arXiv subjects

Yiwei Li

Publications and source records attributed to Yiwei Li.

At least 55 records · Page 3Linked to original sources

arXiv2025

CLAQS: Compact Learnable All-Quantum Token Mixer with Shared-ansatz for Text Classification

Quantum compute is scaling fast, from cloud QPUs to high throughput GPU simulators, making it timely to prototype quantum NLP beyond toy tasks. However, devices remain qubit limited and depth limited, training can be unstable, and classical attention is compute and memory heavy. This motivates compact, phase aware quantum token mixers that stabilize amplitudes and scale to long sequences. We present CLAQS, a compact, fully quantum token mixer for text classification that jointly learns complex-valued mixing and nonlinear transformations within a unified quantum circuit. To enable stable end-to-end optimization, we apply l1 normalization to regulate amplitude scaling and introduce a two-stage parameterized quantum architecture that decouples shared token embeddings from a window-level quantum feed-forward module. Operating under a sliding-window regime with document-level aggregation, CLAQS requires only eight data qubits and shallow circuits, yet achieves 91.64% accuracy on SST-2 and 87.08% on IMDB, outperforming both classical Transformer baselines and strong hybrid quantum-classical counterparts.

quant-ph↗

arXiv2025

InsBank: Evolving Instruction Subset for Ongoing Alignment

Large language models (LLMs) typically undergo instruction tuning to enhance alignment. Recent studies emphasize that quality and diversity of instruction data are more crucial than quantity, highlighting the need to select diverse, high-quality subsets to reduce training costs. However, how to evolve these selected subsets alongside the development of new instruction data remains insufficiently explored. To achieve LLMs' ongoing alignment, we introduce Instruction Bank (\textbf{InsBank}), a continuously updated repository that integrates the latest valuable instruction data. We further propose Progressive Instruction Bank Evolution (\textbf{PIBE}), a novel framework designed to evolve InsBank effectively and efficiently over time. PIBE employs a gradual data selection strategy to maintain long-term efficiency, leveraging a representation-based diversity score to capture relationships between data points and retain historical information for comprehensive diversity evaluation. This also allows for flexible combination of diversity and quality scores during data selection and ranking. Extensive experiments demonstrate that PIBE significantly outperforms baselines in InsBank evolution and is able to extract budget-specific subsets, demonstrating its effectiveness and adaptability.

cs.CL↗

arXiv2025

Memory Enhanced Fractional-Order Dung Beetle Optimization for Photovoltaic Parameter Identification

Accurate parameter identification in photovoltaic (PV) models is crucial for performance evaluation but remains challenging due to their nonlinear, multimodal, and high-dimensional nature. Although the Dung Beetle Optimization (DBO) algorithm has shown potential in addressing such problems, it often suffers from premature convergence. To overcome these issues, this paper proposes a Memory Enhanced Fractional-Order Dung Beetle Optimization (MFO-DBO) algorithm that integrates three coordinated strategies. Firstly, fractional-order (FO) calculus introduces memory into the search process, enhancing convergence stability and solution quality. Secondly, a fractional-order logistic chaotic map improves population diversity during initialization. Thirdly, a chaotic perturbation mechanism helps elite solutions escape local optima. Numerical results on the CEC2017 benchmark suite and the PV parameter identification problem demonstrate that MFO-DBO consistently outperforms advanced DBO variants, CEC competition winners, FO-based optimizers, enhanced classical algorithms, and recent metaheuristics in terms of accuracy, robustness, convergence speed, while also maintaining an excellent balance between exploration and exploitation compared to the standard DBO algorithm.

cs.NE↗

arXiv2025

Bridging Classical and Quantum Computing for Next-Generation Language Models

Integrating Large Language Models (LLMs) with quantum computing is a critical challenge, hindered by the severe constraints of Noisy Intermediate-Scale Quantum (NISQ) devices, including barren plateaus and limited coherence. Current approaches often fail due to static quantum-classical partitioning. We introduce Adaptive Quantum-Classical Fusion (AQCF), the first framework to bridge this gap through dynamic, quantum-classical co-design. AQCF's core principle is real-time adaptation: it analyzes input complexity to orchestrate seamless transitions between classical and quantum processing. The framework features three key innovations: (1) entropy-driven adaptive circuits that circumvent barren plateaus; (2) quantum memory banks that unify classical attention with quantum state-based similarity retrieval; and (3) intelligent fusion controllers that allocate tasks for optimal performance. This architecture maintains full compatibility with classical Transformers while progressively incorporating quantum advantages. Experiments on sentiment analysis demonstrate that AQCF achieves competitive performance, significantly improves quantum resource efficiency, and operates successfully within typical NISQ constraints. By providing a seamless integration pathway, AQCF offers both immediate practical value on current quantum hardware and a clear evolution path toward mature Quantum LLMs.

