arXiv subjects

Liang Zhang

Publications and source records attributed to Liang Zhang.

At least 145 records · Page 8Linked to original sources

arXiv2025

Time-R1: Post-Training Large Vision Language Model for Temporal Video Grounding

Temporal Video Grounding (TVG), the task of locating specific video segments based on language queries, is a core challenge in long-form video understanding. While recent Large Vision-Language Models (LVLMs) have shown early promise in tackling TVG through supervised fine-tuning (SFT), their abilities to generalize remain limited. To address this, we propose a novel post-training framework that enhances the generalization capabilities of LVLMs via reinforcement learning (RL). Specifically, our contributions span three key directions: (1) Time-R1: we introduce a reasoning-guided post-training framework via RL with verifiable reward to enhance the capabilities of LVLMs on the TVG task. (2) TimeRFT: we explore data-efficient post-training strategies on our curated RL-friendly dataset, which trains the model to progressively comprehend difficult samples, leading to better generalization. (3) TVGBench: we carefully construct a small yet comprehensive benchmark for LVLM evaluation, assessing 11 types of queries and featuring balanced distributions across both videos and queries. Extensive experiments demonstrate that Time-R1 achieves state-of-the-art performance across multiple downstream datasets using only 2.5K training data, while improving its general video understanding capabilities.

cs.CV↗

arXiv2025

VoteSplat: Hough Voting Gaussian Splatting for 3D Scene Understanding

3D Gaussian Splatting (3DGS) has become horsepower in high-quality, real-time rendering for novel view synthesis of 3D scenes. However, existing methods focus primarily on geometric and appearance modeling, lacking deeper scene understanding while also incurring high training costs that complicate the originally streamlined differentiable rendering pipeline. To this end, we propose VoteSplat, a novel 3D scene understanding framework that integrates Hough voting with 3DGS. Specifically, Segment Anything Model (SAM) is utilized for instance segmentation, extracting objects, and generating 2D vote maps. We then embed spatial offset vectors into Gaussian primitives. These offsets construct 3D spatial votes by associating them with 2D image votes, while depth distortion constraints refine localization along the depth axis. For open-vocabulary object localization, VoteSplat maps 2D image semantics to 3D point clouds via voting points, reducing training costs associated with high-dimensional CLIP features while preserving semantic unambiguity. Extensive experiments demonstrate effectiveness of VoteSplat in open-vocabulary 3D instance localization, 3D point cloud understanding, click-based 3D object localization, hierarchical segmentation, and ablation studies. Our code is available at https://sy-ja.github.io/votesplat/

cs.GR↗

arXiv2025

Learning to utilize image second-order derivative information for crisp edge detection

Edge detection is a fundamental task in computer vision. It has made great progress under the development of deep convolutional neural networks (DCNNs), some of which have achieved a beyond human-level performance. However, recent top-performing edge detection methods tend to generate thick and noisy edge lines. In this work, we solve this problem from two aspects: (1) the lack of prior knowledge regarding image edges, and (2) the issue of imbalanced pixel distribution. We propose a second-order derivative-based multi-scale contextual enhancement module (SDMCM) to help the model locate true edge pixels accurately by introducing the edge prior knowledge. We also construct a hybrid focal loss function (HFL) to alleviate the imbalanced distribution issue. In addition, we employ the conditionally parameterized convolution (CondConv) to develop a novel boundary refinement module (BRM), which can further refine the final output edge maps. In the end, we propose a U-shape network named LUS-Net which is based on the SDMCM and BRM for crisp edge detection. We perform extensive experiments on three standard benchmarks, and the experiment results illustrate that our method can predict crisp and clean edge maps and achieves state-of-the-art performance on the BSDS500 dataset (ODS=0.829), NYUD-V2 dataset (ODS=0.768), and BIPED dataset (ODS=0.903).

cs.CV↗

arXiv2025

Fast Transitions of X-ray Variability in the Neutron Star Low Mass X-ray Binary Cygnus X-2

