arXiv subjects

Chang Sun

Publications and source records attributed to Chang Sun.

At least 55 records · Page 3Linked to original sources

arXiv2025

Sub-microsecond Transformers for Jet Tagging on FPGAs

We present the first sub-microsecond transformer implementation on an FPGA achieving competitive performance for state-of-the-art high-energy physics benchmarks. Transformers have shown exceptional performance on multiple tasks in modern machine learning applications, including jet tagging at the CERN Large Hadron Collider (LHC). However, their computational complexity prohibits use in real-time applications, such as the hardware trigger system of the collider experiments up until now. In this work, we demonstrate the first application of transformers for jet tagging on FPGAs, achieving $\mathcal{O}(100)$ nanosecond latency with superior performance compared to alternative baseline models. We leverage high-granularity quantization and distributed arithmetic optimization to fit the entire transformer model on a single FPGA, achieving the required throughput and latency. Furthermore, we add multi-head attention and linear attention support to hls4ml, making our work accessible to the broader fast machine learning community. This work advances the next-generation trigger systems for the High Luminosity LHC, enabling the use of transformers for real-time applications in high-energy physics and beyond.

physics.ins-det↗

arXiv2025

Enabling Granular Subgroup Level Model Evaluations by Generating Synthetic Medical Time Series

We present a novel framework for leveraging synthetic ICU time-series data not only to train but also to rigorously and trustworthily evaluate predictive models, both at the population level and within fine-grained demographic subgroups. Building on prior diffusion and VAE-based generators (TimeDiff, HealthGen, TimeAutoDiff), we introduce \textit{Enhanced TimeAutoDiff}, which augments the latent diffusion objective with distribution-alignment penalties. We extensively benchmark all models on MIMIC-III and eICU, on 24-hour mortality and binary length-of-stay tasks. Our results show that Enhanced TimeAutoDiff reduces the gap between real-on-synthetic and real-on-real evaluation (``TRTS gap'') by over 70\%, achieving $Δ_{TRTS} \leq 0.014$ AUROC, while preserving training utility ($Δ_{TSTR} \approx 0.01$). Crucially, for 32 intersectional subgroups, large synthetic cohorts cut subgroup-level AUROC estimation error by up to 50\% relative to small real test sets, and outperform them in 72--84\% of subgroups. This work provides a practical, privacy-preserving roadmap for trustworthy, granular model evaluation in critical care, enabling robust and reliable performance analysis across diverse patient populations without exposing sensitive EHR data, contributing to the overall trustworthiness of Medical AI.

cs.LG↗

arXiv2025

A Central Differential Flux with High-Order Dissipation for Robust Simulations of Transcritical Flows

The simulation of transcritical flows remains challenging due to strong thermodynamic nonlinearities that induce spurious pressure oscillations in conventional schemes.While primitive-variable formulations offer improved robustness under such conditions, they are always limited by energy conservation errors and the absence of systematic high-order treatments for numerical fluxes. In this paper, we introduce the Central Differential flux with High-Order Dissipation (CDHD), a novel numerical flux solver designed for primitive-variable discretization. This method combines a central flux for advection with a minimal, upwind-biased dissipation term to stabilize the simulation while maintaining formal accuracy. The dissipation term effectively suppresses oscillations and improves stability in transcritical flows. Compared to traditional primitive-variable approaches, CDHD reduces the energy conservation error in two order of magnitude. When incorporated into a hybrid framework with a conservative shock-capturing scheme, the method robustly handles both smooth transcritical phenomena and shock waves. Numerical tests validate the accuracy, stability, and energy-preserving capabilities of CDHD, demonstrating its potential as a reliable tool for complex real-gas flow simulations.

physics.flu-dyn↗

arXiv2025

Bilateral Collaboration with Large Vision-Language Models for Open Vocabulary Human-Object Interaction Detection

