Search arXiv⌕ Search

arXiv subjects

Cheng Chen

Publications and source records attributed to Cheng Chen.

At least 163 records · Page 9Linked to original sources
arXiv2025

Generalized Category Discovery via Token Manifold Capacity Learning

Generalized category discovery (GCD) is essential for improving deep learning models' robustness in open-world scenarios by clustering unlabeled data containing both known and novel categories. Traditional GCD methods focus on minimizing intra-cluster variations, often sacrificing manifold capacity, which limits the richness of intra-class representations. In this paper, we propose a novel approach, Maximum Token Manifold Capacity (MTMC), that prioritizes maximizing the manifold capacity of class tokens to preserve the diversity and complexity of data. MTMC leverages the nuclear norm of singular values as a measure of manifold capacity, ensuring that the representation of samples remains informative and well-structured. This method enhances the discriminability of clusters, allowing the model to capture detailed semantic features and avoid the loss of critical information during clustering. Through theoretical analysis and extensive experiments on coarse- and fine-grained datasets, we demonstrate that MTMC outperforms existing GCD methods, improving both clustering accuracy and the estimation of category numbers. The integration of MTMC leads to more complete representations, better inter-class separability, and a reduction in dimensional collapse, establishing MTMC as a vital component for robust open-world learning. Code is in github.com/lytang63/MTMC.

cs.LG↗
arXiv2025

RouteNator: A Router-Based Multi-Modal Architecture for Generating Synthetic Training Data for Function Calling LLMs

This paper addresses fine-tuning Large Language Models (LLMs) for function calling tasks when real user interaction data is unavailable. In digital content creation tools, where users express their needs through natural language queries that must be mapped to API calls, the lack of real-world task-specific data and privacy constraints for training on it necessitate synthetic data generation. Existing approaches to synthetic data generation fall short in diversity and complexity, failing to replicate real-world data distributions and leading to suboptimal performance after LLM fine-tuning. We present a novel router-based architecture that leverages domain resources like content metadata and structured knowledge graphs, along with text-to-text and vision-to-text language models to generate high-quality synthetic training data. Our architecture's flexible routing mechanism enables synthetic data generation that matches observed real-world distributions, addressing a fundamental limitation of traditional approaches. Evaluation on a comprehensive set of real user queries demonstrates significant improvements in both function classification accuracy and API parameter selection. Models fine-tuned with our synthetic data consistently outperform traditional approaches, establishing new benchmarks for function calling tasks.

cs.LG↗
arXiv2025

Aquarius: A Family of Industry-Level Video Generation Models for Marketing Scenarios

This report introduces Aquarius, a family of industry-level video generation models for marketing scenarios designed for thousands-xPU clusters and models with hundreds of billions of parameters. Leveraging efficient engineering architecture and algorithmic innovation, Aquarius demonstrates exceptional performance in high-fidelity, multi-aspect-ratio, and long-duration video synthesis. By disclosing the framework's design details, we aim to demystify industrial-scale video generation systems and catalyze advancements in the generative video community. The Aquarius framework consists of five components: Distributed Graph and Video Data Processing Pipeline: Manages tens of thousands of CPUs and thousands of xPUs via automated task distribution, enabling efficient video data processing. Additionally, we are about to open-source the entire data processing framework named "Aquarius-Datapipe". Model Architectures for Different Scales: Include a Single-DiT architecture for 2B models and a Multimodal-DiT architecture for 13.4B models, supporting multi-aspect ratios, multi-resolution, and multi-duration video generation. High-Performance infrastructure designed for video generation model training: Incorporating hybrid parallelism and fine-grained memory optimization strategies, this infrastructure achieves 36% MFU at large scale. Multi-xPU Parallel Inference Acceleration: Utilizes diffusion cache and attention optimization to achieve a 2.35x inference speedup. Multiple marketing-scenarios applications: Including image-to-video, text-to-video (avatar), video inpainting and video personalization, among others. More downstream applications and multi-dimensional evaluation metrics will be added in the upcoming version updates.