We present a spectral-timing analysis of two NICER observations of the weakly magnetized neutron star low-mass X-ray binary Cygnus X-2. During these observations, we detect a rapid transition from a narrow 50-Hz horizontal-branch oscillation to a broad 5-Hz normal-branch oscillation, accompanied by an increase in source flux and a decrease in spectral hardness. Thanks to the large effective area of NICER, we are able to conduct a detailed comparison of the spectra associated with different types of quasi-periodic oscillations (QPOs) on short timescales. By fitting the spectra with a model that includes a disc and Comptonization components plus two emission lines, we find that the parameters of the disc component do not change significantly during the transition. However, assuming a fixed electron temperature, the optical depth of the Comptonization component decreases significantly. This drop in optical depth may be attributed to the expansion of the boundary layer or spreading layer.In addition, we find that the rms spectra for both the HBO and NBO are hard, suggesting that the boundary layer or spreading layer is driving the variability. We discuss the potential physical origin of the different types of QPOs.

astro-ph.HE↗

arXiv2025

A Hessian-Aware Stochastic Differential Equation for Modelling SGD

Continuous-time approximation of Stochastic Gradient Descent (SGD) is a crucial tool to study its escaping behaviors from stationary points. However, existing stochastic differential equation (SDE) models fail to fully capture these behaviors, even for simple quadratic objectives. Built on a novel stochastic backward error analysis framework, we derive the Hessian-Aware Stochastic Modified Equation (HA-SME), an SDE that incorporates Hessian information of the objective function into both its drift and diffusion terms. Our analysis shows that HA-SME achieves the order-best approximation error guarantee among existing SDE models in the literature, while significantly reducing the dependence on the smoothness parameter of the objective. Empirical experiments on neural network-based loss functions further validate this improvement. Further, for quadratic objectives, under mild conditions, HA-SME is proved to be the first SDE model that recovers exactly the SGD dynamics in the distributional sense. Consequently, when the local landscape near a stationary point can be approximated by quadratics, HA-SME provides a more precise characterization of the local escaping behaviors of SGD. With the enhanced approximation guarantee, we further conduct an escape time analysis using HA-SME, showcasing how it can be employed to analytically study the escaping behavior of SGD for general function classes.

stat.ML↗

arXiv2025

A Low Power Monolithic Active Pixel Sensor Prototype for the STCF Inner Tracker

The Super Tau-Charm Facility (STCF) is a proposed $e^+e^-$ collider with a peak luminosity 100 times higher than that of the present tau-charm factory. The inner tracker (ITK) of STCF should feature a low material budget and high readout speed. Under these requirements, the monolithic active pixel sensor (MAPS) is considered as a promising candidate for the ITK. To minimize the power consumption of MAPS (for low material budget), larger-size sensors are proposed to reduce the scale of the readout circuitry while preserving the required position resolution. Multiple sensors with varying dimensions and structures were designed and integrated in several prototype chips for performance comparison, fabricated in a 180~nm CIS process. The in-pixel readout circuit can also provide time of arrival (ToA) and time-over-threshold (ToT) of the hit signal, with a least significant bit (LSB) of 50 ns. The peripheral readout circuit performs operations including timestamp correction, data aggregation, caching, framing, 8b/10b encoding, and serialization. According to simulation, the power consumption for a full-scale chip is about 55.7 mW/cm2. Preliminary measurements have been conducted on the prototype chips.

physics.ins-det↗

arXiv2025

LLMs Can Achieve High-quality Simultaneous Machine Translation as Efficiently as Offline

When the complete source sentence is provided, Large Language Models (LLMs) perform excellently in offline machine translation even with a simple prompt "Translate the following sentence from [src lang] into [tgt lang]:". However, in many real scenarios, the source tokens arrive in a streaming manner and simultaneous machine translation (SiMT) is required, then the efficiency and performance of decoder-only LLMs are significantly limited by their auto-regressive nature. To enable LLMs to achieve high-quality SiMT as efficiently as offline translation, we propose a novel paradigm that includes constructing supervised fine-tuning (SFT) data for SiMT, along with new training and inference strategies. To replicate the token input/output stream in SiMT, the source and target tokens are rearranged into an interleaved sequence, separated by special tokens according to varying latency requirements. This enables powerful LLMs to learn read and write operations adaptively, based on varying latency prompts, while still maintaining efficient auto-regressive decoding. Experimental results show that, even with limited SFT data, our approach achieves state-of-the-art performance across various SiMT benchmarks, and preserves the original abilities of offline translation. Moreover, our approach generalizes well to document-level SiMT setting without requiring specific fine-tuning, even beyond the offline translation model.