Open vocabulary Human-Object Interaction (HOI) detection is a challenging task that detects all triplets of interest in an image, even those that are not pre-defined in the training set. Existing approaches typically rely on output features generated by large Vision-Language Models (VLMs) to enhance the generalization ability of interaction representations. However, the visual features produced by VLMs are holistic and coarse-grained, which contradicts the nature of detection tasks. To address this issue, we propose a novel Bilateral Collaboration framework for open vocabulary HOI detection (BC-HOI). This framework includes an Attention Bias Guidance (ABG) component, which guides the VLM to produce fine-grained instance-level interaction features according to the attention bias provided by the HOI detector. It also includes a Large Language Model (LLM)-based Supervision Guidance (LSG) component, which provides fine-grained token-level supervision for the HOI detector by the LLM component of the VLM. LSG enhances the ability of ABG to generate high-quality attention bias. We conduct extensive experiments on two popular benchmarks: HICO-DET and V-COCO, consistently achieving superior performance in the open vocabulary and closed settings. The code will be released in Github.

cs.CV↗

arXiv2025

Adapting Lightweight Vision Language Models for Radiological Visual Question Answering

Recent advancements in vision-language systems have improved the accuracy of Radiological Visual Question Answering (VQA) Models. However, some challenges remain across each stage of model development: limited expert-labeled images hinders data procurement at scale; the intricate and nuanced patterns of radiological images make modeling inherently difficult; and the lack of evaluation evaluation efforts makes it difficult to identify cases where the model might be ill-conditioned. In this study, we fine-tune a lightweight 3B parameter vision-language model for Radiological VQA, demonstrating that small models, when appropriately tuned with curated data, can achieve robust performance across both open- and closed-ended questions. We propose a cost-effective training pipeline from synthetic question-answer pair generation to multi-stage fine-tuning on specialised radiological domain-targeted datasets (e.g., ROCO v2.0, MedPix v2.0). Our results show that despite operating at a fraction of the scale of state-of-the-art models such as LLaVA-Med, our model achieves promising performance given its small parameter size and the limited scale of training data. We introduce a lightweight saliency-based diagnostic tool that enables domain experts to inspect VQA model performance and identify ill-conditioned failure modes through saliency analysis.

cs.CV↗

arXiv2025

LPCM: Learning-based Predictive Coding for LiDAR Point Cloud Compression

Since the data volume of LiDAR point clouds is very huge, efficient compression is necessary to reduce their storage and transmission costs. However, existing learning-based compression methods do not exploit the inherent angular resolution of LiDAR and ignore the significant differences in the correlation of geometry information at different bitrates. The predictive geometry coding method in the geometry-based point cloud compression (G-PCC) standard uses the inherent angular resolution to predict the azimuth angles. However, it only models a simple linear relationship between the azimuth angles of neighboring points. Moreover, it does not optimize the quantization parameters for residuals on each coordinate axis in the spherical coordinate system. We propose a learning-based predictive coding method (LPCM) with both high-bitrate and low-bitrate coding modes. LPCM converts point clouds into predictive trees using the spherical coordinate system. In high-bitrate coding mode, we use a lightweight Long-Short-Term Memory-based predictive (LSTM-P) module that captures long-term geometry correlations between different coordinates to efficiently predict and compress the elevation angles. In low-bitrate coding mode, where geometry correlation degrades, we introduce a variational radius compression (VRC) module to directly compress the point radii. Then, we analyze why the quantization of spherical coordinates differs from that of Cartesian coordinates and propose a differential evolution (DE)-based quantization parameter selection method, which improves rate-distortion performance without increasing coding time. Experimental results on the LiDAR benchmark \textit{SemanticKITTI} and the MPEG-specified \textit{Ford} datasets show that LPCM outperforms G-PCC and other learning-based methods.

eess.IV↗

arXiv2025

DualStream Contextual Fusion Network: Efficient Target Speaker Extraction by Leveraging Mixture and Enrollment Interactions

