arXiv subjects

Yu Sun

Publications and source records attributed to Yu Sun.

At least 127 records · Page 7Linked to original sources

arXiv2025

UltraUNet: Real-Time Ultrasound Tongue Segmentation for Diverse Linguistic and Imaging Conditions

Ultrasound tongue imaging (UTI) is a non-invasive and cost-effective tool for studying speech articulation, motor control, and related disorders. However, real-time tongue contour segmentation remains challenging due to low signal-to-noise ratios, imaging variability, and computational demands. We propose UltraUNet, a lightweight encoder-decoder architecture optimized for real-time segmentation of tongue contours in ultrasound images. UltraUNet incorporates domain-specific innovations such as lightweight Squeeze-and-Excitation blocks, Group Normalization for small-batch stability, and summation-based skip connections to reduce memory and computational overhead. It achieves 250 frames per second and integrates ultrasound-specific augmentations like denoising and blur simulation. Evaluations on 8 datasets demonstrate high accuracy and robustness, with single-dataset Dice = 0.855 and MSD = 0.993px, and cross-dataset Dice averaging 0.734 and 0.761. UltraUNet provides a fast, accurate solution for speech research, clinical diagnostics, and analysis of speech motor disorders.

cs.CV↗

arXiv2025

PRISM: Probabilistic and Robust Inverse Solver with Measurement-Conditioned Diffusion Prior for Blind Inverse Problems

Diffusion models are now commonly used to solve inverse problems in computational imaging. However, most diffusion-based inverse solvers require complete knowledge of the forward operator to be used. In this work, we introduce a novel probabilistic and robust inverse solver with measurement-conditioned diffusion prior (PRISM) to effectively address blind inverse problems. PRISM offers a technical advancement over current methods by incorporating a powerful measurement-conditioned diffusion model into a theoretically principled posterior sampling scheme. Experiments on blind image deblurring validate the effectiveness of the proposed method, demonstrating the superior performance of PRISM over state-of-the-art baselines in both image and blur kernel recovery.

eess.IV↗

arXiv2025

CritiQ: Mining Data Quality Criteria from Human Preferences

Language model heavily depends on high-quality data for optimal performance. Existing approaches rely on manually designed heuristics, the perplexity of existing models, training classifiers, or careful prompt engineering, which require significant expert experience and human annotation effort while introduce biases. We introduce CritiQ, a novel data selection method that automatically mines criteria from human preferences for data quality with only ~30 human-annotated pairs and performs efficient data selection. The main component, CritiQ Flow, employs a manager agent to evolve quality criteria and worker agents to make pairwise judgments. We build a knowledge base that extracts quality criteria from previous work to boost CritiQ Flow. Compared to perplexity- and classifier- based methods, verbal criteria are more interpretable and possess reusable value. After deriving the criteria, we train the CritiQ Scorer to give quality scores and perform efficient data selection. We demonstrate the effectiveness of our method in the code, math, and logic domains, achieving high accuracy on human-annotated test sets. To validate the quality of the selected data, we continually train Llama 3.1 models and observe improved performance on downstream tasks compared to uniform sampling. Ablation studies validate the benefits of the knowledge base and the reflection process. We analyze how criteria evolve and the effectiveness of majority voting.

cs.CL↗

arXiv2025

Understanding the Landscape of Ampere GPU Memory Errors

Graphics Processing Units (GPUs) have become a de facto solution for accelerating high-performance computing (HPC) applications. Understanding their memory error behavior is an essential step toward achieving efficient and reliable HPC systems. In this work, we present a large-scale cross-supercomputer study to characterize GPU memory reliability, covering three supercomputers - Delta, Polaris, and Perlmutter - all equipped with NVIDIA A100 GPUs. We examine error logs spanning 67.77 million GPU device-hours across 10,693 GPUs. We compare error rates and mean-time-between-errors (MTBE) and highlight both shared and distinct error characteristics among these three systems. Based on these observations and analyses, we discuss the implications and lessons learned, focusing on the reliable operation of supercomputers, the choice of checkpointing interval, and the comparison of reliability characteristics with those of previous-generation GPUs. Our characterization study provides valuable insights into fault-tolerant HPC system design and operation, enabling more efficient execution of HPC applications.

cs.DC↗

arXiv2025

Flexible Base Station Sleeping and Resource Allocation for Green Uplink Fully-Decoupled RAN

