arXiv subjects

Yan Liu

Publications and source records attributed to Yan Liu.

At least 127 records · Page 7Linked to original sources

arXiv2025

Gait-Adaptive Perceptive Humanoid Locomotion with Real-Time Under-Base Terrain Reconstruction

For full-size humanoid robots, even with recent advances in reinforcement learning-based control, achieving reliable locomotion on complex terrains, such as long staircases, remains challenging. In such settings, limited perception, ambiguous terrain cues, and insufficient adaptation of gait timing can cause even a single misplaced or mistimed step to result in rapid loss of balance. We introduce a perceptive locomotion framework that merges terrain sensing, gait regulation, and whole-body control into a single reinforcement learning policy. A downward-facing depth camera mounted under the base observes the support region around the feet, and a compact U-Net reconstructs a dense egocentric height map from each frame in real time, operating at the same frequency as the control loop. The perceptual height map, together with proprioceptive observations, is processed by a unified policy that produces joint commands and a global stepping-phase signal, allowing gait timing and whole-body posture to be adapted jointly to the commanded motion and local terrain geometry. We further adopt a single-stage successive teacher-student training scheme for efficient policy learning and knowledge transfer. Experiments conducted on a 31-DoF, 1.65 m humanoid robot demonstrate robust locomotion in both simulation and real-world settings, including forward and backward stair ascent and descent, as well as crossing a 46 cm gap. Project Page:https://ga-phl.github.io/

cs.RO↗

arXiv2025

CartoMapQA: A Fundamental Benchmark Dataset Evaluating Vision-Language Models on Cartographic Map Understanding

The rise of Visual-Language Models (LVLMs) has unlocked new possibilities for seamlessly integrating visual and textual information. However, their ability to interpret cartographic maps remains largely unexplored. In this paper, we introduce CartoMapQA, a benchmark specifically designed to evaluate LVLMs' understanding of cartographic maps through question-answering tasks. The dataset includes over 2000 samples, each composed of a cartographic map, a question (with open-ended or multiple-choice answers), and a ground-truth answer. These tasks span key low-, mid- and high-level map interpretation skills, including symbol recognition, embedded information extraction, scale interpretation, and route-based reasoning. Our evaluation of both open-source and proprietary LVLMs reveals persistent challenges: models frequently struggle with map-specific semantics, exhibit limited geospatial reasoning, and are prone to Optical Character Recognition (OCR)-related errors. By isolating these weaknesses, CartoMapQA offers a valuable tool for guiding future improvements in LVLM architectures. Ultimately, it supports the development of models better equipped for real-world applications that depend on robust and reliable map understanding, such as navigation, geographic search, and urban planning. Our source code and data are openly available to the research community at: https://github.com/ungquanghuy-kddi/CartoMapQA.git

cs.CV↗

arXiv2025

On the Tension Between Optimality and Adversarial Robustness in Policy Optimization

Achieving optimality and adversarial robustness in deep reinforcement learning has long been regarded as conflicting goals. Nonetheless, recent theoretical insights presented in CAR suggest a potential alignment, raising the important question of how to realize this in practice. This paper first identifies a key gap between theory and practice by comparing standard policy optimization (SPO) and adversarially robust policy optimization (ARPO). Although they share theoretical consistency, a fundamental tension between robustness and optimality arises in practical policy gradient methods. SPO tends toward convergence to vulnerable first-order stationary policies (FOSPs) with strong natural performance, whereas ARPO typically favors more robust FOSPs at the expense of reduced returns. Furthermore, we attribute this tradeoff to the reshaping effect of the strongest adversary in ARPO, which significantly complicates the global landscape by inducing deceptive sticky FOSPs. This improves robustness but makes navigation more challenging. To alleviate this, we develop the BARPO, a bilevel framework unifying SPO and ARPO by modulating adversary strength, thereby facilitating navigability while preserving global optima. Extensive empirical results demonstrate that BARPO consistently outperforms vanilla ARPO, providing a practical approach to reconcile theoretical and empirical performance.

