arXiv subjects

Bo Wang

Publications and source records attributed to Bo Wang.

At least 235 records · Page 13Linked to original sources

arXiv2025

EvoFormer: Learning Dynamic Graph-Level Representations with Structural and Temporal Bias Correction

Dynamic graph-level embedding aims to capture structural evolution in networks, which is essential for modeling real-world scenarios. However, existing methods face two critical yet under-explored issues: Structural Visit Bias, where random walk sampling disproportionately emphasizes high-degree nodes, leading to redundant and noisy structural representations; and Abrupt Evolution Blindness, the failure to effectively detect sudden structural changes due to rigid or overly simplistic temporal modeling strategies, resulting in inconsistent temporal embeddings. To overcome these challenges, we propose EvoFormer, an evolution-aware Transformer framework tailored for dynamic graph-level representation learning. To mitigate Structural Visit Bias, EvoFormer introduces a Structure-Aware Transformer Module that incorporates positional encoding based on node structural roles, allowing the model to globally differentiate and accurately represent node structures. To overcome Abrupt Evolution Blindness, EvoFormer employs an Evolution-Sensitive Temporal Module, which explicitly models temporal evolution through a sequential three-step strategy: (I) Random Walk Timestamp Classification, generating initial timestamp-aware graph-level embeddings; (II) Graph-Level Temporal Segmentation, partitioning the graph stream into segments reflecting structurally coherent periods; and (III) Segment-Aware Temporal Self-Attention combined with an Edge Evolution Prediction task, enabling the model to precisely capture segment boundaries and perceive structural evolution trends, effectively adapting to rapid temporal shifts. Extensive evaluations on five benchmark datasets confirm that EvoFormer achieves state-of-the-art performance in graph similarity ranking, temporal anomaly detection, and temporal segmentation tasks, validating its effectiveness in correcting structural and temporal biases.

cs.LG↗

arXiv2025

Bioinspired underwater soft robots: from biology to robotics and back

The ocean vast unexplored regions and diverse soft-bodied marine organisms have spurred interest in bio-inspired underwater soft robotics. Recent advances have enabled new capabilities in underwater movement, sensing, and interaction. However, these efforts are largely unidirectional, with biology guiding robotics while insights from robotics rarely feed back into biology. Here we propose a holistic, bidirectional framework that integrates biological principles, robotic implementation, and biological validation. We show that soft robots can serve as experimental tools to probe biological functions and even test evolutionary hypotheses. Their inherent compliance also allows them to outperform rigid systems in unstructured environments, supporting applications in marine exploration, manipulation, and medicine. Looking forward, we introduce bio-universal-inspired robotics, a paradigm that transcends species-specific mimicry by identifying convergent principles across species to inspire more adaptable designs. Despite rapid progress, challenges persist in material robustness, actuation efficiency, autonomy, and intelligence. By uniting biology and engineering, soft robots can advance ocean exploration and deepen scientific discovery.

cs.RO↗

arXiv2025

CAS-IQA: Teaching Vision-Language Models for Synthetic Angiography Quality Assessment

Synthetic X-ray angiographies generated by modern generative models hold great potential to reduce the use of contrast agents in vascular interventional procedures. However, low-quality synthetic angiographies can significantly increase procedural risk, underscoring the need for reliable image quality assessment (IQA) methods. Existing IQA models, however, fail to leverage auxiliary images as references during evaluation and lack fine-grained, task-specific metrics necessary for clinical relevance. To address these limitations, this paper proposes CAS-IQA, a vision-language model (VLM)-based framework that predicts fine-grained quality scores by effectively incorporating auxiliary information from related images. In the absence of angiography datasets, CAS-3K is constructed, comprising 3,565 synthetic angiographies along with score annotations. To ensure clinically meaningful assessment, three task-specific evaluation metrics are defined. Furthermore, a Multi-path featUre fuSion and rouTing (MUST) module is designed to enhance image representations by adaptively fusing and routing visual tokens to metric-specific branches. Extensive experiments on the CAS-3K dataset demonstrate that CAS-IQA significantly outperforms state-of-the-art IQA methods by a considerable margin.