cs.CL↗

arXiv2025

Optimizing Case-Based Reasoning System for Functional Test Script Generation with Large Language Models

In this work, we explore the potential of large language models (LLMs) for generating functional test scripts, which necessitates understanding the dynamically evolving code structure of the target software. To achieve this, we propose a case-based reasoning (CBR) system utilizing a 4R cycle (i.e., retrieve, reuse, revise, and retain), which maintains and leverages a case bank of test intent descriptions and corresponding test scripts to facilitate LLMs for test script generation. To improve user experience further, we introduce Re4, an optimization method for the CBR system, comprising reranking-based retrieval finetuning and reinforced reuse finetuning. Specifically, we first identify positive examples with high semantic and script similarity, providing reliable pseudo-labels for finetuning the retriever model without costly labeling. Then, we apply supervised finetuning, followed by a reinforcement learning finetuning stage, to align LLMs with our production scenarios, ensuring the faithful reuse of retrieved cases. Extensive experimental results on two product development units from Huawei Datacom demonstrate the superiority of the proposed CBR+Re4. Notably, we also show that the proposed Re4 method can help alleviate the repetitive generation issues with LLMs.

cs.SE↗

arXiv2025

SubGCache: Accelerating Graph-based RAG with Subgraph-level KV Cache

Graph-based retrieval-augmented generation (RAG) enables large language models (LLMs) to incorporate structured knowledge via graph retrieval as contextual input, enhancing more accurate and context-aware reasoning. We observe that for different queries, it could retrieve similar subgraphs as prompts, and thus we propose SubGCache, which aims to reduce inference latency by reusing computation across queries with similar structural prompts (i.e., subgraphs). Specifically, SubGCache clusters queries based on subgraph embeddings, constructs a representative subgraph for each cluster, and pre-computes the key-value (KV) cache of the representative subgraph. For each query with its retrieved subgraph within a cluster, it reuses the pre-computed KV cache of the representative subgraph of the cluster without computing the KV tensors again for saving computation. Experiments on two new datasets across multiple LLM backbones and graph-based RAG frameworks demonstrate that SubGCache consistently reduces inference latency with comparable and even improved generation quality, achieving up to 6.68$\times$ reduction in time-to-first-token (TTFT).

cs.LG↗

arXiv2025

X-ray and Radio Campaign of the Z-source GX 340+0 II: the X-ray polarization in the normal branch

We present the first X-ray polarization measurement of the neutron star low-mass X-ray binary and Z-source, GX 340$+$0, in the normal branch (NB) using a 200 ks observation with the Imaging X-ray Polarimetric Explorer (IXPE). This observation was performed in 2024 August. Along with IXPE, we also conducted simultaneous observations with NICER, AstroSat, Insight-HXMT, ATCA, and GMRT to investigate the broadband spectral and timing properties in the X-ray and radio wavelengths. During the campaign, the source traced a complete Z-track during the IXPE observation but spent most of the time in the NB. We measure X-ray polarization degree (PD) of $1.22\pm0.25\%$ in the 2-8 keV energy band with a polarization angle (PA) of $38\pm6^\circ$. The PD in the NB is observed to be weaker than in the horizontal branch (HB) but aligned in the same direction. The PD of the source exhibits a marginal increase with energy while the PA shows no energy dependence. The joint spectro-polarimetric modeling is consistent with the observed X-ray polarization originating from a single spectral component from the blackbody or the Comptonized emission while the disk emission does not contribute towards the X-ray polarization. GMRT observations at 1.26 GHz during HB had a tentative detection at 4.5$\pm$0.7 mJy while ATCA observations a day later during the NB detected the source at 0.70$\pm$0.05 mJy and 0.59$\pm$0.05 mJy in the 5.5 & 9 GHz bands, respectively, suggesting an evolving jet structure depending on the Z-track position.

astro-ph.HE↗

arXiv2025

A Deep Learning Scheme of Electromagnetic Scattering From Scatterers With Incomplete Profiles