cs.CV↗
arXiv2025

Beam test performance of a prototype muon trigger detector for the PSI muEDM experiment

We report on the performance evaluation of a prototype muon trigger detector for the PSI muEDM experiment, conducted as a proof-of-principle test at the $π$E1 beamline of the Paul Scherrer Institute (PSI) using \SI{27.5}{MeV/c} muons. The detector is designed to identify muons within the acceptance phase space of a compact storage solenoid and activate a pulsed magnetic kicker for muon storage; it was tested without the application of a magnetic field. It comprises a telescope made up of four scintillators in anticoincidence with a gate scintillator, all read out by silicon photomultipliers. The study focused on characterizing the detector's response to various muon trajectories and the light yield of its plastic scintillators. Experimental results demonstrated strong agreement with Geant4 Monte Carlo simulations that incorporate optical photon modeling, confirming the detector's concept and its potential for meeting the stringent requirements of the muEDM experiment.

physics.ins-det↗
arXiv2025

Enhancing Diagnostic Accuracy in Rare and Common Fundus Diseases with a Knowledge-Rich Vision-Language Model

Previous foundation models for fundus images were pre-trained with limited disease categories and knowledge base. Here we introduce a knowledge-rich vision-language model (RetiZero) that leverages knowledge from more than 400 fundus diseases. For RetiZero's pretraining, we compiled 341,896 fundus images paired with texts, sourced from public datasets, ophthalmic literature, and online resources, encompassing a diverse range of diseases across multiple ethnicities and countries. RetiZero exhibits remarkable performance in several downstream tasks, including zero-shot disease recognition, image-to-image retrieval, AI-assisted clinical diagnosis,few-shot fine-tuning, and internal- and cross-domain disease identification. In zero-shot scenarios, RetiZero achieves Top-5 accuracies of 0.843 for 15 diseases and 0.756 for 52 diseases. For image retrieval, it achieves Top-5 scores of 0.950 and 0.886 for the same sets, respectively. AI-assisted clinical diagnosis results show that RetiZero's Top-3 zero-shot performance surpasses the average of 19 ophthalmologists from Singapore, China, and the United States. RetiZero substantially enhances clinicians' accuracy in diagnosing fundus diseases, in particularly rare ones. These findings underscore the value of integrating the RetiZero into clinical settings, where various fundus diseases are encountered.

eess.IV↗
arXiv2025

CADCrafter: Generating Computer-Aided Design Models from Unconstrained Images

Creating CAD digital twins from the physical world is crucial for manufacturing, design, and simulation. However, current methods typically rely on costly 3D scanning with labor-intensive post-processing. To provide a user-friendly design process, we explore the problem of reverse engineering from unconstrained real-world CAD images that can be easily captured by users of all experiences. However, the scarcity of real-world CAD data poses challenges in directly training such models. To tackle these challenges, we propose CADCrafter, an image-to-parametric CAD model generation framework that trains solely on synthetic textureless CAD data while testing on real-world images. To bridge the significant representation disparity between images and parametric CAD models, we introduce a geometry encoder to accurately capture diverse geometric features. Moreover, the texture-invariant properties of the geometric features can also facilitate the generalization to real-world scenarios. Since compiling CAD parameter sequences into explicit CAD models is a non-differentiable process, the network training inherently lacks explicit geometric supervision. To impose geometric validity constraints, we employ direct preference optimization (DPO) to fine-tune our model with the automatic code checker feedback on CAD sequence quality. Furthermore, we collected a real-world dataset, comprised of multi-view images and corresponding CAD command sequence pairs, to evaluate our method. Experimental results demonstrate that our approach can robustly handle real unconstrained CAD images, and even generalize to unseen general objects.

cs.CV↗
arXiv2025

Investigation of the baryon time-like electromagnetic form factors in the electron-positron annihilation reactions

