About

Professor Wenxi Wang is a faculty member at the University of Virginia leading a research group dedicated to advancing Neuro-Symbolic AI. His work focuses on bridging the strengths of learning-based and logic-based approaches to develop AI techniques that enhance the scalability and efficiency of automated reasoning systems. Additionally, Professor Wang's research aims to enable AI models themselves to acquire reasoning and verification capabilities, pushing the boundaries of what AI can achieve in terms of logical understanding and reliability. Beyond fundamental research, Professor Wang's group develops specialized Neuro-Symbolic AI methods targeted at reliable software. This includes efforts to improve the reliability and robustness of systems ranging from large-scale cloud infrastructures to modern AI applications, which are treated as complex software systems. His research also explores verifiable code generation to make software development more trustworthy, transparent, and error-resistant. Through these contributions, Professor Wang advances both the theoretical foundations and practical applications of Neuro-Symbolic AI in software reliability and automated reasoning.

Research topics

• Quantum mechanics
• Optics
• Optoelectronics
• Physics
• Mathematics

Selected publications

• Hybrid GA-ANN-LSSVM algorithm for adaptive Brillouin frequency shift extraction in long-distance oil and gas pipeline monitoring

  Optics Express · 2025-05-15 · 3 citations

  articleOpen access1st authorCorresponding

  Long-distance oil and gas pipelines require high-precision safety monitoring systems to prevent failures caused by environmental changes and operational strains. Although Brillouin optical time-domain analysis (BOTDA) systems are widely used for distributed sensing, existing algorithms face challenges in handling multi-peak Brillouin gain spectra (BGS) and high signal-to-noise ratio (SNR) requirements. This paper proposes a hybrid intelligent algorithm (GA-ANN-LSSVM) combining a genetic algorithm-optimized artificial neural network (GA-ANN) and least squares support vector machine (LSSVM) to improve BOTDA performance. The algorithm adaptively identifies single/multi-peak BGS and dynamically invokes GA-ANN or LSSVM for Brillouin frequency shift (BFS) extraction. Experimental results demonstrate that the proposed method achieves temperature and strain measurement errors below 1 °C and 2 με, respectively, with 4.4% higher peak recognition accuracy than conventional algorithms. Additionally, the proposed method reduces the initialization SNR threshold by 5 dB and processing time by 3 seconds, significantly enhancing the practicality of BOTDA in complex pipeline environments.

  PublisherDOI

• Mode splitting regulation and sensing implemented in SNAP microresonators

  Photonics Research · 2025-07-28

  article

  The mode splitting phenomenon in a high-quality whispering gallery mode (WGM) microresonator coupled to sub-wavelength scatterers can help resolve nanoparticle information. In this study, we analyzed the characteristics of nanoparticle-coupling-induced mode splitting, particularly in multiple-particle insertions relevant to practical biochemical sensing applications. The surface nanoscale axial photonics (SNAP) microresonator supports axial mode field distribution. We fabricated a fiber probe and used it to scan along the longitudinal axis of the SNAP microresonator, sequentially adjusting the axial mode to doublets, confirming the regulatability of mode splitting. The sensing region of the SNAP microresonator was immersed in particle-aqueous environments, and we observed mode splitting induced by particle scattering coupling. By analyzing splitting events, information such as particle binding number, average polarizability, and particle size can be resolved. The SNAP microresonator not only avoids interference from the detection environment on the coupling region through spatial separation between the coupling region and the sensing region, but also enables precise nanoparticle calculation based on splitting spectra, significantly enhancing the practicality of WGM microresonators in biochemical sensing applications.

  PublisherDOI

• Affine $$\imath $$Quantum Groups and Twisted Yangians in Drinfeld Presentations

  Communications in Mathematical Physics · 2025-04-04 · 1 citations

  preprintOpen access
  PublisherOA PDFDOI

• Narrow-Linewidth Semiconductor Laser with Hybrid Feedback

  Photonics · 2025-09-02 · 1 citations

  articleOpen accessSenior author

  Narrow-linewidth semiconductor lasers have become indispensable devices in high-precision measurement and detection. Among various available technologies, self-injection locking plays a crucial role due to its significant ability to reduce linewidth and enhance coherence. Here, we demonstrate a hybrid feedback narrow-linewidth laser based on fixed external cavity feedback combined with self-injection locking feedback. The laser consists of a semiconductor gain chip, fiber Bragg grating, and micro-ring resonator, achieving laser mode selection and linewidth compression. Ultimately, a single longitudinal mode narrow-linewidth laser output with a Lorentzian linewidth of 149 Hz and a side-mode suppression ratio of 65 dB was obtained. The demonstrated laser can be applied in applications such as coherent optical communication and high-precision coherent detection.