A deep learning scheme is proposed to solve the electromagnetic (EM) scattering problems where the profile of the dielectric scatterer of interest is incomplete. As a compensation, a limited amount of scattering data is provided, which is in principle containing sufficient information associated with the missing part of the profile. The existing solvers can hardly realize the compensation if the known part of the profile and the scattering data are combined straightforwardly. On one hand, the well-developed forward solvers have no mechanism to accept the scattering data, which can recover the unknown part of the profile if properly used. On the other hand, the existing solvers for inverse problems cannot retrieve the complete profile with an acceptable accuracy from the limited amount of scattering data, even when the available part of the profile can be fed into the solvers. This work aims to handle the difficulty. To this end, the EM forward scattering from an incompletely known dielectric scatterer is derived. A scheme based on DL is then proposed where the forward and inverse scattering problems are solved simultaneously. Numerical experiments are conducted to demonstrate the performance of the proposed DL-based scheme for both two-dimensional (2-D) and three-dimensional (3-D) EM scattering problems.

cs.CE↗

arXiv2025

Exploring Communication Strategies for Collaborative LLM Agents in Mathematical Problem-Solving

Large Language Model (LLM) agents are increasingly utilized in AI-aided education to support tutoring and learning. Effective communication strategies among LLM agents improve collaborative problem-solving efficiency and facilitate cost-effective adoption in education. However, little research has systematically evaluated the impact of different communication strategies on agents' problem-solving. Our study examines four communication modes, \textit{teacher-student interaction}, \textit{peer-to-peer collaboration}, \textit{reciprocal peer teaching}, and \textit{critical debate}, in a dual-agent, chat-based mathematical problem-solving environment using the OpenAI GPT-4o model. Evaluated on the MATH dataset, our results show that dual-agent setups outperform single agents, with \textit{peer-to-peer collaboration} achieving the highest accuracy. Dialogue acts like statements, acknowledgment, and hints play a key role in collaborative problem-solving. While multi-agent frameworks enhance computational tasks, effective communication strategies are essential for tackling complex problems in AI education.

cs.HC↗

arXiv2025

Hidden Zeros and $2$-split via BCFW Recursion Relation

In this paper, we provide another angle to understand recent discoveries, i.e., the hidden zeros and corresponding 2-split behavior using the BCFW recursion relation. For the hidden zeros, we show that although the BCFW recursion relation is not directly applicable for computing amplitudes of the non-linear sigma model, we can indeed prove the zeros using the modified BCFW recursion relation. Our work also indicates that for the 2-split to hold, the current should be carefully defined.

hep-th↗

arXiv2025

LoRA-Based Continual Learning with Constraints on Critical Parameter Changes

LoRA-based continual learning represents a promising avenue for leveraging pre-trained models in downstream continual learning tasks. Recent studies have shown that orthogonal LoRA tuning effectively mitigates forgetting. However, this work unveils that under orthogonal LoRA tuning, the critical parameters for pre-tasks still change notably after learning post-tasks. To address this problem, we directly propose freezing the most critical parameter matrices in the Vision Transformer (ViT) for pre-tasks before learning post-tasks. In addition, building on orthogonal LoRA tuning, we propose orthogonal LoRA composition (LoRAC) based on QR decomposition, which may further enhance the plasticity of our method. Elaborate ablation studies and extensive comparisons demonstrate the effectiveness of our proposed method. Our results indicate that our method achieves state-of-the-art (SOTA) performance on several well-known continual learning benchmarks. For instance, on the Split CIFAR-100 dataset, our method shows a 6.35\% improvement in accuracy and a 3.24\% reduction in forgetting compared to previous methods. Our code is available at https://github.com/learninginvision/LoRAC-IPC.

cs.CV↗

arXiv2025

Saga: Capturing Multi-granularity Semantics from Massive Unlabelled IMU Data for User Perception

Inertial measurement units (IMUs), have been prevalently used in a wide range of mobile perception applications such as activity recognition and user authentication, where a large amount of labelled data are normally required to train a satisfactory model. However, it is difficult to label micro-activities in massive IMU data due to the hardness of understanding raw IMU data and the lack of ground truth. In this paper, we propose a novel fine-grained user perception approach, called Saga, which only needs a small amount of labelled IMU data to achieve stunning user perception accuracy. The core idea of Saga is to first pre-train a backbone feature extraction model, utilizing the rich semantic information of different levels embedded in the massive unlabelled IMU data. Meanwhile, for a specific downstream user perception application, Bayesian Optimization is employed to determine the optimal weights for pre-training tasks involving different semantic levels. We implement Saga on five typical mobile phones and evaluate Saga on three typical tasks on three IMU datasets. Results show that when only using about 100 training samples per class, Saga can achieve over 90% accuracy of the full-fledged model trained on over ten thousands training samples with no additional system overhead.