Target speaker extraction focuses on extracting a target speech signal from an environment with multiple speakers by leveraging an enrollment. Existing methods predominantly rely on speaker embeddings obtained from the enrollment, potentially disregarding the contextual information and the internal interactions between the mixture and enrollment. In this paper, we propose a novel DualStream Contextual Fusion Network (DCF-Net) in the time-frequency (T-F) domain. Specifically, DualStream Fusion Block (DSFB) is introduced to obtain contextual information and capture the interactions between contextualized enrollment and mixture representation across both spatial and channel dimensions, and then rich and consistent representations are utilized to guide the extraction network for better extraction. Experimental results demonstrate that DCF-Net outperforms state-of-the-art (SOTA) methods, achieving a scale-invariant signal-to-distortion ratio improvement (SI-SDRi) of 21.6 dB on the benchmark dataset, and exhibits its robustness and effectiveness in both noise and reverberation scenarios. In addition, the wrong extraction results of our model, called target confusion problem, reduce to 0.4%, which highlights the potential of DCF-Net for practical applications.

cs.SD↗

arXiv2025

Towards Computer-Using Personal Agents

Computer-Using Agents (CUA) enable users to automate increasingly-complex tasks using graphical interfaces such as browsers. As many potential tasks require personal data, we propose Computer-Using Personal Agents (CUPAs) that have access to an external repository of the user's personal data. Compared with CUAs, CUPAs offer users better control of their personal data, the potential to automate more tasks involving personal data, better interoperability with external sources of data, and better capabilities to coordinate with other CUPAs in order to solve collaborative tasks involving the personal data of multiple users.

cs.HC↗

arXiv2025

Equivalent Conditions for Domination of $\mathrm{M}(2,\mathbb{C})$-sequences

It is well known that a $\mathrm{SL}(2,\mathbb{C})$-sequence is uniformly hyperbolic if and only it satisfies a uniform exponential growth condition. Similarly, for $\mathrm{GL}(2,\mathbb{C})$-sequences whose determinants are uniformly bounded away from zero, it has dominated splitting if and only if it satisfies a uniform exponential gap condition between the two singular values. Inspired by [QTZ], we provide a similar equivalent description in terms of singular values for $\mathrm{M}(2,\mathbb{C})$-sequences that admit dominated splitting. We also prove a version of the Avalanche Principle for such sequences.

math.DS↗

arXiv2024

X-CrossNet: A complex spectral mapping approach to target speaker extraction with cross attention speaker embedding fusion

Target speaker extraction (TSE) is a technique for isolating a target speaker's voice from mixed speech using auxiliary features associated with the target speaker. It is another attempt at addressing the cocktail party problem and is generally considered to have more practical application prospects than traditional speech separation methods. Although academic research in this area has achieved high performance and evaluation scores on public datasets, most models exhibit significantly reduced performance in real-world noisy or reverberant conditions. To address this limitation, we propose a novel TSE model, X-CrossNet, which leverages CrossNet as its backbone. CrossNet is a speech separation network specifically optimized for challenging noisy and reverberant environments, achieving state-of-the-art performance in tasks such as speaker separation under these conditions. Additionally, to enhance the network's ability to capture and utilize auxiliary features of the target speaker, we integrate a Cross-Attention mechanism into the global multi-head self-attention (GMHSA) module within each CrossNet block. This facilitates more effective integration of target speaker features with mixed speech features. Experimental results show that our method performs superior separation on the WSJ0-2mix and WHAMR! datasets, demonstrating strong robustness and stability.

cs.SD↗

arXiv2024

Empirical Privacy Evaluations of Generative and Predictive Machine Learning Models -- A review and challenges for practice

Synthetic data generators, when trained using privacy-preserving techniques like differential privacy, promise to produce synthetic data with formal privacy guarantees, facilitating the sharing of sensitive data. However, it is crucial to empirically assess the privacy risks associated with the generated synthetic data before deploying generative technologies. This paper outlines the key concepts and assumptions underlying empirical privacy evaluation in machine learning-based generative and predictive models. Then, this paper explores the practical challenges for privacy evaluations of generative models for use cases with millions of training records, such as data from statistical agencies and healthcare providers. Our findings indicate that methods designed to verify the correct operation of the training algorithm are effective for large datasets, but they often assume an adversary that is unrealistic in many scenarios. Based on the findings, we highlight a crucial trade-off between the computational feasibility of the evaluation and the level of realism of the assumed threat model. Finally, we conclude with ideas and suggestions for future research.