  PublisherOA PDFDOI

• Mixture-of-Queries Transformer: Camouflaged Instance Segmentation via Queries Cooperation and Frequency Enhancement

  2025-09-01

  articleSenior author

  Due to the high similarity between camouflaged instances and the surroundings and the widespread camouflage-like scenarios, the recently proposed camouflaged instance segmentation (CIS) is a challenging and relevant task. Previous approaches achieve some progress on CIS, while many overlook camouflaged objects’ color and contour nature and then decide on each candidate instinctively. In this paper, we contribute a Mixture-of-Queries Transformer (MoQT) in an end-to-end manner for CIS based on two key designs (a Frequency Enhancement Feature Extractor and a Mixture-of-Queries Decoder). First, the Frequency Enhancement Feature Extractor is responsible for capturing the camouflaged clues in the frequency domain. To expose camouflaged instances, the extractor enhances the effectiveness of contour, eliminates the interference color, and obtains suitable features simultaneously. Second, a Mixture-of-Queries Decoder utilizes multiple newly initialized experts of queries (a group of queries considered an expert) in each layer for spotting camouflaged characteristics with cooperation. These experts collaborate to generate outputs with the mixture-of-queries mechanism, refined hierarchically to a fine-grained level for more accurate instance masks. Coupling these two components enables MoQT to use multiple experts to integrate effective clues of camouflaged objects in both spatial and frequency domains. Extensive experimental results demonstrate our MoQT outperforms 19 state-of-the-art CIS approaches on both COD10K and NC4K datasets.

  PublisherDOI

• Braid group action and quasi-split affine iquantum groups III

  ArXiv.org · 2025-11-02

  preprintOpen access

  This is the last of three papers on Drinfeld presentations of quasi-split affine iquantum groups $\widetilde{\mathbf U}^\imath$, settling the remaining type ${\rm AIII}^{(τ)}_{2r}$. This type distinguishes itself among all quasi-split affine types in having 3 relative root lengths. Various basic real and imaginary $v$-root vectors for $\widetilde{\mathbf U}^\imath$ are constructed, giving rise to affine rank one subalgebras of $\widetilde{\mathbf U}^\imath$ associated with simple roots in the finite relative root system. We establish the relations among these $v$-root vectors and show that they provide a Drinfeld presentation of $\widetilde{\mathbf U}^\imath$.

  PublisherOA PDFDOI

• Affine and cyclotomic $q$-Schur categories via webs

  ArXiv.org · 2025-04-14

  preprintOpen accessSenior author

  We formulate two new $\mathbb Z[q,q^{-1}]$-linear diagrammatic monoidal categories, the affine $q$-web category and the affine $q$-Schur category, as well as their respective cyclotomic quotient categories. Diagrammatic integral bases for the Hom-spaces of all these categories are established. In addition, we establish the following isomorphisms, providing diagrammatic presentations of these $q$-Schur algebras for the first time: (i)~ the path algebras of the affine $q$-web category to R.~Green's affine $q$-Schur algebras, (ii)~ the path algebras of the affine $q$-Schur category to Maksimau-Stroppel's higher level affine $q$-Schur algebras, and most significantly, (iii)~ the path algebras of the cyclotomic $q$-Schur categories to Dipper-James-Mathas' cyclotomic $q$-Schur algebras.