cs.LG↗

arXiv2025

SUPERChem: A Multimodal Reasoning Benchmark in Chemistry

Current benchmarks for evaluating the chemical reasoning capabilities of Large Language Models (LLMs) are limited by oversimplified tasks, lack of process-level evaluation, and misalignment with expert-level chemistry skills. To address these issues, we introduce SUPERChem, a benchmark of 500 expert-curated reasoning-intensive chemistry problems, covering diverse subfields and provided in both multimodal and text-only formats. Original content and an iterative curation pipeline eliminate flawed items and mitigate data contamination. Each problem is paired with an expert-authored solution path, enabling Reasoning Path Fidelity (RPF) scoring to evaluate reasoning quality beyond final-answer accuracy. Evaluations against a human baseline of 40.3% accuracy show that even the best-performing model, GPT-5 (High), reaches only 38.5%, followed closely by Gemini 2.5 Pro (37.9%) and DeepSeek-V3.1-Think (37.3%). SUPERChem elicits multi-step, multimodal reasoning, reveals model-dependent effects of visual information, and distinguishes high-fidelity reasoners from heuristic ones. By providing a challenging benchmark and a reliable evaluation framework, SUPERChem aims to facilitate the advancement of LLMs toward expert-level chemical intelligence. The dataset of the benchmark is available at https://huggingface.co/datasets/ZehuaZhao/SUPERChem.

cs.CL↗

arXiv2025

Spatially Parallel All-optical Neural Networks

All-optical neural networks (AONNs) have emerged as a promising paradigm for ultrafast and energy-efficient computation. These networks typically consist of multiple serially connected layers between input and output layers--a configuration we term spatially series AONNs, with deep neural networks (DNNs) being the most prominent examples. However, such series architectures suffer from progressive signal degradation during information propagation and critically require additional nonlinearity designs to model complex relationships effectively. Here we propose a spatially parallel architecture for all-optical neural networks (SP-AONNs). Unlike series architecture that sequentially processes information through consecutively connected optical layers, SP-AONNs divide the input signal into identical copies fed simultaneously into separate optical layers. Through coherent interference between these parallel linear sub-networks, SP-AONNs inherently enable nonlinear computation without relying on active nonlinear components or iterative updates. We implemented a modular 4F optical system for SP-AONNs and evaluated its performance across multiple image classification benchmarks. Experimental results demonstrate that increasing the number of parallel sub-networks consistently enhances accuracy, improves noise robustness, and expands model expressivity. Our findings highlight spatial parallelism as a practical and scalable strategy for advancing the capabilities of optical neural computing.

physics.optics↗

arXiv2025

RAGFort: Dual-Path Defense Against Proprietary Knowledge Base Extraction in Retrieval-Augmented Generation

Retrieval-Augmented Generation (RAG) systems deployed over proprietary knowledge bases face growing threats from reconstruction attacks that aggregate model responses to replicate knowledge bases. Such attacks exploit both intra-class and inter-class paths, progressively extracting fine-grained knowledge within topics and diffusing it across semantically related ones, thereby enabling comprehensive extraction of the original knowledge base. However, existing defenses target only one path, leaving the other unprotected. We conduct a systematic exploration to assess the impact of protecting each path independently and find that joint protection is essential for effective defense. Based on this, we propose RAGFort, a structure-aware dual-module defense combining "contrastive reindexing" for inter-class isolation and "constrained cascade generation" for intra-class protection. Experiments across security, performance, and robustness confirm that RAGFort significantly reduces reconstruction success while preserving answer quality, offering comprehensive defense against knowledge base extraction attacks.

cs.AI↗

arXiv2025

SIG-Chat: Spatial Intent-Guided Conversational Gesture Generation Involving How, When and Where

The accompanying actions and gestures in dialogue are often closely linked to interactions with the environment, such as looking toward the interlocutor or using gestures to point to the described target at appropriate moments. Speech and semantics guide the production of gestures by determining their timing (WHEN) and style (HOW), while the spatial locations of interactive objects dictate their directional execution (WHERE). Existing approaches either rely solely on descriptive language to generate motions or utilize audio to produce non-interactive gestures, thereby lacking the characterization of interactive timing and spatial intent. This significantly limits the applicability of conversational gesture generation, whether in robotics or in the fields of game and animation production. To address this gap, we present a full-stack solution. We first established a unique data collection method to simultaneously capture high-precision human motion and spatial intent. We then developed a generation model driven by audio, language, and spatial data, alongside dedicated metrics for evaluating interaction timing and spatial accuracy. Finally, we deployed the solution on a humanoid robot, enabling rich, context-aware physical interactions.