Based on the experimental measurements of the electron-positron annihilation reactions into a baryon ($B$) and anti-baryon ($\bar{B}$) pair, the electromagnetic form factors of hyperons in the time-like region can be investigated within the vector meson dominance model. The theoretical model parameters are determined by fitting them to the total cross sections of the process $e^+e^-\to B \bar{B}$, and it is found that the current experimental data on the baryon electromagnetic form factors in the time-like region can be well reproduced. In addition to the total cross sections, the electromagnetic form factors $G_E$ and $G_M$, and the charge radii of those baryons are also estimated, which are in agreement with the experimental data. On the other hand, we have also investigated the nonmonotonic behavior of the time-like baryon electromagnetic form factors, and it is found that the previously proposed periodic behaviour of the nucleon time-like electromagnetic form factor is not confirmed. However, we do observe the nonmonotonic structures in the line shape of the baryon effective form factors or the ratio $|G_E/G_M|$ for the charmed baryon $Λ^+_c$. These features can be naturally explained by incorporating contributions from excited vector states. More precise measurements of the $e^+e^-\to B \bar{B}$ reaction offer a valuable opportunity to probe the properties of excited vector states, which are at present poorly known. Additionally, comprehensive theoretical and experimental studies of baryon timelike electromagnetic form factors can provide critical insights into the underlying mechanisms of electron-positron annihilation processes.

hep-ph↗
arXiv2025

Transformer neural networks and quantum simulators: a hybrid approach for simulating strongly correlated systems

Owing to their great expressivity and versatility, neural networks have gained attention for simulating large two-dimensional quantum many-body systems. However, their expressivity comes with the cost of a challenging optimization due to the in general rugged and complicated loss landscape. Here, we present a hybrid optimization scheme for neural quantum states (NQS), involving a data-driven pretraining with numerical or experimental data and a second, Hamiltonian-driven optimization stage. By using both projective measurements from the computational basis as well as expectation values from other measurement configurations such as spin-spin correlations, our pretraining gives access to the sign structure of the state, yielding improved and faster convergence that is robust w.r.t. experimental imperfections and limited datasets. We apply the hybrid scheme to the ground state search for the 2D transverse field Ising model and dipolar XY model on $6\times 6$ and $10\times 10$ square lattices with a patched transformer wave function, using numerical data as well as experimental data from a programmable Rydberg quantum simulator [Chen et al., Nature 616 (2023)], and show that the information from a second measurement basis highly improves the performance. Our work paves the way for a reliable and efficient optimization of neural quantum states.

cond-mat.dis-nn↗
arXiv2025

Error Source Sensitivity Analysis in Model-Based Coherence Scanning Interferometry for Thin Film Metrology

In semiconductor manufacturing processes, silicon dioxide films are commonly used as barrier layers, insulating layers, and protective layers. Coherence scanning interferometry (CSI) offers thin film thickness measurements with a millimeter-scale field of view and micrometer-scale lateral resolution. When the film thickness is less than the coherence length of a CSI system, a model-based film thickness measurement method is typically employed, which relies on a priori information about the thin film and the instrument. This study quantitatively analyzes how the accuracy of a priori information would affect the accuracy of thickness measurement when using a model-based CSI method. The influence factors include camera noise, numerical aperture (NA), pupil apodization, light source spectrum, and thin film refractive index. A series of SiO$_2$/Si thin films with varying thicknesses are analyzed by combining simulation and experimental approaches. The results reveal that the accuracy of thickness measurements exhibits varying sensitivity to different a priori information. The refractive index of the thin film is identified as the most sensitive source of error, where 1\% deviation in refractive index may cause 1\% relative thickness error, whereas 5\% deviation in NA results in less than 1\% relative thickness error. The simulation and experimental results show good agreement, validating the correctness and effectiveness of this study.