  PublisherOA PDFDOI

• Precision LIDAR Using Dual‐Comb Breathing Spectra

  Laser & Photonics Review · 2025-06-16 · 3 citations

  articleOpen access

  Abstract Light detection and ranging (Lidar) is indispensable in a variety of fields, encompassing basic science, manufacturing, production, and daily life. Here, from a different perspective, A phenomenon is observed occurring between the optical frequency comb (OFC) and obstacles within the optical frequency domain, which is referred to “breathing spectra,” inspired by the dynamic shape alterations with varying lengths, reminiscent of the oscillatory patterns seen during breathing. Precision length metrology is achieved by retracing the peak positions of dual‐microcomb breathing spectra (DBS) with different repetition rates back to the stable comb optical modes, enabling the attainment of nanoscale accuracy across long distance in a single‐shot measurement while consuming fewer computational resources. Minimum Allan deviations of 1.08 nm at a distance of 0.5 m, and 21.8 nm at a distance of 217 m are experimentally demonstrated. The DBS methodology eliminates the need for auxiliary ranging and other complex steps while being CMOS‐compatible and offering the potential for single‐chip integration, will thus emerge as a competitive and novel alternative in the realm of length metrology applications.

  PublisherOA PDFDOI

• AuthGlass: Benchmarking Voice Liveness Detection and Authentication on Smart Glasses via Comprehensive Acoustic Features

  ArXiv.org · 2025-09-25

  preprintOpen access

  With the rapid advancement of smart glasses, voice interaction has been widely adopted due to its naturalness and convenience. However, its practical deployment is often undermined by vulnerability to spoofing attacks, while no public dataset currently exists for voice liveness detection and authentication in smart-glasses scenarios. To address this challenge, we first collect a multi-acoustic-modal dataset comprising 16-channel audio data from 42 subjects, along with corresponding attack samples covering two attack categories. Based on insights derived from this collected data, we propose AuthG-Live, a sound-field-based voice liveness detection method, and AuthG-Net, a multi-acoustic-modal authentication model. We further benchmark seven voice liveness detection methods and four authentication methods across diverse acoustic modalities. The results demonstrate that our proposed approach achieves state-of-the-art performance on four benchmark tasks, and extensive ablation studies validate the generalizability of our methods \red{under real-world constraints}. Finally, we release this dataset, termed AuthGlass, to facilitate future research on voice liveness detection and authentication for smart glasses.

  PublisherOA PDFDOI

• Precise Distance Measurement by Multi-Wavelength Interferometry Using a Soliton Microcomb

  IEEE Photonics Technology Letters · 2025-03-05 · 4 citations

  article

  In this work, we present a method for precise distance measurement through microcomb-based multi-wavelength interferometry. The high repetition frequency of the microcomb results in a non-ambiguity range confined to the millimeter or centimeter scale. To address this limitation, we leverage the microwave signal inherently carried by the microcomb to significantly extend the non-ambiguity range. Initial coarse distance measurements are obtained via the phases of the microwave signal. These measurements are subsequently refined using multi-wavelength interferometry, with the microcomb serving as the direct signal source. Experimental results indicate that the comparison with the reference distance meter can be within 44 nm. The Allan deviation can reach 43 nm at 4 s, and 8.6 nm at 100 s averaging time. Our system is able to provide a ranging method with nanometric precision and extremely large non-ambiguity range.

  PublisherDOI

Recent grants

• Duality between representations of Lie superalgebras and Lie algebras via Kazhdan-Lusztig theory

  NSF · $100k · 2005–2008

• Representation theory and quantum symmetric pairs

  NSF · $206k · 2014–2017

• Canonical Bases, Categorification, and Modular Representations

  NSF · $318k · 2017–2020

• Quantum Symmetric Pairs, Categorification, and Geometry

  NSF · $345k · 2020–2024

• Affine algebras, Lie superalgebras, Hecke algebras, and representations

  NSF · $171k · 2008–2011

Frequent coauthors

• Wenfu Zhang

  Xi'an Institute of Optics and Precision Mechanics

  114 shared

• Wei Zhao

  Xi'an Institute of Optics and Precision Mechanics

  68 shared

• Liao Chen

  60 shared

• Hao Hu

  60 shared

• Chi Zhang

  60 shared

• Xinliang Zhang

  Hebei Medical University

  60 shared

• Brent E. Little

  52 shared

• Wenfu Zhang

  Chinese Academy of Sciences

  45 shared

Labs

• Hiprel GroupPI

  Advancing Neuro-Symbolic AI, bridging the strengths of learning-based and logic-based approaches.

Education

• Ph.D.

  University of Texas at Austin

Awards & honors

• MIT EECS Rising Stars 2022