arXiv subjects

Yong Wang

Publications and source records attributed to Yong Wang.

At least 181 records · Page 10Linked to original sources

arXiv2025

Preparing Spin Squeezed States via Adaptive Genetic Algorithm

We introduce a novel strategy employing an adaptive genetic algorithm (GA) for iterative optimization of control sequences to generate quantum nonclassical states. Its efficacy is demonstrated by preparing spin-squeezed states in an open collective spin model governed by a linear control field. Inspired by Darwinian evolution, the algorithm iteratively refines control sequences using crossover, mutation, and elimination strategies, starting from a coherent spin state within a dissipative and dephasing environment. We rigorously benchmark our method against constant control protocols and reinforcement learning, demonstrating competitive and robust performance. Furthermore, we showcase the GA's versatility by directly optimizing for metrologically relevant squeezing, achieving scalable performance, even in the presence of dissipation and thermal noise. The proposed strategy demonstrates a high state-preparation fidelity, exceeding 0.99, and provides a long time window for maintaining the spin squeezed state, even under dissipative conditions. We discuss feasible experimental implementations and potential extensions to alternative quantum systems, and the adaptability of the GA module. This research establishes the foundation for utilizing GA-like strategies in controlling quantum systems and achieving desired nonclassical states.

quant-ph↗

arXiv2025

A Sequence-Form Characterization and Differentiable Path-Following Method for Computing Normal-Form Perfect Equilibria in Extensive-Form Games

The sequence form, owing to its compact and holistic strategy representation, has demonstrated significant efficiency in computing normal-form perfect equilibria for two-player extensive-form games with perfect recall. Nevertheless, the examination of $n$-player games remains underexplored. To tackle this challenge, we present a sequence-form characterization of normal-form perfect equilibria for $n$-player extensive-form games, achieved through a class of perturbed games formulated in sequence form. Based on this characterization, we develop a differentiable path-following method for computing normal-form perfect equilibria and prove its convergence. This method formulates an artificial logarithmic-barrier game in sequence form, introducing an additional variable to regulate the impact of logarithmic-barrier terms on the payoff functions, as well as the transition of the strategy space. We prove the existence of a smooth equilibrium path defined by the artificial game, starting from an arbitrary positive realization plan and converging to a normal-form perfect equilibrium of the original game as the additional variable approaches zero. Furthermore, we extend Harsanyi's linear and logarithmic tracing procedures to the sequence form and develop two alternative methods for computing normal-form perfect equilibria. Numerical experiments further substantiate the effectiveness and computational efficiency of our methods.

cs.GT↗

arXiv2025

AdaCuRL: Adaptive Curriculum Reinforcement Learning with Invalid Sample Mitigation and Historical Revisiting

Reinforcement learning (RL) has demonstrated considerable potential for enhancing reasoning in large language models (LLMs). However, existing methods suffer from Gradient Starvation and Policy Degradation when training directly on samples with mixed difficulty. To mitigate this, prior approaches leverage Chain-of-Thought (CoT) data, but the construction of high-quality CoT annotations remains labor-intensive. Alternatively, curriculum learning strategies have been explored but frequently encounter challenges, such as difficulty mismatch, reliance on manual curriculum design, and catastrophic forgetting. To address these issues, we propose AdaCuRL, a Adaptive Curriculum Reinforcement Learning framework that integrates coarse-to-fine difficulty estimation with adaptive curriculum scheduling. This approach dynamically aligns data difficulty with model capability and incorporates a data revisitation mechanism to mitigate catastrophic forgetting. Furthermore, AdaCuRL employs adaptive reference and sparse KL strategies to prevent Policy Degradation. Extensive experiments across diverse reasoning benchmarks demonstrate that AdaCuRL consistently achieves significant performance improvements on both LLMs and MLLMs.

cs.LG↗

arXiv2025

Where and What Matters: Sensitivity-Aware Task Vectors for Many-Shot Multimodal In-Context Learning