The fully-decoupled radio access network (FD-RAN) is an innovative architecture designed for next-generation mobile communication networks, featuring decoupled control and data planes as well as separated uplink and downlink transmissions. To further enhance energy efficiency, this paper explores a green approach to FD-RAN by incorporating adaptive base station (BS) sleeping and resource allocation. First, we introduce a holistic power consumption model and formulate a energy efficiency maximization problem for FD-RAN, involving joint optimization of user equipment (UE) association, BS sleeping, and power control. Subsequently, the optimization problem is decomposed into two subproblems. The first subproblem, involving UE power control, is solved using a successive lower-bound maximization approach based on Dinkelbach's algorithm. The second subproblem, addressing UE association and BS sleeping, is tackled via a modified, low-complexity many-to-many swap matching algorithm. Extensive simulation results demonstrate the superior effectiveness of FD-RAN with our proposed algorithms, revealing the sources of energy efficiency gains.

cs.NI↗

arXiv2025

Learning to (Learn at Test Time): RNNs with Expressive Hidden States

Self-attention performs well in long context but has quadratic complexity. Existing RNN layers have linear complexity, but their performance in long context is limited by the expressive power of their hidden states. We present a practical framework for instantiating sequence modeling layers with linear complexity and expressive hidden states. The key idea is to make the hidden state a machine learning model itself, and the update rule a step of self-supervised learning. Since the hidden state is updated by training even on test sequences, our layers are called Test-Time Training (TTT) layers. We consider two instantiations: TTT-Linear and TTT-MLP, whose hidden state is a linear model and a two-layer MLP respectively. We evaluate our instantiations at the scale of 125M to 1.3B parameters, comparing with a strong Transformer and Mamba, a modern RNN. Similar to Transformer, TTT-Linear and TTT-MLP can keep reducing perplexity by conditioning on more tokens, while Mamba cannot after 16k context. TTT-MLP still faces challenges in memory I/O, but shows larger potential in long context, pointing to a promising direction for future research.

cs.LG↗

arXiv2025

Orthogonal Finetuning for Direct Preference Optimization

DPO is an effective preference optimization algorithm. However, the DPO-tuned models tend to overfit on the dispreferred samples, manifested as overly long generations lacking diversity. While recent regularization approaches have endeavored to alleviate this issue by modifying the objective function, they achieved that at the cost of alignment performance degradation. In this paper, we innovatively incorporate regularization from the perspective of weight updating to curb alignment overfitting. Through the pilot experiment, we discovered that there exists a positive correlation between overfitting and the hyperspherical energy fluctuation. Hence, we introduce orthogonal finetuning for DPO via a weight-Rotated Preference Optimization (RoPO) method, which merely conducts rotational and magnitude-stretching updates on the weight parameters to maintain the hyperspherical energy invariant, thereby preserving the knowledge encoded in the angle between neurons. Extensive experiments demonstrate that our model aligns perfectly with human preferences while retaining the original expressive capacity using only 0.0086% of the trainable parameters, suggesting an effective regularization against overfitting. Specifically, RoPO outperforms DPO by up to 10 points on MT-Bench and by up to 2.8 points on AlpacaEval 2, while enhancing the generation diversity by an average of 6 points.

cs.CL↗

arXiv2025

Learning to Optimally Stop Diffusion Processes, with Financial Applications

We study optimal stopping for diffusion processes with unknown model primitives within the continuous-time reinforcement learning (RL) framework developed by Wang et al. (2020), and present applications to option pricing and portfolio choice. By penalizing the corresponding variational inequality formulation, we transform the stopping problem into a stochastic optimal control problem with two actions. We then randomize controls into Bernoulli distributions and add an entropy regularizer to encourage exploration. We derive a semi-analytical optimal Bernoulli distribution, based on which we devise RL algorithms using the martingale approach established in Jia and Zhou (2022a). We establish a policy improvement theorem and prove the fast convergence of the resulting policy iterations. We demonstrate the effectiveness of the algorithms in pricing finite-horizon American put options, solving Merton's problem with transaction costs, and scaling to high-dimensional optimal stopping problems. In particular, we show that both the offline and online algorithms achieve high accuracy in learning the value functions and characterizing the associated free boundaries.

math.OC↗

arXiv2025

An Intelligent Mobile Application to Monitor and Correct Sitting Posture Using Raspberry Pi and MediaPipe Pose Detection

Poor posture has become an increasingly prevalent concern due to students and workers spending extended amounts of time sitting at a desk. To address this issue, we developed PoseTrack, a mobile application that uses a Raspberry Pi Camera and Mediapipe Pose landmarks to monitor the userś posture and provide real time feedback. The system detects poor posture, including forward lean, slouching, hunched shoulders, crossed legs, etc. Some challenges we faced were obtaining posture data, transferring data from the Raspberry Pi to the App, and safely storing user data. We used a Flask server to pass data from the Raspberry Pi to the mobile application, Firebase to store user data, and the Flutter framework to create the app. To test the analysis system viability, we designed an experiment that tested the system accuracy across several different perspectives and postures. The results indicate that the system is able to effectively detect poor posture whenever the userś joints are not blocked by the table or their limbs. The results demonstrate the potential for the system to be further improved and used on a larger scale for poor posture monitoring.

