Searcharxiv⌕ Search

arXiv subjects

Huijun Zhao

Publications and source records attributed to Huijun Zhao.

9 recordsLinked to original sources
arXiv2026

LegalDrill: Diagnosis-Driven Synthesis for Legal Reasoning in Small Language Models

Small language models (SLMs) are promising for real-world deployment due to their efficiency and low operational cost. However, their limited capacity struggles with high-stakes legal reasoning tasks that require coherent statute interpretation and logically consistent deduction. Furthermore, training SLMs for such tasks demands high-quality, concise reasoning trajectories, which are prohibitively expensive to manually collect and difficult to curate via standard rejection sampling, lacking granularity beyond final verdicts. To address these challenges, we propose {LegalDrill}, a diagnosis-driven synthesis framework that extracts and iteratively refines reasoning trajectories from a capable teacher via fine-grained prompting, then a self-reflective verification is employed to adaptively select the most effective data for the SLM student. The resulting data empower SLM training through supervised fine-tuning and direct preference optimization. Extensive experiments on several legal benchmarks demonstrate that {LegalDrill} significantly bolsters the legal reasoning capabilities of representative SLMs while bypassing the need for scarce expert annotations, paving a scalable path toward practical legal reasoning systems.

cs.CL↗
arXiv2026

Do VLMs Have a Moral Backbone? A Study on the Fragile Morality of Vision-Language Models

Despite substantial efforts toward improving the moral alignment of Vision-Language Models (VLMs), it remains unclear whether their ethical judgments are stable in realistic settings. This work studies moral robustness in VLMs, defined as the ability to preserve moral judgments under textual and visual perturbations that do not alter the underlying moral context. We systematically probe VLMs with a diverse set of model-agnostic multimodal perturbations and find that their moral stances are highly fragile, frequently flipping under simple manipulations. Our analysis reveals systematic vulnerabilities across perturbation types, moral domains, and model scales, including a sycophancy trade-off where stronger instruction-following models are more susceptible to persuasion. We further show that lightweight inference-time interventions can partially restore moral stability. These results demonstrate that moral alignment alone is insufficient and that moral robustness is a necessary criterion for the responsible deployment of VLMs.

cs.CY↗
arXiv2025

MORALISE: A Structured Benchmark for Moral Alignment in Visual Language Models

Warning: This paper contains examples of harmful language and images. Reader discretion is advised. Recently, vision-language models have demonstrated increasing influence in morally sensitive domains such as autonomous driving and medical analysis, owing to their powerful multimodal reasoning capabilities. As these models are deployed in high-stakes real-world applications, it is of paramount importance to ensure that their outputs align with human moral values and remain within moral boundaries. However, existing work on moral alignment either focuses solely on textual modalities or relies heavily on AI-generated images, leading to distributional biases and reduced realism. To overcome these limitations, we introduce MORALISE, a comprehensive benchmark for evaluating the moral alignment of vision-language models (VLMs) using diverse, expert-verified real-world data. We begin by proposing a comprehensive taxonomy of 13 moral topics grounded in Turiel's Domain Theory, spanning the personal, interpersonal, and societal moral domains encountered in everyday life. Built on this framework, we manually curate 2,481 high-quality image-text pairs, each annotated with two fine-grained labels: (1) topic annotation, identifying the violated moral topic(s), and (2) modality annotation, indicating whether the violation arises from the image or the text. For evaluation, we encompass two tasks, \textit{moral judgment} and \textit{moral norm attribution}, to assess models' awareness of moral violations and their reasoning ability on morally salient content. Extensive experiments on 19 popular open- and closed-source VLMs show that MORALISE poses a significant challenge, revealing persistent moral limitations in current state-of-the-art models. The full benchmark is publicly available at https://huggingface.co/datasets/Ze1025/MORALISE.

cs.CV↗
arXiv2023

A Surface Acoustic Wave based Single Photon Shifter for Solid-state Sources

Controlling the frequency of nonclassical light is indispensable for implementing quantum computation, communication and bridging various quantum systems. However, frequency-shift devices for solid state single-photon sources that are easy to integrate are practically absent. Here, we propose an integrated single-photon frequency shifter based on acousto-optic modulation. The device consists of two Interdigital Transducers (IDTs) for surface acoustic wave (SAW) generation and a silicon waveguide periodically placed at the nodes the SAW to increase the interaction length. The Vπ*L of the device is 1.2v.cm. Under 133.2MHz driving frequency and 10 volt driving voltage, a shift up to 65.7GHz is achieved with near unity conversion efficiency. Our results demonstrate the feasibility of on-chip deterministic quantum spectral control in constructing hybrid quantum networks.

