Author name cluster

Man Li

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

4 papers

1 author row

EAAI Journal 2026 Journal Article

Multi-view knowledge graph recommendation with dynamic transformer modeling

Mengting Zhang
Man Li
Wenming Cao
Mingming Yang
Bing Li
Guangzhen Zhu
Lu Cheng

Recommender systems are essential for alleviating information overload but continue to face challenges such as data sparsity, cold-start issues, and noisy knowledge graph structures. To address these problems, we propose a novel recommendation framework that integrates a Dynamic Tanh Transformer (DyT-Transformer), a multi-view knowledge graph propagation strategy, and an inter-layer attention mechanism. The DyT-Transformer enhances user-item representations by dynamically weighting key features without normalization layers, improving stability and efficiency. Multi-view propagation is employed to generate diverse graph views through randomized perturbations, which mitigates noise and improves robustness. Inter-layer attention adaptively aggregates embedding across propagation depths to capture hierarchical semantics. Extensive experiments on five real-world datasets, including Book-Crossing, MovieLens-1M, Amazon-Book, Yelp2018, and Alibaba, demonstrate that the proposed model consistently outperforms state-of-the-art baselines in both click-through rate prediction and Top-K recommendation tasks. These results confirm the effectiveness and scalability of combining DyT-Transformer with multi-view knowledge graph learning for robust recommendation. We have made the source code of our proposed model available at https: //github. com/tttterter/MVLDT.

Details DOI

JBHI Journal 2025 Journal Article

A Multimodal Consistency-Based Self-Supervised Contrastive Learning Framework for Automated Sleep Staging in Patients With Disorders of Consciousness

Jiahui Pan
Yangzuyi Yu
Man Li
Wanxin Wei
Shuyu Chen
Heyi Zheng
Yanbin He
Yuanqing Li

Sleep is a fundamental human activity, and automated sleep staging holds considerable investigational potential. Despite numerous deep learning methods proposed for sleep staging that exhibit notable performance, several challenges remain unresolved, including inadequate representation and generalization capabilities, limitations in multimodal feature extraction, the scarcity of labeled data, and the restricted practical application for patients with disorder of consciousness (DOC). This paper proposes MultiConsSleepNet, a multimodal consistency-based sleep staging network. This network comprises a unimodal feature extractor and a multimodal consistency feature extractor, aiming to explore universal representations of electroencephalograms (EEGs) and electrooculograms (EOGs) and extract the consistency of intra- and intermodal features. Additionally, self-supervised contrastive learning strategies are designed for unimodal and multimodal consistency learning to address the current situation in clinical practice where it is difficult to obtain high-quality labeled data but has a huge amount of unlabeled data. It can effectively alleviate the model's dependence on labeled data, and improve the model's generalizability for effective migration to DOC patients. Experimental results on three publicly available datasets demonstrate that MultiConsSleepNet achieves state-of-the-art performance in sleep staging with limited labeled data and effectively utilizes unlabeled data, enhancing its practical applicability. Furthermore, the proposed model yields promising results on a self-collected DOC dataset, offering a novel perspective for sleep staging research in patients with DOC.

Details DOI

YNIMG Journal 2025 Journal Article

Screening tools for subjective cognitive decline and mild cognitive impairment based on task-state prefrontal functional connectivity: a functional near-infrared spectroscopy study

Zhengping Pu
Hongna Huang
Man Li
Hongyan Li
Xiaoyan Shen
Lizhao Du
Qingfeng Wu
Xiaomei Fang

BACKGROUND: Subjective cognitive decline (SCD) and mild cognitive impairment (MCI) carry the risk of progression to dementia, and accurate screening methods for these conditions are urgently needed. Studies have suggested the potential ability of functional near-infrared spectroscopy (fNIRS) to identify MCI and SCD. The present fNIRS study aimed to develop an early screening method for SCD and MCI based on activated prefrontal functional connectivity (FC) during the performance of cognitive scales and subject-wise cross-validation via machine learning. METHODS: Activated prefrontal FC data measured by fNIRS were collected from 55 normal controls, 80 SCD patients, and 111 MCI patients. Differences in FC were analyzed among the groups, and FC strength and cognitive scale performance were extracted as features to build classification and predictive models through machine learning. Model performance was assessed based on accuracy, specificity, sensitivity, and area under the curve (AUC) with 95 % confidence interval (CI) values. RESULTS: Statistical analysis revealed a trend toward more impaired prefrontal FC with declining cognitive function. Prediction models were built by combining features of prefrontal FC and cognitive scale performance and applying machine learning models, The models showed generally satisfactory abilities to differentiate among the three groups, especially those employing linear discriminant analysis, logistic regression, and support vector machine. Accuracies of 92.0 % for MCI vs. NC, 80.0 % for MCI vs. SCD, and 76.1 % for SCD vs. NC were achieved, and the highest AUC values were 97.0 % (95 % CI: 94.6 %-99.3 %) for MCI vs. NC, 87.0 % (95 % CI: 81.5 %-92.5 %) for MCI vs. SCD, and 79.2 % (95 % CI: 71.0 %-87.3 %) for SCD vs. NC. CONCLUSION: The developed screening method based on fNIRS and machine learning has the potential to predict early-stage cognitive impairment based on prefrontal FC data collected during cognitive scale-induced activation.

Details DOI

AAAI Conference 2023 Conference Paper

Wiener Graph Deconvolutional Network Improves Graph Self-Supervised Learning

Jiashun Cheng
Man Li
Jia Li
Fugee Tsung

Graph self-supervised learning (SSL) has been vastly employed to learn representations from unlabeled graphs. Existing methods can be roughly divided into predictive learning and contrastive learning, where the latter one attracts more research attention with better empirical performance. We argue that, however, predictive models weaponed with powerful decoder could achieve comparable or even better representation power than contrastive models. In this work, we propose a Wiener Graph Deconvolutional Network (WGDN), an augmentation-adaptive decoder empowered by graph wiener filter to perform information reconstruction. Theoretical analysis proves the superior reconstruction ability of graph wiener filter. Extensive experimental results on various datasets demonstrate the effectiveness of our approach.

PDF Details DOI