cs.CV↗

arXiv2025

Multi-Treatment-DML: Causal Estimation for Multi-Dimensional Continuous Treatments with Monotonicity Constraints in Personal Loan Risk Optimization

Optimizing credit limits, interest rates, and loan terms is crucial for managing borrower risk and lifetime value (LTV) in personal loan platform. However, counterfactual estimation of these continuous, multi-dimensional treatments faces significant challenges: randomized trials are often prohibited by risk controls and long repayment cycles, forcing reliance on biased observational data. Existing causal methods primarily handle binary/discrete treatments and struggle with continuous, multi-dimensional settings. Furthermore, financial domain knowledge mandates provably monotonic treatment-outcome relationships (e.g., risk increases with credit limit).To address these gaps, we propose Multi-Treatment-DML, a novel framework leveraging Double Machine Learning (DML) to: (i) debias observational data for causal effect estimation; (ii) handle arbitrary-dimensional continuous treatments; and (iii) enforce monotonic constraints between treatments and outcomes, guaranteeing adherence to domain requirements.Extensive experiments on public benchmarks and real-world industrial datasets demonstrate the effectiveness of our approach. Furthermore, online A/B testing conducted on a realworld personal loan platform, confirms the practical superiority of Multi-Treatment-DML in real-world loan operations.

cs.LG↗

arXiv2025

O1NumHess: a fast and accurate seminumerical Hessian algorithm using only O(1) gradients

In this work, we describe a new algorithm, O1NumHess, to calculate the Hessian of a molecular system by finite differentiation of gradients calculated at displaced geometries. Different from the conventional seminumerical Hessian algorithm, which requires gradients at $O(N_{\mathrm{atom}})$ displaced geometries (where $N_{\mathrm{atom}}$ is the number of atoms), the present approach only requires $O(1)$ gradients. Key to the reduction of the number of gradients is the exploitation of the off-diagonal low-rank (ODLR) property of Hessians, namely the blocks of the Hessian that correspond to two distant groups of atoms have low rank. This property reduces the number of independent entries of the Hessian from $O(N_{\mathrm{atom}}^2)$ to $O(N_{\mathrm{atom}})$, such that $O(1)$ gradients already contain enough information to uniquely determine the Hessian. Numerical results on model systems (long alkanes and polyenes), transition metal reactions (WCCR10) and non-covalent complexes (S30L-CI) using the BDF program show that O1NumHess gives frequency, zero-point energy, enthalpy and Gibbs free energy errors that are only about two times those of conventional double-sided seminumerical Hessians. Moreover, O1NumHess is always faster than the conventional numerical Hessian algorithm, frequently even faster than the analytic Hessian, and requires only about 100 gradients for sufficiently large systems. An open-source implementation of this method, which can also be applied to problems irrelevant to computational chemistry, is available on GitHub.

physics.chem-ph↗

arXiv2025

ASM-UNet: Adaptive Scan Mamba Integrating Group Commonalities and Individual Variations for Fine-Grained Segmentation

Precise lesion resection depends on accurately identifying fine-grained anatomical structures. While many coarse-grained segmentation (CGS) methods have been successful in large-scale segmentation (e.g., organs), they fall short in clinical scenarios requiring fine-grained segmentation (FGS), which remains challenging due to frequent individual variations in small-scale anatomical structures. Although recent Mamba-based models have advanced medical image segmentation, they often rely on fixed manually-defined scanning orders, which limit their adaptability to individual variations in FGS. To address this, we propose ASM-UNet, a novel Mamba-based architecture for FGS. It introduces adaptive scan scores to dynamically guide the scanning order, generated by combining group-level commonalities and individual-level variations. Experiments on two public datasets (ACDC and Synapse) and a newly proposed challenging biliary tract FGS dataset, namely BTMS, demonstrate that ASM-UNet achieves superior performance in both CGS and FGS tasks. Our code and dataset are available at https://github.com/YqunYang/ASM-UNet.

