Author name cluster

Hong Sun

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.

11 papers

2 author rows

JBHI Journal 2026 Journal Article

Dual-Branch Self-Supervised Contrastive Pre-Training Framework for Sleep Stage Classification

Jie Ouyang
Yuanwang Wei
Shuxia Qian
Fried-Michael Dahlweid
Hong Sun
Zou Lai
Xianchao Zhang

Accurate sleep staging is vital for evaluating sleep quality and diagnosing sleep disorders. Yet most automated sleep staging methods rely on large datasets labeled by experts. However, clinical annotation is both time-consuming and subjective, making it difficult to obtain sufficient high-quality data for automated sleep staging research. To address this bottleneck, we propose a few-shot, dual-branch contrastive pre-training framework for single-channel electroencephalogram (EEG)–based sleep staging. The framework first conducts fully self-supervised pre-training on unlabeled data, then performs fine-tuning that requires only a small set of labeled samples. We developed and evaluated our solution with the public Sleep-EDF-v2 EEG dataset, achieving state-of-the-art results despite using limited labeled data. Specifically, with only 1% labeled data, our method delivers an accuracy of 76. 10% and Macro F1-score of 61. 34%, comparable to supervised models trained on 100% labeled data. We further validated our approach on the ISRUC-1 and ISRUC-3 datasets, where similar robust results were consistently observed. The ability to effectively develop sleep classification models using minimal labeled data demonstrates the potential value of our framework across diverse clinical settings.

Details DOI

AAAI Conference 2026 Conference Paper

Efficient, Secure, Differentially Private Deep Learning in the Two-Server Model

Jun Feng
Hong Sun
Pengfei Zhang
Bocheng Ren
Shunli Zhang

Existing solutions on differentially private deep learning (DPDL) either require the assumption of a trusted data server (centralized DPDL) or suffer from poor utility (local DPDL); and hence their adoptions are hampered in real-world scenarios.We present CRYPTDP, a crypto-assisted differentially private deep learning approach in the two-server model. CRYPTDP employs two non-colluding servers to collaboratively and efficiently train differentially private deep learning over the secret shares of data owners' private data while protecting the confidentiality of the data from untrusted servers. CRYPTDP is the first approach with the best of both local DPDL and centralized DPDL models, which does not resort to trusted server like local DPDL and has the utility like centralized DPDL. In particular, we also make innovations for addressing the major challenges like poor performance and security that beset CRYPTDP: We introduce a new secure computation and differential privacy friendly activation function; we propose a novel garbled-circuits-free most significant bit extraction protocol, and using the protocol we propose an efficient and secure garbled-circuits-free protocol for activation function over secret shares. Exhaustive experiments show that CRYPTDP delivers significantly better performance than the state-of-the-art local DPDL, yields higher accuracy than the state-of-the-art centralized DPDL, and can achieve two orders of magnitude faster runtime than the state-of-the-art approach.

PDF Details DOI

AAAI Conference 2026 Conference Paper

Preference Is More than Comparisons: Rethinking Dueling Bandits with Augmented Human Feedback

Shengbo Wang
Hong Sun
Ke Li

Interactive preference elicitation (IPE) aims to substantially reduce human effort while acquiring human preferences in wide personalization systems. Dueling bandit (DB) algorithms enable optimal decision-making in IPE building on pairwise comparisons. However, they remain inefficient when human feedback is sparse. Existing methods address sparsity by heavily relying on parametric reward models, whose rigid assumptions are vulnerable to misspecification. In contrast, we explore an alternative perspective based on feedback augmentation, and introduce critical improvements to the model-free DB framework. Specifically, we introduce augmented confidence bounds to integrate augmented human feedback under generalized concentration properties, and analyze the multi-factored performance trade-off via regret analysis. Our prototype algorithm achieves competitive performance across several IPE benchmarks, including recommendation, multi-objective optimization, and response optimization for large language models, demonstrating the potential of our approach for provably efficient IPE in broader applications.