physics.optics↗
arXiv2023

Revisiting the Electrified Pt(111)/Water Interfaces through an Affordable Double-Reference ab-initio Approach

The electrified solid-liquid interface plays an essential role in many renewable energy-related applications, including hydrogen production and utilization. Limitations in computational modelling of the electrified solid-liquid interface have held back the understanding of its properties at the atomic-scale level. In this study, we applied the grand canonical density functional theory (GC-DFT) combined with a hybrid implicit/explicit solvation model to reinvestigate the widely studied electrified platinum-water interface affordably. The calculated double layer capacitances of the Pt(111)-water interface over the applied bias potential closely match the experimental and previous theoretical data from expensive ab-initio molecular dynamics simulations. The structural analysis of the interface models reveals that the applied bias potential can significantly affect the Pt(111)-water atomic interface configurations. The orientation of the water molecules next to the Pt(111) surface is vital for correctly describing the potential of zero charge (PZC) and capacitance. Additionally, the GC-DFT results confirm that the absorption of the hydrogen atom under applied bias potential can significantly affect the electrified interfacial properties. The presented affordable GC-DFT approach, therefore, offers an efficient and accurate means to enhance the understanding of electrified solid-liquid interfaces.

cond-mat.mtrl-sci↗
arXiv2023

Variant Plateau Law in Atomically Thin Transition Metal Dichalcogenide Dome Networks

Since its fundamental inception from soap bubbles, Plateau law has sparked extensive research in equilibrated states. However, most studies primarily relied on liquids, foams or cellular structures, whereas its applicability has yet to be explored in nano-scale solid films. Here, we observed a variant Plateau law in networks of atomically thin domes made of solid two-dimensional (2D) transition metal dichalcogenides (TMDs). Discrete layer-dependent van der Waals (vdWs) interaction energies were experimentally and theoretically obtained for domes protruding in different TMD layers. Significant surface tension differences from layer-dependent vdWs interaction energies manifest in a variant of this fundamental law. Meanwhile, the remarkable mechanical properties, gas impermeability and interlayer vdWs interaction energy of TMD films enable domes and the networks to sustain high gas pressure and exist in a fundamentally variant nature for several years. Our findings pave the way towards exploring variant discretised states with applications in opto-electro-mechanical devices.

cond-mat.mtrl-sci↗
arXiv2021

Sim2Real Docs: Domain Randomization for Documents in Natural Scenes using Ray-traced Rendering

In the past, computer vision systems for digitized documents could rely on systematically captured, high-quality scans. Today, transactions involving digital documents are more likely to start as mobile phone photo uploads taken by non-professionals. As such, computer vision for document automation must now account for documents captured in natural scene contexts. An additional challenge is that task objectives for document processing can be highly use-case specific, which makes publicly-available datasets limited in their utility, while manual data labeling is also costly and poorly translates between use cases. To address these issues we created Sim2Real Docs - a framework for synthesizing datasets and performing domain randomization of documents in natural scenes. Sim2Real Docs enables programmatic 3D rendering of documents using Blender, an open source tool for 3D modeling and ray-traced rendering. By using rendering that simulates physical interactions of light, geometry, camera, and background, we synthesize datasets of documents in a natural scene context. Each render is paired with use-case specific ground truth data specifying latent characteristics of interest, producing unlimited fit-for-task training data. The role of machine learning models is then to solve the inverse problem posed by the rendering pipeline. Such models can be further iterated upon with real-world data by either fine tuning or making adjustments to domain randomization parameters.

cs.CV↗
arXiv2018

Representations of Cyclotomic Rational Cherednik Algebras with Aspherical Parameters

In this article, we describe all two sided ideals of a cyclotomic rational Cherednik algebra $H_\mathbf{c}$ and its spherical subalgebra $eH_\mathbf{c} e$ with a Weil generic aspherical parameter $\mathbf{c}$, and further describe the simple modules in the category $\mathcal{O}^{sph}_\mathbf{c}$ . The main tools we use are categorical Kac-Moody actions on catogories $\mathcal{O}_\mathbf{c}$ and restriction functors for Harish-Chandra bimodules.

math.RT↗