cs.CV↗

arXiv2025

Extremum Seeking Control for Antenna Pointing via Symmetric Product Approximation

This paper investigates extremum seeking control for a torque-controlled antenna pointing system without direct angular measurements. We consider a two-degree-of-freedom (2-DOF) antenna system that receives an unknown signal from its environment, where the signal strength varies with the antenna orientation. It is assumed that only real-time measurements of the signal are available. We develop an extremum seeking control strategy that enables the antenna to autonomously adjust its direction to maximize the received signal strength based on the symmetric product approximation. Under suitable assumptions on the signal function, we prove local practical uniform asymptotic stability for the closed-loop system.

eess.SY↗

arXiv2025

AnomalyControl: Learning Cross-modal Semantic Features for Controllable Anomaly Synthesis

Anomaly synthesis is a crucial approach to augment abnormal data for advancing anomaly inspection. Based on the knowledge from the large-scale pre-training, existing text-to-image anomaly synthesis methods predominantly focus on textual information or coarse-aligned visual features to guide the entire generation process. However, these methods often lack sufficient descriptors to capture the complicated characteristics of realistic anomalies (e.g., the fine-grained visual pattern of anomalies), limiting the realism and generalization of the generation process. To this end, we propose a novel anomaly synthesis framework called AnomalyControl to learn cross-modal semantic features as guidance signals, which could encode the generalized anomaly cues from text-image reference prompts and improve the realism of synthesized abnormal samples. Specifically, AnomalyControl adopts a flexible and non-matching prompt pair (i.e., a text-image reference prompt and a targeted text prompt), where a Cross-modal Semantic Modeling (CSM) module is designed to extract cross-modal semantic features from the textual and visual descriptors. Then, an Anomaly-Semantic Enhanced Attention (ASEA) mechanism is formulated to allow CSM to focus on the specific visual patterns of the anomaly, thus enhancing the realism and contextual relevance of the generated anomaly features. Treating cross-modal semantic features as the prior, a Semantic Guided Adapter (SGA) is designed to encode effective guidance signals for the adequate and controllable synthesis process. Extensive experiments indicate that AnomalyControl can achieve state-of-the-art results in anomaly synthesis compared with existing methods while exhibiting superior performance for downstream tasks.

cs.CV↗

arXiv2025

3DTTNet: Multimodal Fusion-Based 3D Traversable Terrain Modeling for Off-Road Environments

Off-road environments remain significant challenges for autonomous ground vehicles, due to the lack of structured roads and the presence of complex obstacles, such as uneven terrain, vegetation, and occlusions. Traditional perception algorithms, primarily designed for structured environments, often fail in unstructured scenarios. In this paper, traversable area recognition is achieved through semantic scene completion. A novel multimodal method, 3DTTNet, is proposed to generate dense traversable terrain estimations by integrating LiDAR point clouds with monocular images from a forward-facing perspective. By integrating multimodal data, environmental feature extraction is strengthened, which is crucial for accurate terrain modeling in complex terrains. Furthermore, RELLIS-OCC, a dataset with 3D traversable annotations, is introduced, incorporating geometric features such as step height, slope, and unevenness. Through a comprehensive analysis of vehicle obsta cle-crossing conditions and the incorporation of vehicle body structure constraints, four traversability cost labels are generated: lethal, medium-cost, low-cost, and free. Experimental results demonstrate that 3DTTNet outperforms the comparison approaches in 3D traversable area recognition, particularly in off-road environments with irregular geometries and partial occlusions. Specifically, 3DTTNet achieves a 42\% improvement in scene completion IoU compared to other models. The proposed framework is scalable and adaptable to various vehicle platforms, allowing for adjustments to occupancy grid parameters and the integration of advanced dynamic models for traversability cost estimation.