cs.GR↗

arXiv2025

U2F: Encouraging SWE-Agent to Seize Novelty without Losing Feasibility

Large language models (LLMs) have shown strong capabilities in software engineering tasks, yet most existing LLM-based SWE-Agents mainly tackle well-defined problems using conventional methods, often overlooking alternative or innovative solutions beyond their predefined frameworks. This limitation is evident in open-world software environments, where emerging challenges transcend established paradigms. We propose U2F (Unknown Unknowns to Functional solutions), a cognitive-inspired, uncertainty-embracing multi-agent framework that systematically surfaces "Unknown Unknowns" - novel solution pathways absent from initial formulations but holding innovative potential. U2F consists of two key components: (1) a Discovery-Exploration-Integration agent system for uncovering and synthesizing potential solutions, and (2) cognitive enhancement mechanisms across three dimensions: cross-domain analogical reasoning, reverse thinking, and external validation, which strategically reframe and extend conventional solution boundaries. Applied to 218 real-world software enabler stories curated from authentic engineering tasks, U2F achieved notable improvements: human experts reported a 14 percent increase in overall novelty, 51 percent improvement in semantic novelty, and stable feasibility (4.02/5.0), corroborated by an LLM-based evaluator. These results highlight the potential of embracing uncertainty as a catalyst for innovation in software engineering.

cs.SE↗

arXiv2025

Higher-Order Causal Structure Learning with Additive Models

Causal structure learning has long been the central task of inferring causal insights from data. Despite the abundance of real-world processes exhibiting higher-order mechanisms, however, an explicit treatment of interactions in causal discovery has received little attention. In this work, we focus on extending the causal additive model (CAM) to additive models with higher-order interactions. This second level of modularity we introduce to the structure learning problem is most easily represented by a directed acyclic hypergraph which extends the DAG. We introduce the necessary definitions and theoretical tools to handle the novel structure we introduce and then provide identifiability results for the hyper DAG, extending the typical Markov equivalence classes. We next provide insights into why learning the more complex hypergraph structure may actually lead to better empirical results. In particular, more restrictive assumptions like CAM correspond to easier-to-learn hyper DAGs and better finite sample complexity. We finally develop an extension of the greedy CAM algorithm which can handle the more complex hyper DAG search space and demonstrate its empirical usefulness in synthetic experiments.

cs.LG↗

arXiv2025

Temporal filtered quantum sensing with the nitrogen-vacancy center in diamond

Nitrogen vacancy centers in diamond are among the leading solid state quantum platforms, offering exceptional spatial resolution and sensitivity for applications such as magnetic field sensing, thermometry, and bioimaging. However, in high background environments,such as those encountered in in vitro diagnostics, the performance of NV based sensors can be compromised by strong background fluorescence, particularly from substrates such as nitrocellulose. In this work, we analytically and experimentally investigate the use of pulsed laser excitation combined with time gating techniques to suppress background fluorescence and enhance the signal to noise ratio in NV based quantum sensing, with an emphasis on spin enhanced biosensing. Through experimental studies using mixed ensembles of silicon vacancy and NV centers in bulk diamond, as well as fluorescent nanodiamonds on NC substrates, we demonstrate significant improvements in NV spin resonance visibility, demonstrated by an increase of the SNR by up to 4x, and a resulting measurement time reduction by 16x. The presented technique and results here can help significantly increase the readout efficiency and speed in future applications of NV centers in high background environments, such as in IVD, where the NV centers are used as a fluorescent label for biomolecules.

quant-ph↗

arXiv2025

Revisiting PSF models: unifying framework and high-performance implementation

Localization microscopy often relies on detailed models of point spread functions. For applications such as deconvolution or PSF engineering, accurate models for light propagation in imaging systems with high numerical aperture are required. Different models have been proposed based on 2D Fourier transforms or 1D Bessel integrals. The most precise ones combine a vectorial description of the electric field and precise aberration models. However, it may be unclear which model to choose, as there is no comprehensive comparison between the Fourier and Bessel approaches yet. Moreover, many existing libraries are written in Java (e.g. our previous PSF generator software) or MATLAB, which hinders the integration into deep learning algorithms. In this work, we start from the original Richards-Wolf integral and revisit both approaches in a systematic way. We present a unifying framework in which we prove the equivalence between the Fourier and Bessel strategies and detail a variety of correction factors applicable to both of them. Then, we provide a high-performance implementation of our theoretical framework in the form of an open-source library that is built on top of PyTorch, a popular library for deep learning. It enables us to benchmark the accuracy and computational speed of different models, thus allowing for an in-depth comparison of the existing models for the first time. We show that the Bessel strategy is optimal for axisymmetric beams while the Fourier approach can be applied to more general scenarios. Our work enables efficient PSF computation on CPU or GPU, which can then be included in simulation and optimization pipelines.