physics.optics↗
arXiv2025

Prior-enhanced reflection spectra imaging with a white light interfermetry

We present a simple but effective method to extract the reflectivity spectra from the interference signals with a large NA white light interferometry. Numerical simulations and experiments have demonstrated the effectiveness of our method. Furthermore, some insights are also disclosed

physics.optics↗
arXiv2025

Optimal Approximate Matrix Multiplication over Sliding Windows

We explore the problem of approximate matrix multiplication (AMM) within the sliding window model, where algorithms utilize limited space to perform large-scale matrix multiplication in a streaming manner. This model has garnered increasing attention in the fields of machine learning and data mining due to its ability to handle time sensitivity and reduce the impact of outdated data. However, despite recent advancements, determining the optimal space bound for this problem remains an open question. In this paper, we introduce the DS-COD algorithm for AMM over sliding windows. This novel and deterministic algorithm achieves optimal performance regarding the space-error tradeoff. We provide theoretical error bounds and the complexity analysis for the proposed algorithm, and establish the corresponding space lower bound for the AMM sliding window problem. Additionally, we present an adaptive version of DS-COD, termed aDS-COD, which improves computational efficiency and demonstrates superior empirical performance. Extensive experiments conducted on both synthetic and real-world datasets validate our theoretical findings and highlight the practical effectiveness of our methods.

cs.DS↗
arXiv2025

Text-to-Image Rectified Flow as Plug-and-Play Priors

Large-scale diffusion models have achieved remarkable performance in generative tasks. Beyond their initial training applications, these models have proven their ability to function as versatile plug-and-play priors. For instance, 2D diffusion models can serve as loss functions to optimize 3D implicit models. Rectified flow, a novel class of generative models, enforces a linear progression from the source to the target distribution and has demonstrated superior performance across various domains. Compared to diffusion-based methods, rectified flow approaches surpass in terms of generation quality and efficiency, requiring fewer inference steps. In this work, we present theoretical and experimental evidence demonstrating that rectified flow based methods offer similar functionalities to diffusion models - they can also serve as effective priors. Besides the generative capabilities of diffusion priors, motivated by the unique time-symmetry properties of rectified flow models, a variant of our method can additionally perform image inversion. Experimentally, our rectified flow-based priors outperform their diffusion counterparts - the SDS and VSD losses - in text-to-3D generation. Our method also displays competitive performance in image inversion and editing.

cs.CV↗
arXiv2025

Can Post-Training Quantization Benefit from an Additional QLoRA Integration?

Large language models (LLMs) have transformed natural language processing but pose significant challenges for real-world deployment. These models necessitate considerable computing resources, which can be costly and frequently unavailable. Model compression techniques such as quantization are often leveraged to alleviate resource demand, but they may have a negative impact on the generation quality. In this study, we explore the integration of 4-bit Post-training Quantization (PTQ) with QLoRA to address these issues. We demonstrate through extensive experiments that this integration outperforms standard PTQ, and in some cases even 16-bit full-parameter fine-tuning on LLMs, validated across proprietary and public datasets with different quantization algorithms. The results demonstrate the efficacy of PTQ-QLoRA integration, offering a viable solution for deploying powerful LLMs in resource-constrained environments without compromising on performance.

cs.CL↗
arXiv2025

Cascading Bandits Robust to Adversarial Corruptions

Online learning to rank sequentially recommends a small list of items to users from a large candidate set and receives the users' click feedback. In many real-world scenarios, users browse the recommended list in order and click the first attractive item without checking the rest. Such behaviors are usually formulated as the cascade model. Many recent works study algorithms for cascading bandits, an online learning to rank framework in the cascade model. However, the performance of existing methods may drop significantly if part of the user feedback is adversarially corrupted (e.g., click fraud). In this work, we study how to resist adversarial corruptions in cascading bandits. We first formulate the ``\textit{Cascading Bandits with Adversarial Corruptions}" (CBAC) problem, which assumes that there is an adaptive adversary that may manipulate the user feedback. Then we propose two robust algorithms for this problem, which assume the corruption level is known and agnostic, respectively. We show that both algorithms can achieve logarithmic regret when the algorithm is not under attack, and the regret increases linearly with the corruption level. The experimental results also verify the robustness of our methods.