cs.RO↗

arXiv2025

Urban In-Context Learning: Bridging Pretraining and Inference through Masked Diffusion for Urban Profiling

Urban profiling aims to predict urban profiles in unknown regions and plays a critical role in economic and social censuses. Existing approaches typically follow a two-stage paradigm: first, learning representations of urban areas; second, performing downstream prediction via linear probing, which originates from the BERT era. Inspired by the development of GPT style models, recent studies have shown that novel self-supervised pretraining schemes can endow models with direct applicability to downstream tasks, thereby eliminating the need for task-specific fine-tuning. This is largely because GPT unifies the form of pretraining and inference through next-token prediction. However, urban data exhibit structural characteristics that differ fundamentally from language, making it challenging to design a one-stage model that unifies both pretraining and inference. In this work, we propose Urban In-Context Learning, a framework that unifies pretraining and inference via a masked autoencoding process over urban regions. To capture the distribution of urban profiles, we introduce the Urban Masked Diffusion Transformer, which enables each region' s prediction to be represented as a distribution rather than a deterministic value. Furthermore, to stabilize diffusion training, we propose the Urban Representation Alignment Mechanism, which regularizes the model's intermediate features by aligning them with those from classical urban profiling methods. Extensive experiments on three indicators across two cities demonstrate that our one-stage method consistently outperforms state-of-the-art two-stage approaches. Ablation studies and case studies further validate the effectiveness of each proposed module, particularly the use of diffusion modeling.

cs.LG↗

arXiv2025

HOLa: Zero-Shot HOI Detection with Low-Rank Decomposed VLM Feature Adaptation

Zero-shot human-object interaction (HOI) detection remains a challenging task, particularly in generalizing to unseen actions. Existing methods address this challenge by tapping Vision-Language Models (VLMs) to access knowledge beyond the training data. However, they either struggle to distinguish actions involving the same object or demonstrate limited generalization to unseen classes. In this paper, we introduce HOLa (Zero-Shot HOI Detection with Low-Rank Decomposed VLM Feature Adaptation), a novel approach that both enhances generalization to unseen classes and improves action distinction. In training, HOLa decomposes VLM text features for given HOI classes via low-rank factorization, producing class-shared basis features and adaptable weights. These features and weights form a compact HOI representation that preserves shared information across classes, enhancing generalization to unseen classes. Subsequently, we refine action distinction by adapting weights for each HOI class and introducing human-object tokens to enrich visual interaction representations. To further distinguish unseen actions, we guide the weight adaptation with LLM-derived action regularization. Experimental results show that our method sets a new state-of-the-art across zero-shot HOI settings on HICO-DET, achieving an unseen-class mAP of 27.91 in the unseen-verb setting. Our code is available at https://github.com/ChelsieLei/HOLa.

cs.CV↗

arXiv2025

Coflex: Enhancing HW-NAS with Sparse Gaussian Processes for Efficient and Scalable DNN Accelerator Design

Hardware-Aware Neural Architecture Search (HW-NAS) is an efficient approach to automatically co-optimizing neural network performance and hardware energy efficiency, making it particularly useful for the development of Deep Neural Network accelerators on the edge. However, the extensive search space and high computational cost pose significant challenges to its practical adoption. To address these limitations, we propose Coflex, a novel HW-NAS framework that integrates the Sparse Gaussian Process (SGP) with multi-objective Bayesian optimization. By leveraging sparse inducing points, Coflex reduces the GP kernel complexity from cubic to near-linear with respect to the number of training samples, without compromising optimization performance. This enables scalable approximation of large-scale search space, substantially decreasing computational overhead while preserving high predictive accuracy. We evaluate the efficacy of Coflex across various benchmarks, focusing on accelerator-specific architecture. Our experimental results show that Coflex outperforms state-of-the-art methods in terms of network accuracy and Energy-Delay-Product, while achieving a computational speed-up ranging from 1.9x to 9.5x.