physics.optics↗

arXiv2025

A global inverse-problem approach to quantitative photo-switching optoacoustic mesoscopy

In this paper, we propose a global framework that includes a detailed model of the photo-switching and acoustic processes for photo-switching optoacoustic mesoscopy, based on the underlying physics. We efficiently implement two forward models as matrix-free linear operators and join them as one forward operator. Then, we reconstruct the concentration maps directly from the temporal series of acoustic signals through the resolution of one combined inverse problem. For robustness against noise and clean unmixing results, we adopt a hybrid regularization technique composed of the $l_1$ and total-variation regularizers applied to two different spaces. We use a proximal-gradient algorithm to solve the minimization problem. Our numerical results show that our regularized one-step approach is the most robust in terms of noise and experimental setup. It consistently achieves higher-quality images, as compared to two-step or unregularized methods.

physics.optics↗

arXiv2025

Quantum Corrected Geodesic Motion in Polymer Kerr-like Spacetime

Rotating black holes are prevalent in astrophysical observations, and a Kerr-like solution that incorporates quantum gravity effects is essential for constructing realistic models. In this work, we analyze the geodesic motion of massive particles in a Kerr-like polymer spacetime, incorporating quantum corrections via a parameter $A_λ$. We demonstrate that increasing $A_λ$ allows for additional orbital evolution in extreme mass ratio inspiral (EMRI) systems before merging. Our results show that the radii, energy, and angular momentum of both the innermost stable circular orbit (ISCO) and marginal circular orbit (MCO) decrease as $A_λ$ increases. Furthermore, when the primary object becomes a wormhole, both prograde ISCO and MCO can intersect the transition surface at the wormhole throat and vanish as $A_λ$ grows. Additionally, we find that the eccentricity of periodic geodesic motion decreases monotonically with increasing $A_λ$. Finally, we explore the variation of the rational number that characterizes periodic motion and highlight the influence of the quantum parameter on different types of periodic orbits, classified by a set of integers associated with the rational number. This work contributes to the understanding of quantum gravity effects and offers potential observational signatures, particularly in the study of EMRIs.

gr-qc↗

arXiv2025

Energy transport in holographic junctions

We study energy transport in a conformal junction connecting three 2D conformal field theories using the AdS/CFT correspondence. The holographic dual consists of three AdS$_3$ spacetimes joined along the worldsheet of a tensile string anchored at the junction. Within a specific range of string tension, where the bulk solution is uniquely determined, we find that all energy reflection and transmission coefficients vary monotonically with the string tension. Notably, the total energy transmission, which quantifies the energy flux from one CFT through the junction, is bounded above by the effective central charge associated with both the CFTs and the junction. Results for energy transport in conformal interfaces are recovered in a special limit. Furthermore, we extend our analysis to junctions connecting $N$ 2D CFTs.

hep-th↗

arXiv2025

Semantic Surgery: Zero-Shot Concept Erasure in Diffusion Models

Concept erasure in text-to-image diffusion models is crucial for mitigating harmful content, yet existing methods often compromise generative quality. We introduce Semantic Surgery, a novel training-free, zero-shot framework for concept erasure that operates directly on text embeddings before the diffusion process. It dynamically estimates the presence of target concepts in a prompt and performs a calibrated vector subtraction to neutralize their influence at the source, enhancing both erasure completeness and locality. The framework includes a Co-Occurrence Encoding module for robust multi-concept erasure and a visual feedback loop to address latent concept persistence. As a training-free method, Semantic Surgery adapts dynamically to each prompt, ensuring precise interventions. Extensive experiments on object, explicit content, artistic style, and multi-celebrity erasure tasks show our method significantly outperforms state-of-the-art approaches. We achieve superior completeness and robustness while preserving locality and image quality (e.g., 93.58 H-score in object erasure, reducing explicit content to just 1 instance, and 8.09 H_a in style erasure with no quality degradation). This robustness also allows our framework to function as a built-in threat detection system, offering a practical solution for safer text-to-image generation.

