arXiv subjects

Xin Su

Publications and source records attributed to Xin Su.

At least 37 records · Page 2Linked to original sources

arXiv2026

WARM-CAT: Warm-Started Test-Time Comprehensive Knowledge Accumulation for Compositional Zero-Shot Learning

Compositional Zero-Shot Learning (CZSL) aims to recognize novel attribute-object compositions based on the knowledge learned from seen ones. Existing methods suffer from performance degradation caused by the distribution shift of label space at test time, which stems from the inclusion of unseen compositions recombined from attributes and objects. To overcome the challenge, we propose a novel approach that accumulates comprehensive knowledge in both textual and visual modalities from unsupervised data to update multimodal prototypes at test time. Building on this, we further design an adaptive update weight to control the degree of prototype adjustment, enabling the model to flexibly adapt to distribution shift during testing. Moreover, a dynamic priority queue is introduced that stores high-confidence images to acquire visual prototypes from historical images for inference. Since the model tends to favor compositions already stored in the queue during testing, we warm-start the queue by initializing it with training images for visual prototypes of seen compositions and generating unseen visual prototypes using the mapping learned between seen and unseen textual prototypes. Considering the semantic consistency of multimodal knowledge, we align textual and visual prototypes by multimodal collaborative representation learning. To provide a more reliable evaluation for CZSL, we introduce a new benchmark dataset, C-Fashion, and refine the widely used but noisy MIT-States dataset. Extensive experiments indicate that our approach achieves state-of-the-art performance on four benchmark datasets under both closed-world and open-world settings. The source code and datasets are available at https://github.com/xud-yan/WARM-CAT .

cs.CV↗

arXiv2026

Enhanced Phase Estimation via Photon-Added Two-Mode Squeezed States and Kerr Nonlinearity

Quantum metrology employs quantum resources to achieve measurement precision beyond classical limits. This work investigates a Mach--Zehnder interferometer incorporating a Kerr nonlinear phase shifter, with photon-added two-mode squeezed coherent states generated via four-wave mixing as input. We demonstrate that increasing both the photon-addition order and the input resource strength systematically enhances phase sensitivity, quantum Fisher information, and the corresponding quantum Cramér--Rao bound. The proposed system not only surpasses the standard quantum limit but also approaches or exceeds the Heisenberg limit for linear phase shifts, while Kerr nonlinearity enables surpassing the super-Heisenberg limit. Furthermore, the scheme exhibits enhanced robustness against photon loss, providing a promising pathway toward practical high-precision quantum metrology applications.

quant-ph↗

arXiv2026

Climate-Adaptive and Cascade-Constrained Machine Learning Prediction for Sea Surface Height under Greenhouse Warming

Machine learning (ML) has achieved remarkable success in climate and marine science. Given that greenhouse warming fundamentally reshapes ocean conditions such as stratification, circulation patterns and eddy activity, evaluating the climate adaptability of the ML models is crucial. While physical constraints have been shown to enhance the performance of ML models, kinetic energy (KE) cascade has not been used as a constraint despite its importance in regulating multi-scale ocean motions. Here we develop two sea surface height (SSH) prediction models (with and without KE cascade constraint) and quantify their climate adaptability at the Kuroshio Extension. Both models exhibit only slight performance degradation under greenhouse warming conditions. Incorporating the KE cascade as a physical constraint significantly improves the model performance, reducing eddy kinetic energy errors by 14.7% in the present climate and 15.9% under greenhouse warming. Additional validations using satellite observations and in the Gulf Stream region further confirm the robustness of the proposed models. Compared with the KE spectrum constraint, both constraints improve the cross-scale transfer and spectrum of KE, but the KE cascade constraint yields larger improvements in the cross-scale transfer. This work presents the first application of the KE cascade as a physical constraint for ML-based ocean state prediction and demonstrates its robust adaptability across climates, offering guidance for the further development of global ML models for both present and future conditions.