quant-ph↗

arXiv2025

Analysis of multivariate event times under informative censoring using vine copula

The study of times to nonterminal events of different types and their interrelation is a compelling area of interest. The primary challenge in analyzing such multivariate event times is the presence of informative censoring by the terminal event. While numerous statistical methods have been proposed for a single nonterminal event, i.e., semi-competing risks data, there remains a dearth of tools for analyzing times to multiple nonterminal events. This article introduces a novel analysis framework that leverages the vine copula to directly estimate the joint density of multivariate times to nonterminal and terminal events. Unlike the few existing methods based on multivariate or nested copulas, the developed approach excels in capturing the heterogeneous dependence between each pair of event times (nonterminal-terminal and between-nonterminal) in terms of strength and structure. We propose a likelihood-based estimation and inference procedure, which can be implemented efficiently in sequential stages. Through extensive simulation studies, we demonstrate the satisfactory finite-sample performance of our proposed stage-wise estimators and analytical variance estimators, as well as their advantages over existing methods. We apply the developed approach to data from a crowdfunding platform to investigate the relationship between various types of creator-backer interactions and a creator's lifetime on the platform.

stat.ME↗

arXiv2025

Argus: Leveraging Multiview Images for Improved 3-D Scene Understanding With Large Language Models

Advancements in foundation models have made it possible to conduct applications in various downstream tasks. Especially, the new era has witnessed a remarkable capability to extend Large Language Models (LLMs) for tackling tasks of 3D scene understanding. Current methods rely heavily on 3D point clouds, but the 3D point cloud reconstruction of an indoor scene often results in information loss. Some textureless planes or repetitive patterns are prone to omission and manifest as voids within the reconstructed 3D point clouds. Besides, objects with complex structures tend to introduce distortion of details caused by misalignments between the captured images and the dense reconstructed point clouds. 2D multi-view images present visual consistency with 3D point clouds and provide more detailed representations of scene components, which can naturally compensate for these deficiencies. Based on these insights, we propose Argus, a novel 3D multimodal framework that leverages multi-view images for enhanced 3D scene understanding with LLMs. In general, Argus can be treated as a 3D Large Multimodal Foundation Model (3D-LMM) since it takes various modalities as input(text instructions, 2D multi-view images, and 3D point clouds) and expands the capability of LLMs to tackle 3D tasks. Argus involves fusing and integrating multi-view images and camera poses into view-as-scene features, which interact with the 3D features to create comprehensive and detailed 3D-aware scene embeddings. Our approach compensates for the information loss while reconstructing 3D point clouds and helps LLMs better understand the 3D world. Extensive experiments demonstrate that our method outperforms existing 3D-LMMs in various downstream tasks.

cs.CV↗

arXiv2025

Sensitive infrared surface photovoltage in quasi-equilibrium in a layered semiconductor at low-intensity low-temperature condition

Benefit to layer-dependent bandgap, van der Waals materials with surface photovoltaic effect (SPV) enable photodetection over a tunable wavelength range with low power consumption. However, sensitive SPV in the infrared region, especially in a quasi-steady illumination condition, is still elusive in layered semiconductors. Here, using angle-resolved photoemission spectroscopy, we report a sensitive SPV in quasi-equilibrium in NbSi0.5Te2, with photoresponsivity up to 2.4*10^6 V/(W*cm^(-2)) at low intensity low temperature condition (LILT). The sensitive SPV is further confirmed by observing the Dember effect, where the photogenerated carrier density is high enough and diffusion currents suppress SPV. Temperature-dependent measurements indicate that intrinsic carriers freezing at low temperature leads to the ultrahigh photoresponse, while a small amount of photon-generated carriers in quasi-equilibrium dominate the system. Our work not only provides a promising layered semiconductor for Infrared optoelectronic devices with strong infrared SPV at LILT, which has application potential in fields such as quantum information and deep-space exploration, but also paves a novel way to enhance light-matter interaction effect by freezing bulk carriers.