PDF Details DOI

AAAI Conference 2026 Conference Paper

Stabilizing Cross-Modal Bidirectional Attribution: Few-Shot Adversarial Prompt Tuning for Robust Vision-Language Models

Jun Feng
Shuhong Wu
Hong Sun
Pengfei Zhang
Bocheng Ren
Shunli Zhang

Large-scale pre-trained vision-language models (VLMs) like CLIP show exceptional performance and zero-shot generalization. However, their reliability may be severely undermined by a critical vulnerability to subtle adversarial perturbations. Our work reveals a critical cross-modal vulnerability: visual-only perturbations induce substantial, synchronous shifts in decision attribution maps across both image and text. This phenomenon signifies a fundamental disruption of the VLM's internal logic, as it alters both the model's perceptual focus and its decision rationale. To counter this vulnerability, we introduce Cross-modal Bidirectional Attribution guided Few-shot Adversarial Prompt Tuning (CBA-FAPT), a novel method that leverages the model's internal decision rationale as a regularizer for robust learning. Our framework's core mechanism is the alignment of a novel bidirectional attribution map. This map is a unique fusion of two components. It combines forward feature attention to capture the model's perceptual focus. It also incorporates backward decision gradients to act as a proxy for the model's decision rationale, quantifying how each feature influences the final outcome. We enforce consistency on this bidirectional map between clean and adversarial examples. This approach corrects the model's internal logic on two fronts and effectively restores its adversarial robustness. Comprehensive experiments on 11 datasets demonstrate that CBA-FAPT outperforms the state-of-the-art, establishing a superior trade-off between robust and natural accuracy.

PDF Details DOI

EAAI Journal 2025 Journal Article

A Dual Two-Stage Attention-based Model for interpretable hard landing prediction from flight data

Jiaxing Shang
Xiaoquan Li
Ruixiang Zhang
Linjiang Zheng
Xu Li
Riquan Zhang
Xinbin Zhao
Fan Li

Hard landings are a significant safety concern in aviation, with potential consequences ranging from poor passenger experiences to serious injuries or fatalities. Predicting and explaining hard landing events are equally important for enhancing flight safety, the former makes it possible to give proactive warnings, while the latter helps pilots identify the reasons and refine their skills. However, existing studies generally lack a comprehensive consideration for the fine-grained characteristics of flight data containing both inter-temporal and inter-parametric relationships, resulting in suboptimal prediction performance. In addition, most of existing approaches aim at improving the prediction performance but fail to provide interpretability for the causes of hard landing. To address the above problems, we propose DUTSAM, a DUal Two-Stage Attention-based interpretable Model for hard landing prediction from quick access recorder (QAR) data. The model consists of dual parallel modules, each of which combines a convolutional feature encoder and a two-stage attention mechanism. The two encoders capture fine-grained characteristics by encoding multivariate data from temporal domain and parametric domain respectively. After that, the dual two-stage attention mechanism captures the inter-temporal and inter-parametric correlations in reverse order to predict hard landing and provide interpretation from both temporal and parametric perspectives. Experimental results on a real QAR dataset with 37, 920 flights show that DUTSAM achieves better prediction performance compared with other state-of-the-art baselines in terms of Precision, Recall, and F1-score. Additionally, case study demonstrates that DUTSAM can uncover key flight parameters and moments strongly correlated to the hard landing events.

Details DOI

AAAI Conference 2025 Conference Paper

SADBA: Self-Adaptive Distributed Backdoor Attack Against Federated Learning

Jun Feng
Yuzhe Lai
Hong Sun
Bocheng Ren

Backdoor attacks in federated learning (FL) face challenges such as lower attack success rates and compromised main task accuracy (MA) compared to local training. Existing methods like distributed backdoor attack (DBA) mitigate these issues by modifying malicious clients’ updates and partitioning global triggers to enhance backdoor persistence and stealth. The recent full combination backdoor attack (FCBA) further improves backdoor efficiency with a full combination strategy. However, these methods are mainly applicable in small-scale FL. In large-scale FL, small trigger patterns weaken impact, and scaling them requires controlling exponentially more clients, which poses significant challenges, while simply reverting to DBA may decrease backdoor performance. To overcome these challenges, we propose the self-adaptive distributed backdoor attack (SADBA), which achieves similar performance to FCBA with a lower percentage of malicious clients (PMC). It also adapts more flexibly through an optimized model poisoning strategy and a self-adaptive data poisoning strategy. Experiments demonstrate SADBA outperforms state-of-the-art methods, achieving higher or comparable backdoor performance and MA across various datasets with limited PMC.