Large Multimodal Models (LMMs) have shown promising in-context learning (ICL) capabilities, but scaling to many-shot settings remains difficult due to limited context length and high inference cost. To address these challenges, task-vector-based methods have been explored by inserting compact representations of many-shot in-context demonstrations into model activations. However, existing task-vector-based methods either overlook the importance of where to insert task vectors or struggle to determine suitable values for each location. To this end, we propose a novel Sensitivity-aware Task Vector insertion framework (STV) to figure out where and what to insert. Our key insight is that activation deltas across query-context pairs exhibit consistent structural patterns, providing a reliable cue for insertion. Based on the identified sensitive-aware locations, we construct a pre-clustered activation bank for each location by clustering the activation values, and then apply reinforcement learning to choose the most suitable one to insert. We evaluate STV across a range of multimodal models (e.g., Qwen-VL, Idefics-2) and tasks (e.g., VizWiz, OK-VQA), demonstrating its effectiveness and showing consistent improvements over previous task-vector-based methods with strong generalization.

cs.AI↗

arXiv2025

MoSa: Motion Generation with Scalable Autoregressive Modeling

We introduce MoSa, a novel hierarchical motion generation framework for text-driven 3D human motion generation that enhances the Vector Quantization-guided Generative Transformers (VQ-GT) paradigm through a coarse-to-fine scalable generation process. In MoSa, we propose a Multi-scale Token Preservation Strategy (MTPS) integrated into a hierarchical residual vector quantization variational autoencoder (RQ-VAE). MTPS employs interpolation at each hierarchical quantization to effectively retain coarse-to-fine multi-scale tokens. With this, the generative transformer supports Scalable Autoregressive (SAR) modeling, which predicts scale tokens, unlike traditional methods that predict only one token at each step. Consequently, MoSa requires only 10 inference steps, matching the number of RQ-VAE quantization layers. To address potential reconstruction degradation from frequent interpolation, we propose CAQ-VAE, a lightweight yet expressive convolution-attention hybrid VQ-VAE. CAQ-VAE enhances residual block design and incorporates attention mechanisms to better capture global dependencies. Extensive experiments show that MoSa achieves state-of-the-art generation quality and efficiency, outperforming prior methods in both fidelity and speed. On the Motion-X dataset, MoSa achieves an FID of 0.06 (versus MoMask's 0.20) while reducing inference time by 27 percent. Moreover, MoSa generalizes well to downstream tasks such as motion editing, requiring no additional fine-tuning. The code is available at https://mosa-web.github.io/MoSa-web

cs.CV↗

arXiv2025

Non-Abelian Gauge Enhances Self-Healing for Non-Hermitian Su--Schrieffer--Heeger Chain

We investigate a non-Hermitian extension of the Su--Schrieffer--Heeger model that incorporates spin-dependent SU(2) gauge fields, represented by non-Abelian couplings between lattice sites, as well as independent nonreciprocal hopping amplitudes. This framework gives rise to a rich phase structure characterized by complex-energy braiding and tunable non-Hermitian skin effects. By employing the generalized Brillouin zone approach, we analyze the bulk-boundary correspondence and identify topological transitions protected by chiral symmetry. Notably, we demonstrate that non-Abelian gauge fields significantly enhance the dynamical resilience of the system, enabling robust self-healing under a moving scattering potential. These results clarify the role of SU(2) gauge fields in stabilizing non-Hermitian topological phases and indicate that the proposed model can be realized with currently available photonic, atomic, and superconducting experimental platforms.

cond-mat.mes-hall↗

arXiv2025

Density estimation for compositional data using nonparametric mixtures

Compositional data, representing proportions constrained to the simplex, arise in diverse fields such as geosciences, ecology, genomics, and microbiome research. Existing nonparametric density estimation methods often rely on transformations, which may induce substantial bias near the simplex boundary. We propose a nonparametric mixture-based framework for density estimation on compositions. Nonparametric Dirichlet mixtures are employed to naturally accommodate boundary values, thereby avoiding the transformation or zero-replacement, while also identifying components supported on the boundary, providing reliable estimates for data with zero or near-zero values. Bandwidth selection and initialization schemes are addressed. For comparison, nonparametric Gaussian mixtures, coupled with log-ratio transformations, are also considered. Extensive simulations show that the proposed estimators outperform existing approaches. Three real data applications, including GDP data analysis, handwritten digit recognition, and skin detection, demonstrate the usefulness of nonparametric Dirichlet mixtures in practice.