cond-mat.mtrl-sci↗

arXiv2025

Observation of quasi-steady dark excitons and gap phase in a doped semiconductor

Exciton plays an important role in optics and optics-related behaviors and leads to novel correlated phases like charge order, exciton insulator, and exciton-polariton condensation. Dark exciton shows distinct properties from bright one. However, it cannot be directly detected by conventional optic measurements. The electronic modulation effect of dark excitons in quasi-equilibrium distribution, critical for electronic devices in working status, is still elusive. Here, using angle-resolved photoemission spectroscopy, we report creating, detecting, and controlling dark excitons in the quasi-equilibrium distribution in a doped semiconductor SnSe2. Surprisingly, we observe an excitonic gap phase, with a conduction band opening an anisotropic gap. Our results broaden the scope of dark excitons, extending their studies from the picosecond timescale in the ultrafast photoemission process to conditions occurring under quasi-equilibrium. We reveal the light-matter interaction in the engineering of electronic structures and provide a new way to realize the excitonic gap phase in semiconductors with large band gaps.

cond-mat.mtrl-sci↗

arXiv2025

SAMed-2: Selective Memory Enhanced Medical Segment Anything Model

Recent "segment anything" efforts show promise by learning from large-scale data, but adapting such models directly to medical images remains challenging due to the complexity of medical data, noisy annotations, and continual learning requirements across diverse modalities and anatomical structures. In this work, we propose SAMed-2, a new foundation model for medical image segmentation built upon the SAM-2 architecture. Specifically, we introduce a temporal adapter into the image encoder to capture image correlations and a confidence-driven memory mechanism to store high-certainty features for later retrieval. This memory-based strategy counters the pervasive noise in large-scale medical datasets and mitigates catastrophic forgetting when encountering new tasks or modalities. To train and evaluate SAMed-2, we curate MedBank-100k, a comprehensive dataset spanning seven imaging modalities and 21 medical segmentation tasks. Our experiments on both internal benchmarks and 10 external datasets demonstrate superior performance over state-of-the-art baselines in multi-task scenarios. The code is available at: https://github.com/ZhilingYan/Medical-SAM-Bench.

cs.CV↗

arXiv2025

Revisiting Self-Consistency from Dynamic Distributional Alignment Perspective on Answer Aggregation

Self-consistency improves reasoning by aggregating diverse stochastic samples, yet the dynamics behind its efficacy remain underexplored. We reframe self-consistency as a dynamic distributional alignment problem, revealing that decoding temperature not only governs sampling randomness but also actively shapes the latent answer distribution. Given that high temperatures require prohibitively large sample sizes to stabilize, while low temperatures risk amplifying biases, we propose a confidence-driven mechanism that dynamically calibrates temperature: sharpening the sampling distribution under uncertainty to align with high-probability modes, and promoting exploration when confidence is high. Experiments on mathematical reasoning tasks show this approach outperforms fixed-diversity baselines under limited samples, improving both average and best-case performance across varying initial temperatures without additional data or modules. This establishes self-consistency as a synchronization challenge between sampling dynamics and evolving answer distributions.

cs.CL↗

arXiv2025

From Sub-Ability Diagnosis to Human-Aligned Generation: Bridging the Gap for Text Length Control via MARKERGEN

Despite the rapid progress of large language models (LLMs), their length-controllable text generation (LCTG) ability remains below expectations, posing a major limitation for practical applications. Existing methods mainly focus on end-to-end training to reinforce adherence to length constraints. However, the lack of decomposition and targeted enhancement of LCTG sub-abilities restricts further progress. To bridge this gap, we conduct a bottom-up decomposition of LCTG sub-abilities with human patterns as reference and perform a detailed error analysis. On this basis, we propose MarkerGen, a simple-yet-effective plug-and-play approach that:(1) mitigates LLM fundamental deficiencies via external tool integration;(2) conducts explicit length modeling with dynamically inserted markers;(3) employs a three-stage generation scheme to better align length constraints while maintaining content quality. Comprehensive experiments demonstrate that MarkerGen significantly improves LCTG across various settings, exhibiting outstanding effectiveness and generalizability.