physics.ao-ph↗

arXiv2025

Phase sensitivity of lossy Mach-Zehnder interferometer via photon addition operation

Photon addition operations applied to squeezed states have been shown to significantly enhance phase sensitivity. In this study, we extend this approach by applying photon addition not only to coherent states but also within a Mach--Zehnder interferometer setup, using coherent and squeezed vacuum states as input. Both intensity-difference and homodyne detection are used to evaluate photon addition schemes, and their phase sensitivities are compared under ideal and lossy conditions, respectively. We also analyze the quantum Fisher information of these two schemes. Results show both schemes improve phase sensitivity, quantum Fisher information, and loss resistance. In particular, photon addition within the interferometer performs better. Homodyne detection outperforms intensity difference detection under photon losses. Notably, each scheme has different parameter dependencies, making them suitable for different application scenarios. When the squeezing parameter is small, photon addition employed at the coherent input with intensity difference detection can approach the Heisenberg limit in ideal conditions and can exceed the standard quantum limit in high-loss conditions. Our proposed scheme represents a valuable method for quantum precision measurements.

quant-ph↗

arXiv2025

Phase estimation via photon subtraction at the output of the hybrid interferometer

The hybrid interferometer integrating an optical parametric amplifier and a beam splitter has the potential to outperform the SU(1,1) interferometer. However, photon loss remains a critical limitation for practical implementation. To address this challenge, we propose a quantum metrology scheme utilizing multi-photon subtraction at the output and replacing the conventional 50:50 beam splitter with a variable beam splitter to enhance robustness against photon loss. We employ a coherent state and a vacuum state as inputs and perform homodyne detection. Our results show that the selection of input modes significantly affects phase estimation, and optimizing the beam splitter's transmittance is crucial for maximizing phase sensitivity in lossy conditions. Furthermore, photon subtraction markedly improves phase sensitivity, quantum Fisher information, and robustness against noise. Our scheme achieves sensitivities beyond the Heisenberg limit even under 20% photon loss.

quant-ph↗

arXiv2025

LexSemBridge: Fine-Grained Dense Representation Enhancement through Token-Aware Embedding Augmentation

As queries in retrieval-augmented generation (RAG) pipelines powered by large language models (LLMs) become increasingly complex and diverse, dense retrieval models have demonstrated strong performance in semantic matching. Nevertheless, they often struggle with fine-grained retrieval tasks, where precise keyword alignment and span-level localization are required, even in cases with high lexical overlap that would intuitively suggest easier retrieval. To systematically evaluate this limitation, we introduce two targeted tasks, keyword retrieval and part-of-passage retrieval, designed to simulate practical fine-grained scenarios. Motivated by these observations, we propose LexSemBridge, a unified framework that enhances dense query representations through fine-grained, input-aware vector modulation. LexSemBridge constructs latent enhancement vectors from input tokens using three paradigms: Statistical (SLR), Learned (LLR), and Contextual (CLR), and integrates them with dense embeddings via element-wise interaction. Theoretically, we show that this modulation preserves the semantic direction while selectively amplifying discriminative dimensions. LexSemBridge operates as a plug-in without modifying the backbone encoder and naturally extends to both text and vision modalities. Extensive experiments across semantic and fine-grained retrieval tasks validate the effectiveness and generality of our approach. All code and models are publicly available at https://github.com/Jasaxion/LexSemBridge/

cs.IR↗

arXiv2025

Investigating the Robustness of Retrieval-Augmented Generation at the Query Level

Large language models (LLMs) are very costly and inefficient to update with new information. To address this limitation, retrieval-augmented generation (RAG) has been proposed as a solution that dynamically incorporates external knowledge during inference, improving factual consistency and reducing hallucinations. Despite its promise, RAG systems face practical challenges-most notably, a strong dependence on the quality of the input query for accurate retrieval. In this paper, we investigate the sensitivity of different components in the RAG pipeline to various types of query perturbations. Our analysis reveals that the performance of commonly used retrievers can degrade significantly even under minor query variations. We study each module in isolation as well as their combined effect in an end-to-end question answering setting, using both general-domain and domain-specific datasets. Additionally, we propose an evaluation framework to systematically assess the query-level robustness of RAG pipelines and offer actionable recommendations for practitioners based on the results of more than 1092 experiments we performed.