cs.CY↗

arXiv2025

Graph-Structured Data Analysis of Component Failure in Autonomous Cargo Ships Based on Feature Fusion

To address the challenges posed by cascading reactions caused by component failures in autonomous cargo ships (ACS) and the uncertainties in emergency decision-making, this paper proposes a novel hybrid feature fusion framework for constructing a graph-structured dataset of failure modes. By employing an improved cuckoo search algorithm (HN-CSA), the literature retrieval efficiency is significantly enhanced, achieving improvements of 7.1% and 3.4% compared to the NSGA-II and CSA search algorithms, respectively. A hierarchical feature fusion framework is constructed, using Word2Vec encoding to encode subsystem/component features, BERT-KPCA to process failure modes/reasons, and Sentence-BERT to quantify the semantic association between failure impact and emergency decision-making. The dataset covers 12 systems, 1,262 failure modes, and 6,150 propagation paths. Validation results show that the GATE-GNN model achieves a classification accuracy of 0.735, comparable to existing benchmarks. Additionally, a silhouette coefficient of 0.641 indicates that the features are highly distinguishable. In the label prediction results, the Shore-based Meteorological Service System achieved an F1 score of 0.93, demonstrating high prediction accuracy. This paper not only provides a solid foundation for failure analysis in autonomous cargo ships but also offers reliable support for fault diagnosis, risk assessment, and intelligent decision-making systems. The link to the dataset is https://github.com/wojiufukele/Graph-Structured-about-CSA.

cs.LG↗

arXiv2025

MARL-MambaContour: Unleashing Multi-Agent Deep Reinforcement Learning for Active Contour Optimization in Medical Image Segmentation

We introduce MARL-MambaContour, the first contour-based medical image segmentation framework based on Multi-Agent Reinforcement Learning (MARL). Our approach reframes segmentation as a multi-agent cooperation task focused on generate topologically consistent object-level contours, addressing the limitations of traditional pixel-based methods which could lack topological constraints and holistic structural awareness of anatomical regions. Each contour point is modeled as an autonomous agent that iteratively adjusts its position to align precisely with the target boundary, enabling adaptation to blurred edges and intricate morphologies common in medical images. This iterative adjustment process is optimized by a contour-specific Soft Actor-Critic (SAC) algorithm, further enhanced with the Entropy Regularization Adjustment Mechanism (ERAM) which dynamically balance agent exploration with contour smoothness. Furthermore, the framework incorporates a Mamba-based policy network featuring a novel Bidirectional Cross-attention Hidden-state Fusion Mechanism (BCHFM). This mechanism mitigates potential memory confusion limitations associated with long-range modeling in state space models, thereby facilitating more accurate inter-agent information exchange and informed decision-making. Extensive experiments on five diverse medical imaging datasets demonstrate the state-of-the-art performance of MARL-MambaContour, highlighting its potential as an accurate and robust clinical application.

cs.CV↗

arXiv2025

Circle packings and hyperbolic surfaces of finite type

This paper constructs hyperbolic polyhedral metrics via circle packings. We introduce the curvature of circles as a parameter to include all three types of constant curvature curves in the hyperbolic geometry. This provides a unified approach to producing polyhedral metrics for surfaces of broader topological types. The combinatorial total geodesic curvature serves as an effective tool for establishing the existence and uniqueness of the packing.

math.GT↗

arXiv2025

Dataset and Benchmark for Enhancing Critical Retained Foreign Object Detection

Critical retained foreign objects (RFOs), including surgical instruments like sponges and needles, pose serious patient safety risks and carry significant financial and legal implications for healthcare institutions. Detecting critical RFOs using artificial intelligence remains challenging due to their rarity and the limited availability of chest X-ray datasets that specifically feature critical RFOs cases. Existing datasets only contain non-critical RFOs, like necklace or zipper, further limiting their utility for developing clinically impactful detection algorithms. To address these limitations, we introduce "Hopkins RFOs Bench", the first and largest dataset of its kind, containing 144 chest X-ray images of critical RFO cases collected over 18 years from the Johns Hopkins Health System. Using this dataset, we benchmark several state-of-the-art object detection models, highlighting the need for enhanced detection methodologies for critical RFO cases. Recognizing data scarcity challenges, we further explore image synthetic methods to bridge this gap. We evaluate two advanced synthetic image methods, DeepDRR-RFO, a physics-based method, and RoentGen-RFO, a diffusion-based method, for creating realistic radiographs featuring critical RFOs. Our comprehensive analysis identifies the strengths and limitations of each synthetic method, providing insights into effectively utilizing synthetic data to enhance model training. The Hopkins RFOs Bench and our findings significantly advance the development of reliable, generalizable AI-driven solutions for detecting critical RFOs in clinical chest X-rays.