cs.CL↗

arXiv2025

Rapid Bone Scintigraphy Enhancement via Semantic Prior Distillation from Segment Anything Model

Rapid bone scintigraphy is crucial for diagnosing skeletal disorders and detecting tumor metastases in children, as it shortens scan duration and reduces discomfort. However, accelerated acquisition often degrades image quality, impairing the visibility of fine anatomical details and potentially compromising diagnosis. To overcome this limitation, we introduce the first application of SAM-based semantic priors for medical image restoration, utilizing the Segment Anything Model (SAM) to enhance pediatric rapid bone scintigraphy. Our approach employs two cascaded networks, $f^{IR1}$ and $f^{IR2}$, supported by three specialized modules: a Semantic Prior Integration (SPI) module, a Semantic Knowledge Distillation (SKD) module, and a Semantic Consistency Module (SCM). The SPI and SKD modules inject domain-specific semantic cues from a fine-tuned SAM, while the SCM preserves coherent semantic feature representations across both cascaded stages. Moreover, we present RBS, a novel Rapid Bone Scintigraphy dataset comprising paired standard (20 cm/min) and rapid (40 cm/min) scans from 137 pediatric patients aged 0.5 - 16 years, making it the first dataset tailored for pediatric rapid bone scintigraphy restoration. Extensive experiments on both a public endoscopic dataset and our RBS dataset demonstrate that our method consistently surpasses existing techniques in PSNR, SSIM, FID, and LPIPS metrics.

eess.IV↗

arXiv2025

Beyond One-Size-Fits-All: Tailored Benchmarks for Efficient Evaluation

Evaluating models on large benchmarks is very resource-intensive, especially during the period of rapid model evolution. Existing efficient evaluation methods estimate the performance of target models by testing them only on a small and static coreset of the benchmark, which is derived from the publicly available evaluation results of source models. These methods rely on the assumption that target models have high prediction consistency with source models. However, we demonstrate that it doesn't generalize well in practice. To alleviate the inconsistency issue, we present TailoredBench, a method that conducts customized evaluation tailored to each target model. Specifically, a Global-coreset is first constructed as a probe to identify the most consistent source models for each target model with an adaptive source model selection strategy. Afterwards, a scalable K-Medoids clustering algorithm is proposed to extend the Global-coreset to a tailored Native-coreset for each target model. According to the predictions on Native-coresets, we obtain the performance of target models on the whole benchmark with a calibrated estimation strategy. Comprehensive experiments on 5 benchmarks across over 300 models demonstrate that compared to best performing baselines, TailoredBench achieves an average reduction of 31.4% in MAE of accuracy estimates under the same inference budgets, showcasing strong effectiveness and generalizability.

cs.LG↗

arXiv2025

Mind the Quote: Enabling Quotation-Aware Dialogue in LLMs via Plug-and-Play Modules

Human-AI conversation frequently relies on quoting earlier text-"check it with the formula I just highlighted"-yet today's large language models (LLMs) lack an explicit mechanism for locating and exploiting such spans. We formalise the challenge as span-conditioned generation, decomposing each turn into the dialogue history, a set of token-offset quotation spans, and an intent utterance. Building on this abstraction, we introduce a quotation-centric data pipeline that automatically synthesises task-specific dialogues, verifies answer correctness through multi-stage consistency checks, and yields both a heterogeneous training corpus and the first benchmark covering five representative scenarios. To meet the benchmark's zero-overhead and parameter-efficiency requirements, we propose QuAda, a lightweight training-based method that attaches two bottleneck projections to every attention head, dynamically amplifying or suppressing attention to quoted spans at inference time while leaving the prompt unchanged and updating < 2.8% of backbone weights. Experiments across models show that QuAda is suitable for all scenarios and generalises to unseen topics, offering an effective, plug-and-play solution for quotation-aware dialogue.

cs.AI↗

arXiv2025

Silencer: From Discovery to Mitigation of Self-Bias in LLM-as-Benchmark-Generator