stat.ME↗

arXiv2025

HS-STaR: Hierarchical Sampling for Self-Taught Reasoners via Difficulty Estimation and Budget Reallocation

Self-taught reasoners (STaRs) enhance the mathematical reasoning abilities of large language models (LLMs) by leveraging self-generated responses for self-training. Recent studies have incorporated reward models to guide response selection or decoding, aiming to obtain higher-quality data. However, they typically allocate a uniform sampling budget across all problems, overlooking the varying utility of problems at different difficulty levels. In this work, we conduct an empirical study and find that problems near the boundary of the LLM's reasoning capability offer significantly greater learning utility than both easy and overly difficult ones. To identify and exploit such problems, we propose HS-STaR, a Hierarchical Sampling framework for Self-Taught Reasoners. Given a fixed sampling budget, HS-STaR first performs lightweight pre-sampling with a reward-guided difficulty estimation strategy to efficiently identify boundary-level problems. Subsequently, it dynamically reallocates the remaining budget toward these high-utility problems during a re-sampling phase, maximizing the generation of valuable training data. Extensive experiments across multiple reasoning benchmarks and backbone LLMs demonstrate that HS-STaR significantly outperforms other baselines without requiring additional sampling budget.

cs.AI↗

arXiv2025

Spectral Torsion for Nonminimal de-Rham Hodge Operator

In this paper, we investigate some new spectral torsion which is the extension of spectral torsion for Dirac operators, and compute the spectral torsion associated with nonminimal de Rham-Hodge operators on manifolds with (or without) boundary. The novelty of this paper lies in the calculation related to the noncommutative residue for the inverse of the Laplacion with non-scalar principle symbols.

math-ph↗

arXiv2025

A Chain-of-thought Reasoning Breast Ultrasound Dataset Covering All Histopathology Categories

Breast ultrasound (BUS) is an essential tool for diagnosing breast lesions, with millions of examinations per year. However, publicly available high-quality BUS benchmarks for AI development are limited in data scale and annotation richness. In this work, we present BUS-CoT, a BUS dataset for chain-of-thought (CoT) reasoning analysis, which contains 11,439 images of 10,019 lesions from 4,838 patients and covers all 99 histopathology types. To facilitate research on incentivizing CoT reasoning, we construct the reasoning processes based on observation, feature, diagnosis and pathology labels, annotated and verified by experienced experts. Moreover, by covering lesions of all histopathology types, we aim to facilitate robust AI systems in rare cases, which can be error-prone in clinical practice.

eess.IV↗

arXiv2025

Position Bias Mitigates Position Bias:Mitigate Position Bias Through Inter-Position Knowledge Distillation

Positional bias (PB), manifesting as non-uniform sensitivity across different contextual locations, significantly impairs long-context comprehension and processing capabilities. Previous studies have addressed PB either by modifying the underlying architectures or by employing extensive contextual awareness training. However, the former approach fails to effectively eliminate the substantial performance disparities, while the latter imposes significant data and computational overhead. To address PB effectively, we introduce \textbf{Pos2Distill}, a position to position knowledge distillation framework. Pos2Distill transfers the superior capabilities from advantageous positions to less favorable ones, thereby reducing the huge performance gaps. The conceptual principle is to leverage the inherent, position-induced disparity to counteract the PB itself. We identify distinct manifestations of PB under \textbf{\textsc{r}}etrieval and \textbf{\textsc{r}}easoning paradigms, thereby designing two specialized instantiations: \emph{Pos2Distill-R\textsuperscript{1}} and \emph{Pos2Distill-R\textsuperscript{2}} respectively, both grounded in this core principle. By employing the Pos2Distill approach, we achieve enhanced uniformity and significant performance gains across all contextual positions in long-context retrieval and reasoning tasks. Crucially, both specialized systems exhibit strong cross-task generalization mutually, while achieving superior performance on their respective tasks.