cs.LG↗

arXiv2024

Enhancing context models for point cloud geometry compression with context feature residuals and multi-loss

In point cloud geometry compression, context models usually use the one-hot encoding of node occupancy as the label, and the cross-entropy between the one-hot encoding and the probability distribution predicted by the context model as the loss function. However, this approach has two main weaknesses. First, the differences between contexts of different nodes are not significant, making it difficult for the context model to accurately predict the probability distribution of node occupancy. Second, as the one-hot encoding is not the actual probability distribution of node occupancy, the cross-entropy loss function is inaccurate. To address these problems, we propose a general structure that can enhance existing context models. We introduce the context feature residuals into the context model to amplify the differences between contexts. We also add a multi-layer perception branch, that uses the mean squared error between its output and node occupancy as a loss function to provide accurate gradients in backpropagation. We validate our method by showing that it can improve the performance of an octree-based model (OctAttention) and a voxel-based model (VoxelDNN) on the object point cloud datasets MPEG 8i and MVUB, as well as the LiDAR point cloud dataset SemanticKITTI.

eess.IV↗

arXiv2024

Enhancing octree-based context models for point cloud geometry compression with attention-based child node number prediction

In point cloud geometry compression, most octreebased context models use the cross-entropy between the onehot encoding of node occupancy and the probability distribution predicted by the context model as the loss. This approach converts the problem of predicting the number (a regression problem) and the position (a classification problem) of occupied child nodes into a 255-dimensional classification problem. As a result, it fails to accurately measure the difference between the one-hot encoding and the predicted probability distribution. We first analyze why the cross-entropy loss function fails to accurately measure the difference between the one-hot encoding and the predicted probability distribution. Then, we propose an attention-based child node number prediction (ACNP) module to enhance the context models. The proposed module can predict the number of occupied child nodes and map it into an 8- dimensional vector to assist the context model in predicting the probability distribution of the occupancy of the current node for efficient entropy coding. Experimental results demonstrate that the proposed module enhances the coding efficiency of octree-based context models.

eess.IV↗

arXiv2024

Generative AI for Synthetic Data Across Multiple Medical Modalities: A Systematic Review of Recent Developments and Challenges

This paper presents a comprehensive systematic review of generative models (GANs, VAEs, DMs, and LLMs) used to synthesize various medical data types, including imaging (dermoscopic, mammographic, ultrasound, CT, MRI, and X-ray), text, time-series, and tabular data (EHR). Unlike previous narrowly focused reviews, our study encompasses a broad array of medical data modalities and explores various generative models. Our search strategy queries databases such as Scopus, PubMed, and ArXiv, focusing on recent works from January 2021 to November 2023, excluding reviews and perspectives. This period emphasizes recent advancements beyond GANs, which have been extensively covered previously. The survey reveals insights from three key aspects: (1) Synthesis applications and purpose of synthesis, (2) generation techniques, and (3) evaluation methods. It highlights clinically valid synthesis applications, demonstrating the potential of synthetic data to tackle diverse clinical requirements. While conditional models incorporating class labels, segmentation masks and image translations are prevalent, there is a gap in utilizing prior clinical knowledge and patient-specific context, suggesting a need for more personalized synthesis approaches and emphasizing the importance of tailoring generative approaches to the unique characteristics of medical data. Additionally, there is a significant gap in using synthetic data beyond augmentation, such as for validation and evaluation of downstream medical AI models. The survey uncovers that the lack of standardized evaluation methodologies tailored to medical images is a barrier to clinical application, underscoring the need for in-depth evaluation approaches, benchmarking, and comparative studies to promote openness and collaboration.

cs.LG↗

arXiv2024

JEP-KD: Joint-Embedding Predictive Architecture Based Knowledge Distillation for Visual Speech Recognition