cs.LG↗

arXiv2025

The orbital period changes for novae

Cataclysmic variable (CVs) are close interacting binaries in which a white dwarf accretes materials from a low mass main sequence companion. CVs can experience nova eruptions due to low mass transfer rates. In the standard theory of CV evolution, the ejected materials during nova eruptions are assumed to leave the system in the form of fast, isotropic, optically thick winds, which predicts that novae only result in positive variation (expansion) of orbital period (i.e. positive $ΔP$). In addition, the angular momentum losses (magnetic braking and gravitational radiation) only predicts a steady long-term decay in the orbital period of CVs, i.e. $\dot P$ is negative. Interestingly, an observation lasting over 30 years reveals positive and negative values for both $ΔP$ and $\dot P$ in CVs, strongly conflicting with the standard evolutionary patterns. However, it cannot be excluded that these observations originate from short-term phenomena caused by nova eruptions because of a short timescale of observations. In this paper, we model the effect of instantaneous nova eruptions on the evolution of CVs, considering three mechanisms associated with mass loss in nova eruptions, including fast wind, Frank jet and binary-driven mass loss. By assuming that the observed $ΔP$ and $\dot P$ are dominated by short-term phenomena, our results show that the binary-driven mass loss can explain almost all of the observations of normal CVs. However, the Frank jet may be needed for some of long-period CVs with evolved companions.

astro-ph.SR↗

arXiv2025

Fast multipole method for the Laplace equation in half plane with Robin boundary condition

In this paper, we present a fast multipole method (FMM) for solving the two-dimensional Laplace equation in a half-plane with Robin boundary conditions. The method is based on a novel expansion theory for the reaction component of the Green's function. By applying the Fourier transform, the reaction field component is obtained in a Sommerfeld-type integral form. We derive far-field approximations and corresponding shifting and translation operators from the Fourier integral representation. The FMM for the reaction component is then developed by using the new far-field approximations incorporated into the classic FMM framework in which the tree structure is constructed from the original and image charges. Combining this with the standard FMM for the free-space components, we develop a fast algorithm to compute the interaction of the half plane Laplace Green's function. We prove that the method exhibits exponential convergence, similar to the free-space FMM. Finally, numerical examples are presented to validate the theoretical results and demonstrate that the FMM achieves $O(N)$ computational complexity.

math.NA↗

arXiv2025

MindSpeed RL: Distributed Dataflow for Scalable and Efficient RL Training on Ascend NPU Cluster

Reinforcement learning (RL) is a paradigm increasingly used to align large language models. Popular RL algorithms utilize multiple workers and can be modeled as a graph, where each node is the status of a worker and each edge represents dataflow between nodes. Owing to the heavy cross-node dependencies, the RL training system usually suffers from poor cluster scalability and low memory utilization. In this article, we introduce MindSpeed RL, an effective and efficient system for large-scale RL training. Unlike existing centralized methods, MindSpeed RL organizes the essential data dependencies in RL training, i.e., sample flow and resharding flow, from a distributed view. On the one hand, a distributed transfer dock strategy, which sets controllers and warehouses on the basis of the conventional replay buffer, is designed to release the dispatch overhead in the sample flow. A practical allgather--swap strategy is presented to eliminate redundant memory usage in resharding flow. In addition, MindSpeed RL further integrates numerous parallelization strategies and acceleration techniques for systematic optimization. Compared with existing state-of-the-art systems, comprehensive experiments on the RL training of popular Qwen2.5-Dense-7B/32B, Qwen3-MoE-30B, and DeepSeek-R1-MoE-671B show that MindSpeed RL increases the throughput by 1.42 ~ 3.97 times. Finally, we open--source MindSpeed RL and perform all the experiments on a super pod of Ascend with 384 neural processing units (NPUs) to demonstrate the powerful performance and reliability of Ascend.