cs.CV↗

arXiv2025

EvaLearn: Quantifying the Learning Capability and Efficiency of LLMs via Sequential Problem Solving

We introduce EvaLearn, a pioneering benchmark designed to evaluate large language models (LLMs) on their learning capability and efficiency in challenging tasks, a critical, yet underexplored aspect of model potential. EvaLearn contains 648 challenging problems across six task types, grouped into 182 sequences, each sequence dedicated to one task type. Diverging from most existing benchmarks that evaluate models in parallel, EvaLearn requires models to solve problems sequentially, allowing them to leverage the experience gained from previous solutions. EvaLearn provides five comprehensive automated metrics to evaluate models and quantify their learning capability and efficiency. We extensively benchmark nine frontier models and observe varied performance profiles: some models, such as Claude-3.7-sonnet, start with moderate initial performance but exhibit strong learning ability, while some models struggle to benefit from experience and may even show negative transfer. Moreover, we investigate model performance under two learning settings and find that instance-level rubrics and teacher-model feedback further facilitate model learning. Importantly, we observe that current LLMs with stronger static abilities do not show a clear advantage in learning capability across all tasks, highlighting that EvaLearn evaluates a new dimension of model performance. We hope EvaLearn provides a novel evaluation perspective for assessing LLM potential and understanding the gap between models and human capabilities, promoting the development of deeper and more dynamic evaluation approaches. All datasets, the automatic evaluation framework, and the results studied in this paper are available at the GitHub repository.

cs.CL↗

arXiv2025

Efficient small-cell sampling for machine-learning potentials of multi-principal element alloys

Multi-principal element alloys (MPEAs) exhibit exceptional properties but face significant challenges in developing accurate machine-learning potentials (MLPs) due to their vast compositional and configurational complexity. Here, we introduce an efficient small-cell sampling (SCS) method, which allows for generating diverse and representative training datasets for MPEAs using only small-cell structures with just one and two elements, thereby bypassing the computational overhead of iterative active learning cycles and large-cell density functional theory calculations. The efficacy of the method is carefully validated through principal component analysis, extrapolation grades evaluation, and root-mean-square errors and physical properties assessment on the TiZrHfCuNi system. Further demonstrations on TiZrVMo, CoCrFeMnNi, and AlTiZrNbHfTa systems accurately reproduce complex phenomena including phase transitions, chemical orderings, and thermodynamic properties. This work establishes an efficient one-shot protocol for constructing high-quality training datasets across multiple elements, laying a solid foundation for developing universal MLPs for MPEAs.

cond-mat.mtrl-sci↗

arXiv2025

TraceAegis: Securing LLM-Based Agents via Hierarchical and Behavioral Anomaly Detection

LLM-based agents have demonstrated promising adaptability in real-world applications. However, these agents remain vulnerable to a wide range of attacks, such as tool poisoning and malicious instructions, that compromise their execution flow and can lead to serious consequences like data breaches and financial loss. Existing studies typically attempt to mitigate such anomalies by predefining specific rules and enforcing them at runtime to enhance safety. Yet, designing comprehensive rules is difficult, requiring extensive manual effort and still leaving gaps that result in false negatives. As agent systems evolve into complex software systems, we take inspiration from software system security and propose TraceAegis, a provenance-based analysis framework that leverages agent execution traces to detect potential anomalies. In particular, TraceAegis constructs a hierarchical structure to abstract stable execution units that characterize normal agent behaviors. These units are then summarized into constrained behavioral rules that specify the conditions necessary to complete a task. By validating execution traces against both hierarchical and behavioral constraints, TraceAegis is able to effectively detect abnormal behaviors. To evaluate the effectiveness of TraceAegis, we introduce TraceAegis-Bench, a dataset covering two representative scenarios: healthcare and corporate procurement. Each scenario includes 1,300 benign behaviors and 300 abnormal behaviors, where the anomalies either violate the agent's execution order or break the semantic consistency of its execution sequence. Experimental results demonstrate that TraceAegis achieves strong performance on TraceAegis-Bench, successfully identifying the majority of abnormal behaviors.

cs.CR↗