cs.CL↗

arXiv2025

Parametric convergence rate of a non-parametric estimator in multivariate mixtures of power series distributions under conditional independence

The conditional independence assumption has recently appeared in a growing body of literature on the estimation of multivariate mixtures. We consider here conditionally independent multivariate mixtures of power series distributions with infinite support, to which belong Poisson, Geometric or Negative Binomial mixtures. We show that for all these mixtures, the non-parametric maximum likelihood estimator converges to the truth at the rate $(\log (nd))^{1+d/2} n^{-1/2}$ in the Hellinger distance, where $n$ denotes the size of the observed sample and $d$ represents the dimension of the mixture. Using this result, we then construct a new non-parametric estimator based on the maximum likelihood estimator that converges with the parametric rate $n^{-1/2}$ in all $\ell_p$-distances, for $p \ge 1$. These convergences rates are supported by simulations and the theory is illustrated using the famous Vélib dataset of the bike sharing system of Paris. We also introduce a testing procedure for whether the conditional independence assumption is satisfied for a given sample. This testing procedure is applied for several multivariate mixtures, with varying levels of dependence, and is thereby shown to distinguish well between conditionally independent and dependent mixtures. Finally, we use this testing procedure to investigate whether conditional independence holds for Vélib dataset.

math.ST↗

arXiv2025

LSM-OPD: Boosting Scan in LSM-Trees by Enabling Direct Computing on Compressed Data

Scan-based operations, such as backstage compaction and value filtering, have emerged as the main bottleneck for LSM-Trees in supporting contemporary data-intensive applications. For slower external storage devices, such as HDD and SATA SSD, the scan performance is primarily limited by the I/O bandwidth (i.e., I/O bound) due to the substantial read/write amplifications in LSM-Trees. Recent adoption of high-performance storage devices, such as NVMe SSD, has transformed the main limitation to be compute-bound, emerging the impact of computational resource consumption caused by inefficient compactions and filtering. However, when the value size increases, the bottleneck for scan performance in fast devices gradually shifts towards the I/O bandwidth as well, and the overall throughput across all types of devices undergo a dramatic reduction. This paper addresses the core issues by proposing LSM-OPD, a Log-S tructured M erge-O rder- Preserving Dictionary encoding scheme that enables direct computing on compressed data within LSM-Trees. It first enables key-value-separated data flushing to disk in a densely encoded columnar layout, ideally with few bytes for a large string value (e.g., 1024 bytes), thereby significantly alleviating the frequent I/O requests caused by intensive scans. Then, it is capable of offloading the costly scan-based operations on large values, including compaction and value filtering, to lightweight dictionaries due to the order-preserving property. And SIMD-based vectorization can now be employed to maximize the evaluating performance on modern multi-core processors, further breaking the compute-bound limitations in LSM-trees. Extensive experiments demonstrate the superior efficiency of LSM-OPD in processing various workloads that involve intensive scan-based operations on diverse modern storage devices.

cs.DB↗

arXiv2025

Qualitative Study for LLM-assisted Design Study Process: Strategies, Challenges, and Roles

Design studies aim to create visualization solutions for real-world problems of different application domains. Recently, the emergence of large language models (LLMs) has introduced new opportunities to enhance the design study process, providing capabilities such as creative problem-solving, data handling, and insightful analysis. However, despite their growing popularity, there remains a lack of systematic understanding of how LLMs can effectively assist researchers in visualization-specific design studies. In this paper, we conducted a multi-stage qualitative study to fill this gap, involving 30 design study researchers from diverse backgrounds and expertise levels. Through in-depth interviews and carefully-designed questionnaires, we investigated strategies for utilizing LLMs, the challenges encountered, and the practices used to overcome them. We further compiled and summarized the roles that LLMs can play across different stages of the design study process. Our findings highlight practical implications to inform visualization practitioners, and provide a framework for leveraging LLMs to enhance the design study process in visualization research.

cs.HC↗

arXiv2025

Parametric convergence rate of some nonparametric estimators in mixtures of power series distributions