LLM-as-Benchmark-Generator methods have been widely studied as a supplement to human annotators for scalable evaluation, while the potential biases within this paradigm remain underexplored. In this work, we systematically define and validate the phenomenon of inflated performance in models evaluated on their self-generated benchmarks, referred to as self-bias, and attribute it to sub-biases arising from question domain, language style, and wrong labels. On this basis, we propose Silencer, a general framework that leverages the heterogeneity between multiple generators at both the sample and benchmark levels to neutralize bias and generate high-quality, self-bias-silenced benchmark. Experimental results across various settings demonstrate that Silencer can suppress self-bias to near zero, significantly improve evaluation effectiveness of the generated benchmark (with an average improvement from 0.655 to 0.833 in Pearson correlation with high-quality human-annotated benchmark), while also exhibiting strong generalizability.

cs.CL↗

arXiv2025

Communication-Efficient Personalized Distributed Learning with Data and Node Heterogeneity

To jointly tackle the challenges of data and node heterogeneity in decentralized learning, we propose a distributed strong lottery ticket hypothesis (DSLTH), based on which a communication-efficient personalized learning algorithm is developed. In the proposed method, each local model is represented as the Hadamard product of global real-valued parameters and a personalized binary mask for pruning. The local model is learned by updating and fusing the personalized binary masks while the real-valued parameters are fixed among different agents. To further reduce the complexity of hardware implementation, we incorporate a group sparse regularization term in the loss function, enabling the learned local model to achieve structured sparsity. Then, a binary mask aggregation algorithm is designed by introducing an intermediate aggregation tensor and adding a personalized fine-tuning step in each iteration, which constrains model updates towards the local data distribution. The proposed method effectively leverages the relativity among agents while meeting personalized requirements in heterogeneous node conditions. We also provide a theoretical proof for the DSLTH, establishing it as the foundation of the proposed method. Numerical simulations confirm the validity of the DSLTH and demonstrate the effectiveness of the proposed algorithm.

cs.LG↗

arXiv2025

Task-Specific Knowledge Distillation from the Vision Foundation Model for Enhanced Medical Image Segmentation

Large-scale pre-trained models, such as Vision Foundation Models (VFMs), have demonstrated impressive performance across various downstream tasks by transferring generalized knowledge, especially when target data is limited. However, their high computational cost and the domain gap between natural and medical images limit their practical application in medical segmentation tasks. Motivated by this, we pose the following important question: "How can we effectively utilize the knowledge of large pre-trained VFMs to train a small, task-specific model for medical image segmentation when training data is limited?" To address this problem, we propose a novel and generalizable task-specific knowledge distillation framework. Our method fine-tunes the VFM on the target segmentation task to capture task-specific features before distilling the knowledge to smaller models, leveraging Low-Rank Adaptation (LoRA) to reduce the computational cost of fine-tuning. Additionally, we incorporate synthetic data generated by diffusion models to augment the transfer set, enhancing model performance in data-limited scenarios. Experimental results across five medical image datasets demonstrate that our method consistently outperforms task-agnostic knowledge distillation and self-supervised pretraining approaches like MoCo v3 and Masked Autoencoders (MAE). For example, on the KidneyUS dataset, our method achieved a 28% higher Dice score than task-agnostic KD using 80 labeled samples for fine-tuning. On the CHAOS dataset, it achieved an 11% improvement over MAE with 100 labeled samples. These results underscore the potential of task-specific knowledge distillation to train accurate, efficient models for medical image segmentation in data-constrained settings.

cs.CV↗

arXiv2025

Speculative Decoding for Multi-Sample Inference

We propose a novel speculative decoding method tailored for multi-sample reasoning scenarios, such as self-consistency and Best-of-N sampling. Our method exploits the intrinsic consensus of parallel generation paths to synthesize high-quality draft tokens without requiring auxiliary models or external databases. By dynamically analyzing structural patterns across parallel reasoning paths through a probabilistic aggregation mechanism, it identifies consensus token sequences that align with the decoding distribution. Evaluations on mathematical reasoning benchmarks demonstrate a substantial improvement in draft acceptance rates over baselines, while reducing the latency in draft token construction. This work establishes a paradigm shift for efficient multi-sample inference, enabling seamless integration of speculative decoding with sampling-based reasoning techniques.

cs.CL↗