cs.LG↗

arXiv2025

AIR-Bench: Automated Heterogeneous Information Retrieval Benchmark

Evaluation plays a crucial role in the advancement of information retrieval (IR) models. However, current benchmarks, which are based on predefined domains and human-labeled data, face limitations in addressing evaluation needs for emerging domains both cost-effectively and efficiently. To address this challenge, we propose the Automated Heterogeneous Information Retrieval Benchmark (AIR-Bench). AIR-Bench is distinguished by three key features: 1) Automated. The testing data in AIR-Bench is automatically generated by large language models (LLMs) without human intervention. 2) Heterogeneous. The testing data in AIR-Bench is generated with respect to diverse tasks, domains and languages. 3) Dynamic. The domains and languages covered by AIR-Bench are constantly augmented to provide an increasingly comprehensive evaluation benchmark for community developers. We develop a reliable and robust data generation pipeline to automatically create diverse and high-quality evaluation datasets based on real-world corpora. Our findings demonstrate that the generated testing data in AIR-Bench aligns well with human-labeled testing data, making AIR-Bench a dependable benchmark for evaluating IR models. The resources in AIR-Bench are publicly available at https://github.com/AIR-Bench/AIR-Bench.

cs.IR↗

arXiv2025

Fast Multipole Method for Maxwell's Equations in Layered Media

We present a fast multipole method (FMM) for solving Maxwell's equations in three-dimensional (3-D) layered media, based on the magnetic vector potential $\boldsymbol A$ under the Lorenz gauge, to derive the layered dyadic Green's function. The dyadic Green's function is represented using three scalar Helmholtz layered Green's functions, with all interface-induced reaction field components expressed through a unified integral representation. By introducing equivalent polarization images for sources and effective locations for targets to reflect the actual transmission distance of different reaction field components, multiple expansions (MEs) and local expansions (LEs) are derived for the far-field governed by actual transmission distance. To further enhance computational efficiency and numerical stability, we employ a Chebyshev polynomial expansion of the associated Legendre functions to speed up the calculation of multipole-to-local (M2L) expansion translations. Finally, leveraging the FMM framework of the Helmholtz equation in 3-D layered media, we develop a FMM for the dyadic Green's function of Maxwell's equations in layered media. Numerical experiments demonstrate the $\mathcal O(N\log N)$-complexity of the resulting FMM method, and rapid convergence for interactions of low-frequency electromagnetic wave sources in 3-D layered media.

math.NA↗

arXiv2025

Gravitational effects on Hong-Ou-Mandel interference in terrestrial laboratory

In this study, we investigate how Earth's gravitational field affects Hong-Ou-Mandel (HOM) interference experiments conducted in a terrestrial laboratory. To second order, we calculate the relativistic time delay from the null geodesic equation (particle perspective), while the phase shift and the associated effective time delay are derived from the Klein-Gordon equation (wave perspective). Since gravity influences both the temporal and spatial parts of the phase shift, these two time delays differ and lead to different coincidence probabilities. The previous HOM experiment conducted on a rotating platform suggests that the wave perspective can explain the experimental results. We further explore the frame dragging and redshift effects in an arbitrarily oriented rectangular interferometer under two distinct scenarios with different photon paths, measuring one effect in each scenario. We find that both effects can be amplified by increasing the number of light loops. Additionally, we emphasize that the next-to-leading order Sagnac effect, arising from gravitational acceleration, is comparable to the Thomas precession, the geodetic effect, and the Lense-Thirring effect. To detect the leading order Sagnac effect and the redshift effect caused by gravitational acceleration, we estimate the number of loops that photons should travel in the interferometer. Furthermore, we propose that the difference between two HOM patterns can be used as a probe to detect gravitational effects on quantum systems.

gr-qc↗