this work, we introduce a new algorithm, Lin-Confident-FTRL, for learning coarse correlated equilibria (CCE) with local access to the simulator, i.e., one can interact with the underlying environment on the visited states. Up to a logarithmic dependence on the size of the state space, Lin-Confident-FTRL learns $ε$-CCE with a provable optimal accuracy bound $O(ε^{-2})$ and gets rids of the linear dependency on the action space, while scaling polynomially with relevant problem parameters (such as the number of agents and time horizon). Moreover, our analysis of Linear-Confident-FTRL generalizes the virtual policy iteration technique in the single-agent local planning literature, which yields a new computationally efficient algorithm with a tighter sample complexity bound when assuming random access to the simulator.

Yuxuan Han, Meng-Hao Guo, Zhengning Liu et al.

Optimizing GPU kernels manually is a challenging and time-consuming task. With the rapid development of LLMs, automated GPU kernel optimization is gradually becoming a tangible reality. However, current LLM-driven automated optimization methods narrowly focus on machine learning applications, such as PyTorch operator optimization, while overlooking broader domains like sparse matrix operations in scientific computing. Extending to these broader applications brings new challenges for the benchmark and algorithm. Therefore, developing a general-purpose automated kernel optimization method becomes our primary focus. In this paper, we address the absence of systematic evaluation for multi-scenario settings by introducing MSKernelBench, which spans multiple scenarios, including fundamental algebraic operations, common LLM kernels, sparse matrix operators, and scientific computing routines, each supporting both FP32 and BF16 precision. Building on this benchmark, we introduce CUDAMaster, a multi-agent, hardware-aware system for kernel optimization that leverages profiling information and automatically constructs the full compilation and execution toolchain. Experimental results demonstrate that CUDAMaster achieves significant speedups across most operators, outperforming Astra by about 35%. In several cases, its performance matches or surpasses that of highly optimized, closed-source libraries such as cuBLAS. A demo showcasing the original and optimized code for each operator is available at https://hanyx2021.github.io/MSKernelBenchDemo/.

Qinxun Bai, Yuxuan Han, Wei Xu et al.

Off-policy reinforcement learning (RL) with function approximation offers an effective way to improve sample efficiency by reusing past experience. Within this setting, the actor-critic (AC) framework has achieved strong empirical success. However, both the critic and actor learning is challenging for the off-policy AC methods: first of all, in addition to the classic "deadly triad" instability of off-policy evaluation, it also suffers from a "moving target" problem, where the policy being evaluated changes continually; secondly, actor learning becomes less efficient due to the difficulty of estimating the exact off-policy policy gradient. The first challenge essentially reduces the problem to repeatedly performing off-policy evaluation for changing policies. For the second challenge, the off-policy policy gradient theorem requires a complex and often impractical algorithm to estimate an additional emphasis critic, which is typically neglected in practice, thereby reducing to the on-policy policy gradient as an approximation. In this work, we introduce a novel concept of functional critic modeling, which leads to a new AC framework that addresses both challenges for actor-critic learning under the deadly triad setting. We provide a theoretical analysis in the linear function setting, establishing the provable convergence of our framework, which, to the best of our knowledge, is the first convergent off-policy target-based AC algorithm. From a practical perspective, we further propose a carefully designed neural network architecture for the functional critic modeling and demonstrate its effectiveness through preliminary experiments on widely used RL tasks from the DeepMind Control Benchmark.

Yingying Fan, Yuxuan Han, Jinchi Lv et al.

In this paper, we study the behavior of the Upper Confidence Bound-Variance (UCB-V) algorithm for the Multi-Armed Bandit (MAB) problems, a variant of the canonical Upper Confidence Bound (UCB) algorithm that incorporates variance estimates into its decision-making process. More precisely, we provide an asymptotic characterization of the arm-pulling rates for UCB-V, extending recent results for the canonical UCB in Kalvit and Zeevi (2021) and Khamaru and Zhang (2024). In an interesting contrast to the canonical UCB, our analysis reveals that the behavior of UCB-V can exhibit instability, meaning that the arm-pulling rates may not always be asymptotically deterministic. Besides the asymptotic characterization, we also provide non-asymptotic bounds for the arm-pulling rates in the high probability regime, offering insights into the regret analysis. As an application of this high probability result, we establish that UCB-V can achieve a more refined regret bound, previously unknown even for more complicate and advanced variance-aware online decision-making algorithms.

Yuxuan Han, Zhipeng Liang, Yang Wang et al.

Contextual bandit algorithms are useful in personalized online decision-making. However, many applications such as personalized medicine and online advertising require the utilization of individual-specific information for effective learning, while user's data should remain private from the server due to privacy concerns. This motivates the introduction of local differential privacy (LDP), a stringent notion in privacy, to contextual bandits. In this paper, we design LDP algorithms for stochastic generalized linear bandits to achieve the same regret bound as in non-privacy settings. Our main idea is to develop a stochastic gradient-based estimator and update mechanism to ensure LDP. We then exploit the flexibility of stochastic gradient descent (SGD), whose theoretical guarantee for bandit problems is rarely explored, in dealing with generalized linear bandits. We also develop an estimator and update mechanism based on Ordinary Least Square (OLS) for linear bandits. Finally, we conduct experiments with both simulation and real-world datasets to demonstrate the consistently superb performance of our algorithms under LDP constraints with reasonably small parameters $(\varepsilon, δ)$ to ensure strong privacy protection.

Qing Ye, Yuxuan Han, Yanan sun et al.

Parameter updating is an important stage in parallelism-based distributed deep learning. Synchronous methods are widely used in distributed training the Deep Neural Networks (DNNs). To reduce the communication and synchronization overhead of synchronous methods, decreasing the synchronization frequency (e.g., every $n$ mini-batches) is a straightforward approach. However, it often suffers from poor convergence. In this paper, we propose a new algorithm of integrating Particle Swarm Optimization (PSO) into the distributed training process of DNNs to automatically compute new parameters. In the proposed algorithm, a computing work is encoded by a particle, the weights of DNNs and the training loss are modeled by the particle attributes. At each synchronization stage, the weights are updated by PSO from the sub weights gathered from all workers, instead of averaging the weights or the gradients. To verify the performance of the proposed algorithm, the experiments are performed on two commonly used image classification benchmarks: MNIST and CIFAR10, and compared with the peer competitors at multiple different synchronization configurations. The experimental results demonstrate the competitiveness of the proposed algorithm.