Visual Speech Recognition (VSR) tasks are generally recognized to have a lower theoretical performance ceiling than Automatic Speech Recognition (ASR), owing to the inherent limitations of conveying semantic information visually. To mitigate this challenge, this paper introduces an advanced knowledge distillation approach using a Joint-Embedding Predictive Architecture (JEPA), named JEP-KD, designed to more effectively utilize audio features during model training. Central to JEP-KD is the inclusion of a generative network within the embedding layer, which enhances the video encoder's capacity for semantic feature extraction and brings it into closer alignment with the audio features from a pre-trained ASR model's encoder. This approach aims to progressively reduce the performance gap between VSR and ASR. Moreover, a comprehensive multimodal, multistage training regimen for the JEP-KD framework is established, bolstering the robustness and efficacy of the training process. Experiment results demonstrate that JEP-KD significantly improves the performance of VSR models and demonstrates versatility across different VSR platforms, indicating its potential for broader application within other multimodal tasks.

cs.CV↗

arXiv2023

KDSM: An uplift modeling framework based on knowledge distillation and sample matching

Uplift modeling aims to estimate the treatment effect on individuals, widely applied in the e-commerce platform to target persuadable customers and maximize the return of marketing activities. Among the existing uplift modeling methods, tree-based methods are adept at fitting increment and generalization, while neural-network-based models excel at predicting absolute value and precision, and these advantages have not been fully explored and combined. Also, the lack of counterfactual sample pairs is the root challenge in uplift modeling. In this paper, we proposed an uplift modeling framework based on Knowledge Distillation and Sample Matching (KDSM). The teacher model is the uplift decision tree (UpliftDT), whose structure is exploited to construct counterfactual sample pairs, and the pairwise incremental prediction is treated as another objective for the student model. Under the idea of multitask learning, the student model can achieve better performance on generalization and even surpass the teacher. Extensive offline experiments validate the universality of different combinations of teachers and student models and the superiority of KDSM measured against the baselines. In online A/B testing, the cost of each incremental room night is reduced by 6.5\%.

cs.LG↗

arXiv2023

TAPS Responsibility Matrix: A tool for responsible data science by design

Data science is an interdisciplinary research area where scientists are typically working with data coming from different fields. When using and analyzing data, the scientists implicitly agree to follow standards, procedures, and rules set in these fields. However, guidance on the responsibilities of the data scientists and the other involved actors in a data science project is typically missing. While literature shows that novel frameworks and tools are being proposed in support of open-science, data reuse, and research data management, there are currently no frameworks that can fully express responsibilities of a data science project. In this paper, we describe the Transparency, Accountability, Privacy, and Societal Responsibility Matrix (TAPS-RM) as framework to explore social, legal, and ethical aspects of data science projects. TAPS-RM acts as a tool to provide users with a holistic view of their project beyond key outcomes and clarifies the responsibilities of actors. We map the developed model of TAPS-RM with well-known initiatives for open data (such as FACT, FAIR and Datasheets for datasets). We conclude that TAPS-RM is a tool to reflect on responsibilities at a data science project level and can be used to advance responsible data science by design.

cs.CY↗

arXiv2022

Fast Muon Tracking with Machine Learning Implemented in FPGA

In this work, we present a new approach for fast tracking on multiwire proportional chambers with neural networks. The tracking networks are developed and adapted for the first-level trigger at hadron collider experiments. We use Monte Carlo samples generated by Geant4 with a custom muon chamber, which resembles part of the thin gap chambers from the ATLAS experiment, for training and performance evaluations. The chamber has a total of seven gas gaps, where the first and last gas gaps are displaced by ~1.5 m. Each gas gap has 50 channels with a size of 18-20 mm. Two neural network models are developed and presented: a convolutional neural network and a neural network optimized for the detector configuration of this study. In the latter network, a convolution layer is provided for each of three groups formed from 2-3 gas gaps of the chamber, and the outputs are fed into multilayer perceptrons in sequence. Both networks are transformed into hardware description language and implemented in Virtex UltraScale+ FPGA. The angular resolution is 2 mrad, which is comparable to the maximum resolution of the detector estimated by the minimum chi2 method. The latency achieved by the implemented firmware is less than 100 ns, and the throughput rate is 160 MHz.

physics.ins-det↗