cs.CL↗

arXiv2025

A Semantic Parsing Framework for End-to-End Time Normalization

Time normalization is the task of converting natural language temporal expressions into machine-readable representations. It underpins many downstream applications in information retrieval, question answering, and clinical decision-making. Traditional systems based on the ISO-TimeML schema limit expressivity and struggle with complex constructs such as compositional, event-relative, and multi-span time expressions. In this work, we introduce a novel formulation of time normalization as a code generation task grounded in the SCATE framework, which defines temporal semantics through symbolic and compositional operators. We implement a fully executable SCATE Python library and demonstrate that large language models (LLMs) can generate executable SCATE code. Leveraging this capability, we develop an automatic data augmentation pipeline using LLMs to synthesize large-scale annotated data with code-level validation. Our experiments show that small, locally deployable models trained on this augmented data can achieve strong performance, outperforming even their LLM parents and enabling practical, accurate, and interpretable time normalization.

cs.CL↗

arXiv2025

Two-Phase Channel Estimation for RIS-Assisted THz Systems with Beam Split

Reconfigurable intelligent surface (RIS)-assisted terahertz (THz) communication is emerging as a key technology to support ultra-high data rates in future sixth-generation networks. However, the acquisition of accurate channel state information (CSI) in such systems is challenging due to the passive nature of RIS and the hybrid beamforming architecture typically employed in THz systems. To address these challenges, we propose a novel low-complexity two-phase channel estimation scheme for RIS-assisted THz systems with beam split effect. In the proposed scheme, we first estimate the full CSI over a small subset of subcarriers, then extract angular information at both the base station and RIS. Subsequently, we recover the full CSI across remaining subcarriers by determining the corresponding spatial directions and angle-excluded coefficients. Theoretical analysis and simulation results demonstrate that the proposed method achieves superior performance in terms of normalized mean-square error while significantly reducing computational complexity compared to existing algorithms.

cs.IT↗

arXiv2025

SK-VQA: Synthetic Knowledge Generation at Scale for Training Context-Augmented Multimodal LLMs

Multimodal retrieval augmented generation (RAG) plays a crucial role in domains such as knowledge-based visual question answering (KB-VQA), where external knowledge is needed to answer a question. However, existing multimodal LLMs (MLLMs) are not designed for context-augmented generation, limiting their effectiveness in such tasks. While synthetic data generation has recently gained attention for training MLLMs, its application for context-augmented generation remains underexplored. To address this gap, we introduce SK-VQA, a large-scale synthetic multimodal dataset containing over 2 million visual question-answer pairs, each associated with context documents containing information necessary to determine the final answer. Compared to previous datasets, SK-VQA contains 11x more unique questions, exhibits greater domain diversity, and covers a broader spectrum of image sources. Through human evaluations, we confirm the high quality of the generated question-answer pairs and their contextual relevance. Extensive experiments show that SK-VQA serves both as a challenging KB-VQA benchmark and as an effective training resource for adapting MLLMs to context-augmented generation. Our results further indicate that models trained on SK-VQA demonstrate enhanced generalization in both context-aware VQA and multimodal RAG settings. SK-VQA is publicly available via Hugging Face Hub.

cs.CL↗

arXiv2025

Two-parameter estimation via photon subtraction operation within a feedback-assisted interferometer

