Selected work

Representative Papers

BDPM: A Machine Learning-Based Feature Extractor for Parkinson's Disease Classification via Gut Microbiota Analysis

Sep 09, 2025

High clinical misdiagnosis rates persist in Parkinson’s disease (PD), and existing deep learning models leveraging gut microbiome data predominantly rely on single classifiers, neglecting ecological interdependencies among microbial taxa and longitudinal dynamics, while suffering from limited feature interpretability. To address these limitations, we propose RFRE—a biologically informed feature selection framework integrating random forests with recursive feature elimination, explicitly incorporating microbial ecological priors to enhance biological interpretability. Furthermore, we design a spatiotemporal-aware hybrid classifier that jointly models cross-sectional compositional differences and longitudinal temporal evolution of microbial abundances. Evaluated on a cohort of 39 PD patient–healthy spouse pairs, our approach achieves significantly improved classification accuracy and robustly identifies discriminative microbial taxa. This work establishes a novel paradigm for early, interpretable, microbiome-based PD diagnosis.

0 citationsRead paper

Tackling Data Heterogeneity in Federated Learning through Knowledge Distillation with Inequitable Aggregation

Jun 25, 2025

In large-scale federated learning, partial client participation exacerbates data heterogeneity—such as label and quantity skew—leading to degraded model performance. To address this, we propose KDIA, a framework synergizing knowledge distillation with imbalanced aggregation. Its core contributions are: (1) a weighted teacher-model aggregation mechanism that incorporates client participation frequency, count, and local dataset size; (2) a server-side generator that synthesizes near-IID features to facilitate robust teacher–student knowledge transfer; and (3) a unified training objective integrating knowledge distillation, self-distillation, and GAN-based feature generation to enhance generalization under heterogeneity. Extensive experiments on CIFAR-10, CIFAR-100, and CINIC-10 demonstrate that KDIA achieves significantly higher accuracy than baselines under low participation rates and strong data heterogeneity, requiring fewer communication rounds. Notably, performance gains increase with the degree of heterogeneity, confirming KDIA’s effectiveness in challenging real-world FL settings.

0 citationsRead paper

Leveraging Static Relationships for Intra-Type and Inter-Type Message Passing in Video Question Answering

Apr 03, 2025

Existing VideoQA methods suffer from inaccurate and insufficient modeling of static relationships, limiting fine-grained spatiotemporal reasoning. To address this, we propose a dual-graph collaborative modeling framework: (1) an intra-type dual graph enabling fine-grained message passing among objects and relations of the same category; and (2) an inter-type heterogeneous graph explicitly modeling cross-category static relational interactions. We introduce the first static-relationship-driven bidirectional message-passing inference paradigm, integrating graph neural networks, multi-granularity graph updates, and static relationship embedding. Our approach achieves significant accuracy improvements on ANetQA and Next-QA, demonstrating that static-relationship-guided joint graph reasoning substantially enhances video understanding.

0 citationsRead paper

Optimization of BLE Broadcast Mode in Offline Finding Network

Apr 02, 2025

To address high neighbor discovery latency and low success rates in Offline Finding Networks (OFNs) based on Bluetooth Low Energy (BLE), this paper proposes CPBIS—a novel, systematic dual-advertising-interval and interval-ratio optimization mechanism tailored for multi-interval advertisers. CPBIS overcomes the limitations of conventional single-fixed-interval approaches by jointly modeling advertising and scanning timing schedules, integrating probabilistic analysis with an efficient parameter search algorithm. Implemented on the nRF52832 platform, CPBIS achieves co-optimization of discovery latency and success rate. Experimental results demonstrate that CPBIS significantly reduces average discovery latency across diverse scanning modes while improving measured discovery success rate by 27%. Furthermore, end-to-end object-finding responsiveness becomes more stable and reliable, enhancing overall OFN performance.

0 citationsRead paper

Recent publications

Latest Papers

BDPM: A Machine Learning-Based Feature Extractor for Parkinson's Disease Classification via Gut Microbiota Analysis

Sep 09, 2025

High clinical misdiagnosis rates persist in Parkinson’s disease (PD), and existing deep learning models leveraging gut microbiome data predominantly rely on single classifiers, neglecting ecological interdependencies among microbial taxa and longitudinal dynamics, while suffering from limited feature interpretability. To address these limitations, we propose RFRE—a biologically informed feature selection framework integrating random forests with recursive feature elimination, explicitly incorporating microbial ecological priors to enhance biological interpretability. Furthermore, we design a spatiotemporal-aware hybrid classifier that jointly models cross-sectional compositional differences and longitudinal temporal evolution of microbial abundances. Evaluated on a cohort of 39 PD patient–healthy spouse pairs, our approach achieves significantly improved classification accuracy and robustly identifies discriminative microbial taxa. This work establishes a novel paradigm for early, interpretable, microbiome-based PD diagnosis.

0 citationsRead paper

Tackling Data Heterogeneity in Federated Learning through Knowledge Distillation with Inequitable Aggregation

Jun 25, 2025

In large-scale federated learning, partial client participation exacerbates data heterogeneity—such as label and quantity skew—leading to degraded model performance. To address this, we propose KDIA, a framework synergizing knowledge distillation with imbalanced aggregation. Its core contributions are: (1) a weighted teacher-model aggregation mechanism that incorporates client participation frequency, count, and local dataset size; (2) a server-side generator that synthesizes near-IID features to facilitate robust teacher–student knowledge transfer; and (3) a unified training objective integrating knowledge distillation, self-distillation, and GAN-based feature generation to enhance generalization under heterogeneity. Extensive experiments on CIFAR-10, CIFAR-100, and CINIC-10 demonstrate that KDIA achieves significantly higher accuracy than baselines under low participation rates and strong data heterogeneity, requiring fewer communication rounds. Notably, performance gains increase with the degree of heterogeneity, confirming KDIA’s effectiveness in challenging real-world FL settings.

0 citationsRead paper

Leveraging Static Relationships for Intra-Type and Inter-Type Message Passing in Video Question Answering

Apr 03, 2025

Existing VideoQA methods suffer from inaccurate and insufficient modeling of static relationships, limiting fine-grained spatiotemporal reasoning. To address this, we propose a dual-graph collaborative modeling framework: (1) an intra-type dual graph enabling fine-grained message passing among objects and relations of the same category; and (2) an inter-type heterogeneous graph explicitly modeling cross-category static relational interactions. We introduce the first static-relationship-driven bidirectional message-passing inference paradigm, integrating graph neural networks, multi-granularity graph updates, and static relationship embedding. Our approach achieves significant accuracy improvements on ANetQA and Next-QA, demonstrating that static-relationship-guided joint graph reasoning substantially enhances video understanding.

0 citationsRead paper

Optimization of BLE Broadcast Mode in Offline Finding Network

Apr 02, 2025

To address high neighbor discovery latency and low success rates in Offline Finding Networks (OFNs) based on Bluetooth Low Energy (BLE), this paper proposes CPBIS—a novel, systematic dual-advertising-interval and interval-ratio optimization mechanism tailored for multi-interval advertisers. CPBIS overcomes the limitations of conventional single-fixed-interval approaches by jointly modeling advertising and scanning timing schedules, integrating probabilistic analysis with an efficient parameter search algorithm. Implemented on the nRF52832 platform, CPBIS achieves co-optimization of discovery latency and success rate. Experimental results demonstrate that CPBIS significantly reduces average discovery latency across diverse scanning modes while improving measured discovery success rate by 27%. Furthermore, end-to-end object-finding responsiveness becomes more stable and reliable, enhancing overall OFN performance.

0 citationsRead paper