cs.LG↗
arXiv2025

EchoFM: Foundation Model for Generalizable Echocardiogram Analysis

Foundation models have recently gained significant attention because of their generalizability and adaptability across multiple tasks and data distributions. Although medical foundation models have emerged, solutions for cardiac imaging, especially echocardiography videos, are still unexplored. In this paper, we introduce EchoFM, a foundation model specifically designed to represent and analyze echocardiography videos. In EchoFM, we propose a self-supervised learning framework that captures both spatial and temporal variability patterns through a spatio-temporal consistent masking strategy and periodic-driven contrastive learning. This framework can effectively capture the spatio-temporal dynamics of echocardiography and learn the representative video features without any labels. We pre-train our model on an extensive dataset comprising over 290,000 echocardiography videos covering 26 scan views across different imaging modes, with up to 20 million frames of images. The pre-trained EchoFM can then be easily adapted and fine-tuned for a variety of downstream tasks, serving as a robust backbone model. Our evaluation was systemically designed for four downstream tasks after the echocardiography examination routine. Experiment results show that EchoFM surpasses state-of-the-art methods, including specialized echocardiography methods, self-supervised pre-training models, and general-purposed pre-trained foundation models, across all downstream tasks.

cs.CV↗
arXiv2025

Roles of $Δ(1232)$, $N^*(1520)$, and $N^*(1650)$ resonances in $γp\to π^0 π^0 p$ reaction within an effective Lagrangian approach

Roles of the $Δ(1232)$, ${N}^{*}(1520)$, and ${N}^{*}(1650)$ resonances in the $γp\to{π}^{0}{π}^{0}p $ reaction near threshold is investigated within an effective Lagrangian approach. We have calculated the differential cross sections of the $γp\to{π}^{0}{π}^{0}p$ reaction by including the contributions from the $Δ(1232)$, ${N}^{*}(1520)$, and ${N}^{*}(1650)$ intermediate states decaying into $π^0 p$ via the $s$-channel nucleon pole and $t$-channel $ρ$ exchange, and found that the current experimental measurements can be well reproduced. The production of $Δ(1232)$ is mainly from the mechanism of the $s$-channel nucleon pole, while the ${N}^{*}(1520)$ and ${N}^{*}(1650)$ are produced from the mechanism of the $t$-channel $ρ$ exchange. It is expected that more experimental data on the $γp \to π^0 π^0 p$ reaction can be used to explore the properties of the low-lying excited baryon states and also the scalar $f_0(500)$ and $f_0(980)$ mesons.

hep-ph↗
arXiv2025

Tomonaga-Luttinger Liquid Behavior in a Rydberg-encoded Spin Chain

Quantum fluctuations can disrupt long-range order in one-dimensional systems, and replace it with the universal paradigm of the Tomonaga-Luttinger liquid (TLL), a critical phase of matter characterized by power-law decaying correlations and linearly dispersing excitations. Using a Rydberg quantum simulator, we study how TLL physics manifests in the low-energy properties of a spin chain, interacting under either the ferromagnetic or the antiferromagnetic dipolar XY Hamiltonian. Following quasi-adiabatic preparation, we directly observe the power-law decay of spin-spin correlations in real-space, allowing us to extract the Luttinger parameter. In the presence of an impurity, the chain exhibits tunable Friedel oscillations of the local magnetization. Moreover, by utilizing a quantum quench, we directly probe the propagation of correlations, which exhibit a light-cone structure related to the linear sound mode of the underlying TLL. Our measurements demonstrate the influence of the long-range dipolar interactions, renormalizing the parameters of TLL with respect to the case of nearest-neighbor interactions. Finally, comparison to numerical simulations exposes the high sensitivity of TLLs to doping and finite-size effects.

quant-ph↗