In this paper, we analyze how multi-photon subtraction operations in a feedback-assisted interferometer can enhance measurement precision for single-parameter and two-parameter estimation under both ideal and photon-loss conditions. We examine the effects of the feedback strength R, the optical parametric amplifier's gain g, the coherent state amplitude α, and the order of multi-photon subtraction on system performance. We demonstrate that an optimal feedback strength R_{opt} exists in both conditions. Selecting a suitable R can significantly boost the system's robustness to photon loss, and markedly improve measurement precision. And the photon subtraction operations within a feedback-assisted interferometer can further enhance measurement precision effectively. Additionally, we find that increasing intramode correlations while decreasing intermode correlations can improve estimation accuracy. This work investigates a new method through the synergistic integration of feedback topology and non-Gaussian operations into a multiparameter estimation system, along with their systematic study under both ideal and loss conditions. The findings may contribute to improving quantum-enhanced measurements and hold promise for high-precision quantum sensing research.

quant-ph↗

arXiv2025

DPO Learning with LLMs-Judge Signal for Computer Use Agents

Computer use agents (CUA) are systems that automatically interact with graphical user interfaces (GUIs) to complete tasks. CUA have made significant progress with the advent of large vision-language models (VLMs). However, these agents typically rely on cloud-based inference with substantial compute demands, raising critical privacy and scalability concerns, especially when operating on personal devices. In this work, we take a step toward privacy-preserving and resource-efficient agents by developing a lightweight vision-language model that runs entirely on local machines. To train this compact agent, we introduce an LLM-as-Judge framework that automatically evaluates and filters synthetic interaction trajectories, producing high-quality data for reinforcement learning without human annotation. Experiments on the OS-World benchmark demonstrate that our fine-tuned local model outperforms existing baselines, highlighting a promising path toward private, efficient, and generalizable GUI agents.

cs.AI↗

arXiv2025

CLEME2.0: Towards Interpretable Evaluation by Disentangling Edits for Grammatical Error Correction

The paper focuses on the interpretability of Grammatical Error Correction (GEC) evaluation metrics, which received little attention in previous studies. To bridge the gap, we introduce **CLEME2.0**, a reference-based metric describing four fundamental aspects of GEC systems: hit-correction, wrong-correction, under-correction, and over-correction. They collectively contribute to exposing critical qualities and locating drawbacks of GEC systems. Evaluating systems by combining these aspects also leads to superior human consistency over other reference-based and reference-less metrics. Extensive experiments on two human judgment datasets and six reference datasets demonstrate the effectiveness and robustness of our method, achieving a new state-of-the-art result. Our codes are released at https://github.com/THUKElab/CLEME.

cs.CL↗

arXiv2025

Conformal Robust Beamforming via Generative Channel Models

Traditional approaches to outage-constrained beamforming optimization rely on statistical assumptions about channel distributions and estimation errors. However, the resulting outage probability guarantees are only valid when these assumptions accurately reflect reality. This paper tackles the fundamental challenge of providing outage probability guarantees that remain robust regardless of specific channel or estimation error models. To achieve this, we propose a two-stage framework: (i) construction of a channel uncertainty set using a generative channel model combined with conformal prediction, and (ii) robust beamforming via the solution of a min-max optimization problem. The proposed method separates the modeling and optimization tasks, enabling principled uncertainty quantification and robust decision-making. Simulation results confirm the effectiveness and reliability of the framework in achieving model-agnostic outage guarantees.

eess.SP↗

arXiv2025

Improved linear and Kerr nonlinear phase estimation via photon addition operations

The accuracy of quantum measurements can be effectively improved by using both photon-added non-Gaussian operations and Kerr nonlinear phase shifters. Here, we employ coherent state mixed photon-added squeezed vacuum state as input into a Mach-Zehnder interferometer with parity detection, thereby achieving a significant enhancement in phase measurement accuracy. Our research focuses on phase sensitivity of linear phase shift under both ideal conditions and photon loss, as well as quantum Fisher information. The results demonstrate that employing the photon addition operations can markedly enhance phase sensitivity and quantum Fisher information, and the measurement accuracy can even approach the Heisenberg limit. In addition, we delve deeper into the scenario of replacing the linear phase shifter with a Kerr nonlinear one and systematically analyze the quantum Fisher information under both ideal and photon loss conditions. By comparison, it is evident that employing both the photon addition operations and the Kerr nonlinear phase shifter can further significantly enhance phase measurement accuracy while effectively improving the system's robustness against photon loss. These findings are instrumental in facilitating the development and practical application of quantum metrology.