eess.IV↗

arXiv2025

BeamLoRA: Beam-Constraint Low-Rank Adaptation

Due to the demand for efficient fine-tuning of large language models, Low-Rank Adaptation (LoRA) has been widely adopted as one of the most effective parameter-efficient fine-tuning methods. Nevertheless, while LoRA improves efficiency, there remains room for improvement in accuracy. Herein, we adopt a novel perspective to assess the characteristics of LoRA ranks. The results reveal that different ranks within the LoRA modules not only exhibit varying levels of importance but also evolve dynamically throughout the fine-tuning process, which may limit the performance of LoRA. Based on these findings, we propose BeamLoRA, which conceptualizes each LoRA module as a beam where each rank naturally corresponds to a potential sub-solution, and the fine-tuning process becomes a search for the optimal sub-solution combination. BeamLoRA dynamically eliminates underperforming sub-solutions while expanding the parameter space for promising ones, enhancing performance with a fixed rank. Extensive experiments across three base models and 12 datasets spanning math reasoning, code generation, and commonsense reasoning demonstrate that BeamLoRA consistently enhances the performance of LoRA, surpassing the other baseline methods.

cs.CL↗

arXiv2025

Curiosity-Driven Reinforcement Learning from Human Feedback

Reinforcement learning from human feedback (RLHF) has proven effective in aligning large language models (LLMs) with human preferences, but often at the cost of reduced output diversity. This trade-off between diversity and alignment quality remains a significant challenge. Drawing inspiration from curiosity-driven exploration in reinforcement learning, we introduce curiosity-driven RLHF (CD-RLHF), a framework that incorporates intrinsic rewards for novel states, alongside traditional sparse extrinsic rewards, to optimize both output diversity and alignment quality. We demonstrate the effectiveness of CD-RLHF through extensive experiments on a range of tasks, including text summarization and instruction following. Our approach achieves significant gains in diversity on multiple diversity-oriented metrics while maintaining alignment with human preferences comparable to standard RLHF. We make our code publicly available at https://github.com/ernie-research/CD-RLHF.

cs.CL↗

arXiv2025

A Lightweight Hybrid Dual Channel Speech Enhancement System under Low-SNR Conditions

Although deep learning based multi-channel speech enhancement has achieved significant advancements, its practical deployment is often limited by constrained computational resources, particularly in low signal-to-noise ratio (SNR) conditions. In this paper, we propose a lightweight hybrid dual-channel speech enhancement system that combines independent vector analysis (IVA) with a modified version of the dual-channel grouped temporal convolutional recurrent network (GTCRN). IVA functions as a coarse estimator, providing auxiliary information for both speech and noise, while the modified GTCRN further refines the speech quality. We investigate several modifications to ensure the comprehensive utilization of both original and auxiliary information. Experimental results demonstrate the effectiveness of the proposed system, achieving enhanced speech with minimal parameters and low computational complexity.

eess.AS↗

arXiv2025

PromptHMR: Promptable Human Mesh Recovery

Human pose and shape (HPS) estimation presents challenges in diverse scenarios such as crowded scenes, person-person interactions, and single-view reconstruction. Existing approaches lack mechanisms to incorporate auxiliary "side information" that could enhance reconstruction accuracy in such challenging scenarios. Furthermore, the most accurate methods rely on cropped person detections and cannot exploit scene context while methods that process the whole image often fail to detect people and are less accurate than methods that use crops. While recent language-based methods explore HPS reasoning through large language or vision-language models, their metric accuracy is well below the state of the art. In contrast, we present PromptHMR, a transformer-based promptable method that reformulates HPS estimation through spatial and semantic prompts. Our method processes full images to maintain scene context and accepts multiple input modalities: spatial prompts like bounding boxes and masks, and semantic prompts like language descriptions or interaction labels. PromptHMR demonstrates robust performance across challenging scenarios: estimating people from bounding boxes as small as faces in crowded scenes, improving body shape estimation through language descriptions, modeling person-person interactions, and producing temporally coherent motions in videos. Experiments on benchmarks show that PromptHMR achieves state-of-the-art performance while offering flexible prompt-based control over the HPS estimation process.

cs.CV↗

arXiv2025

Neural Inverse Scattering with Score-based Regularization

Inverse scattering is a fundamental challenge in many imaging applications, ranging from microscopy to remote sensing. Solving this problem often requires jointly estimating two unknowns -- the image and the scattering field inside the object -- necessitating effective image prior to regularize the inference. In this paper, we propose a regularized neural field (NF) approach which integrates the denoising score function used in score-based generative models. The neural field formulation offers convenient flexibility to performing joint estimation, while the denoising score function imposes the rich structural prior of images. Our results on three high-contrast simulated objects show that the proposed approach yields a better imaging quality compared to the state-of-the-art NF approach, where regularization is based on total variation.

eess.IV↗