We consider the problem of estimating a mixture of power series distributions with infinite support, to which belong very well-known models such as Poisson, Geometric, Logarithmic or Negative Binomial probability mass functions. We consider the nonparametric maximum likelihood estimator (NPMLE) and show that, under very mild assumptions, it converges to the true mixture distribution $π_0$ at a rate no slower than $(\log n)^{3/2} n^{-1/2}$ in the Hellinger distance. Recent work on minimax lower bounds suggests that the logarithmic factor in the obtained Hellinger rate of convergence can not be improved, at least for mixtures of Poisson distributions. Furthermore, we construct nonparametric estimators that are based on the NPMLE and show that they converge to $π_0$ at the parametric rate $n^{-1/2}$ in the $\ell_p$-norm ($p \in [1, \infty]$ or $p \in [2, \infty])$: The weighted least squares and hybrid estimators. Simulations and a real data application are considered to assess the performance of all estimators we study in this paper and illustrate the practical aspect of the theory. The simulations results show that the NPMLE has the best performance in the Hellinger, $\ell_1$ and $\ell_2$ distances in all scenarios. Finally, to construct confidence intervals of the true mixture probability mass function, both the nonparametric and parametric bootstrap procedures are considered. Their performances are compared with respect to the coverage and length of the resulting intervals.

math.ST↗

arXiv2025

TrajLens: Visual Analysis for Constructing Cell Developmental Trajectories in Cross-Sample Exploration

Constructing cell developmental trajectories is a critical task in single-cell RNA sequencing (scRNA-seq) analysis, enabling the inference of potential cellular progression paths. However, current automated methods are limited to establishing cell developmental trajectories within individual samples, necessitating biologists to manually link cells across samples to construct complete cross-sample evolutionary trajectories that consider cellular spatial dynamics. This process demands substantial human effort due to the complex spatial correspondence between each pair of samples. To address this challenge, we first proposed a GNN-based model to predict cross-sample cell developmental trajectories. We then developed TrajLens, a visual analytics system that supports biologists in exploring and refining the cell developmental trajectories based on predicted links. Specifically, we designed the visualization that integrates features on cell distribution and developmental direction across multiple samples, providing an overview of the spatial evolutionary patterns of cell populations along trajectories. Additionally, we included contour maps superimposed on the original cell distribution data, enabling biologists to explore them intuitively. To demonstrate our system's performance, we conducted quantitative evaluations of our model with two case studies and expert interviews to validate its usefulness and effectiveness.

cs.CG↗

arXiv2025

Analyzing and Mitigating Repetitions in Trip Recommendation

Trip recommendation has emerged as a highly sought-after service over the past decade. Although current studies significantly understand human intention consistency, they struggle with undesired repetitive outcomes that need resolution. We make two pivotal discoveries using statistical analyses and experimental designs: (1) The occurrence of repetitions is intricately linked to the models and decoding strategies. (2) During training and decoding, adding perturbations to logits can reduce repetition. Motivated by these observations, we introduce AR-Trip (Anti Repetition for Trip Recommendation), which incorporates a cycle-aware predictor comprising three mechanisms to avoid duplicate Points-of-Interest (POIs) and demonstrates their effectiveness in alleviating repetition. Experiments on four public datasets illustrate that AR-Trip successfully mitigates repetition issues while enhancing precision.

cs.IR↗

arXiv2025

USP: Unified Self-Supervised Pretraining for Image Generation and Understanding

Recent studies have highlighted the interplay between diffusion models and representation learning. Intermediate representations from diffusion models can be leveraged for downstream visual tasks, while self-supervised vision models can enhance the convergence and generation quality of diffusion models. However, transferring pretrained weights from vision models to diffusion models is challenging due to input mismatches and the use of latent spaces. To address these challenges, we propose Unified Self-supervised Pretraining (USP), a framework that initializes diffusion models via masked latent modeling in a Variational Autoencoder (VAE) latent space. USP achieves comparable performance in understanding tasks while significantly improving the convergence speed and generation quality of diffusion models. Our code will be publicly available at https://github.com/AMAP-ML/USP.

cs.CV↗