quant-ph↗

arXiv2025

TiMo: Spatiotemporal Foundation Model for Satellite Image Time Series

Satellite image time series (SITS) provide continuous observations of the Earth's surface, making them essential for applications such as environmental management and disaster assessment. However, existing spatiotemporal foundation models rely on plain vision transformers, which encode entire temporal sequences without explicitly capturing multiscale spatiotemporal relationships between land objects. This limitation hinders their effectiveness in downstream tasks. To overcome this challenge, we propose TiMo, a novel hierarchical vision transformer foundation model tailored for SITS analysis. At its core, we introduce a spatiotemporal gyroscope attention mechanism that dynamically captures evolving multiscale patterns across both time and space. For pre-training, we curate MillionST, a large-scale dataset of one million images from 100,000 geographic locations, each captured across 10 temporal phases over five years, encompassing diverse geospatial changes and seasonal variations. Leveraging this dataset, we adapt masked image modeling to pre-train TiMo, enabling it to effectively learn and encode generalizable spatiotemporal representations.Extensive experiments across multiple spatiotemporal tasks-including deforestation monitoring, land cover segmentation, crop type classification, and flood detection-demonstrate TiMo's superiority over state-of-the-art methods. Code, model, and dataset will be released at https://github.com/MiliLab/TiMo.

cs.CV↗

arXiv2025

Msmsfnet: a multi-stream and multi-scale fusion net for edge detection

Edge detection is a long-standing problem in computer vision. Despite the efficiency of existing algorithms, their performance, however, rely heavily on the pre-trained weights of the backbone network on the ImageNet dataset. The use of pre-trained weights in previous methods significantly increases the difficulty to design new models for edge detection without relying on existing well-trained ImageNet models, as pre-training the model on the ImageNet dataset is expensive and becomes compulsory to ensure the fairness of comparison. Besides, the pre-training and fine-tuning strategy is not always useful and sometimes even inaccessible. For instance, the pre-trained weights on the ImageNet dataset are unlikely to be helpful for edge detection in Synthetic Aperture Radar (SAR) images due to strong differences in the statistics between optical images and SAR images. Moreover, no dataset has comparable size to the ImageNet dataset for SAR image processing. In this work, we study the performance achievable by state-of-the-art deep learning based edge detectors in publicly available datasets when they are trained from scratch, and devise a new network architecture, the multi-stream and multi-scale fusion net (msmsfnet), for edge detection. We show in our experiments that by training all models from scratch, our model outperforms state-of-the-art edge detectors in three publicly available datasets. We also demonstrate the efficiency of our model for edge detection in SAR images, where no useful pre-trained weight is available. Finally, We show that our model is able to achieve competitive performance on the BSDS500 dataset when the pre-trained weights are used.

cs.CV↗

arXiv2025

Intensification of Oceanic Inverse Energy Cascade Under Global Warming

Kinetic energy (KE) cascade in the turbulent ocean is pivotal in connecting diverse scales of oceanic motions, redistributing energy, and influencing ocean circulation and climate variability. However, its response to global warming remains poorly understood. Using a 24-year satellite altimetry dataset, we identify a pronounced intensification of inverse geostrophic kinetic energy cascade at the sea surface across most ocean regions during 1994-2017, with cascade amplitude increasing by 1% to 2% per decade. This intensification occurs not only in energetic regions but also in expansive quiescent areas. Contributing factors to this intensification of geostrophic KE cascade include enhanced vertical shear of horizontal velocity, deepened mixed layer, strengthened stratification, weakened eddy killing as well as changes in the KE budget. The dominant factors vary across regions. Our findings offer new insights into the ocean's response to global warming and improve understanding of feedback mechanisms for ocean circulation changes.

physics.geo-ph↗