cs.LG↗

arXiv2025

Generative Data Imputation for Sparse Learner Performance Data Using Generative Adversarial Imputation Networks

Learner performance data collected by Intelligent Tutoring Systems (ITSs), such as responses to questions, is essential for modeling and predicting learners' knowledge states. However, missing responses due to skips or incomplete attempts create data sparsity, challenging accurate assessment and personalized instruction. To address this, we propose a generative imputation approach using Generative Adversarial Imputation Networks (GAIN). Our method features a three-dimensional (3D) framework (learners, questions, and attempts), flexibly accommodating various sparsity levels. Enhanced by convolutional neural networks and optimized with a least squares loss function, the GAIN-based method aligns input and output dimensions to question-attempt matrices along the learners' dimension. Extensive experiments using datasets from AutoTutor Adult Reading Comprehension (ARC), ASSISTments, and MATHia demonstrate that our approach significantly outperforms tensor factorization and alternative GAN methods in imputation accuracy across different attempt scenarios. Bayesian Knowledge Tracing (BKT) further validates the effectiveness of the imputed data by estimating learning parameters: initial knowledge (P(L0)), learning rate (P(T)), guess rate (P(G)), and slip rate (P(S)). Results indicate the imputed data enhances model fit and closely mirrors original distributions, capturing underlying learning behaviors reliably. Kullback-Leibler (KL) divergence assessments confirm minimal divergence, showing the imputed data preserves essential learning characteristics effectively. These findings underscore GAIN's capability as a robust imputation tool in ITSs, alleviating data sparsity and supporting adaptive, individualized instruction, ultimately leading to more precise and responsive learner assessments and improved educational outcomes.

cs.LG↗

arXiv2025

3D Heterogeneous Integration of Silicon Nitride and Aluminum Nitride on Sapphire toward Ultra-wideband Photonics Integrated Circuits

Extending two-dimensional photonic integrated circuits (PICs) to three-dimensional (3D) configurations promises great potential for scaling up integration, enhancing functionality, and improving performance of PICs. Silicon-based 3D PICs have made substantial progress due to CMOS compatibility. However, the narrow bandgap of silicon (1.1 eV) limits their use in short-wavelength applications, such as chemical and biological sensing, underwater optical communications, and optical atomic clocks. In this work, we developed a 3D photonics platform by heterogeneously integrating silicon nitride (SiN) and aluminum nitride (AlN) PICs on sapphire (Al$_2$O$_3$). The broadband transparency of these materials allow our platform to operate over a multi-octave wavelength ranging from ultraviolet to infrared. Leveraging this platform, we demonstrated efficient optical nonlinearity in an AlN microcavity, low-loss and tunable SiN waveguide-based optical components, and optical linking between AlN and SiN PICs layers in the visible and near-infrared spectrum, hinting at potential applications in integrated quantum systems. Our work presents an ultra-wideband 3D PICs platform, providing new opportunities for broadband and short-wavelength applications of PICs.

physics.optics↗

arXiv2025

Extending ring polymer molecular dynamics rate theory to reactions with non-separable reactants

The ring polymer molecular dynamics (RPMD) rate theory is an efficient and accurate method for estimating rate coefficients of chemical reactions affected by nuclear quantum effects. The commonly used RPMD treatment of gas-phase bimolecular reactions adopts two dividing surfaces, one at the transition state and another in the reactant asymptote, where partition functions of separated reactants can be readily obtained. With some exceptions, however, this strategy is difficult to implement for processes on surfaces or in liquids, because the reactants are often strongly coupled with the extended medium (surface or solvent) and thus non-separable. Under such circumstances, the RPMD rate theory with a single dividing surface (SDS) is better suited. However, most of its implementations adopted Cartesian forms of the reaction coordinate, which, in many cases, are not ideal for describing complex reactions. Here, we present a SDS-based RPMD implementation, which are able to tackle the aforementioned challenges. This approach is demonstrated in four representative reactions, including the gas phase H + H2 exchange reaction, gas phase CH3NC isomerization, H recombinative desorption from Pt(111), and NO desorption from Pd(111). This implementation, which is applicable to both uni- and bi-molecular reactions, offers a unified treatment of gas-phase and surface reaction rate calculations on the same footing.

physics.chem-ph↗