PDF Details DOI

ICLR Conference 2024 Conference Paper

Efficient Backdoor Attacks for Deep Neural Networks in Real-world Scenarios

Ziqiang Li 0001
Hong Sun
Pengfei Xia
Heng Li 0008
Beihao Xia
Yi Wu 0018
Bin Li 0025

Recent deep neural networks (DNNs) have came to rely on vast amounts of training data, providing an opportunity for malicious attackers to exploit and contaminate the data to carry out backdoor attacks. However, existing backdoor attack methods make unrealistic assumptions, assuming that all training data comes from a single source and that attackers have full access to the training data. In this paper, we introduce a more realistic attack scenario where victims collect data from multiple sources, and attackers cannot access the complete training data. We refer to this scenario as $\textbf{data-constrained backdoor attacks}$. In such cases, previous attack methods suffer from severe efficiency degradation due to the $\textbf{entanglement}$ between benign and poisoning features during the backdoor injection process. To tackle this problem, we introduce three CLIP-based technologies from two distinct streams: $\textit{Clean Feature Suppression}$ and $\textit{Poisoning Feature Augmentation}$. The results demonstrate remarkable improvements, with some settings achieving over $\textbf{100}$% improvement compared to existing attacks in data-constrained scenarios.

Details

IJCAI Conference 2015 Conference Paper

A Hybrid Neural Model for Type Classification of Entity Mentions

Li Dong
Furu Wei
Hong Sun
Ming Zhou
Ke Xu

The semantic class (i. e. , type) of an entity plays a vital role in many natural language processing tasks, such as question answering. However, most of existing type classification systems extensively rely on hand-crafted features. This paper introduces a hybrid neural model which classifies entity mentions to a wide-coverage set of 22 types derived from DBpedia. It consists of two parts. The mention model uses recurrent neural networks to recursively obtain the vector representation of an entity mention from the words it contains. The context model, on the other hand, employs multilayer perceptrons to obtain the hidden representation for contextual information of a mention. Representations obtained by the two parts are used together to predict the type distribution. Using automatically generated data, these two parts are jointly learned. Experimental studies illustrate that the proposed approach outperforms baseline methods. Moreover, when type information provided by our method is used in a question answering system, we observe a 14. 7% relative improvement for the top-1 accuracy of answers.

PDF Details

IJCAI Conference 2013 Conference Paper

Answer Extraction from Passage Graph for Question Answering

Hong Sun
Nan Duan
Yajuan Duan
Ming Zhou

In question answering, answer extraction aims to pin-point the exact answer from passages. However, most previous methods perform such extraction on each passage separately, without considering clues provided in other passages. This paper presents a novel approach to extract answers by fully leveraging connections among different passages. Specially, extraction is performed on a Passage Graph which is built by adding links upon multiple passages. Different passages are connected by linking words with the same stem. We use the factor graph as our model for answer extraction. Experimental results on multiple QA data sets demonstrate that our method significantly improves the performance of answer extraction.

PDF Details DOI

YNIMG Journal 2006 Journal Article

Discriminative analysis of relapsing neuromyelitis optica and relapsing–remitting multiple sclerosis based on two-dimensional histogram from diffusion tensor imaging

Fuchun Lin
Chunshui Yu
Tianzi Jiang
Kuncheng Li
Chaozhe Zhu
Wanlin Zhu
Wen Qin
Yunyun Duan

It is difficult to completely differentiate patients with relapsing neuromyelitis optica (RNMO) from relapsing–remitting multiple sclerosis (RRMS) for their similarities in clinical manifestation. In this study, we proposed a novel approach, using two-dimensional histogram of apparent diffusion coefficient (ADC) and fractional anisotropy (FA) of the brain derived from diffusion tensor imaging (DTI) as classification feature, to discriminate patients with RNMO from RRMS. In this approach, two-dimensional principal component analysis (2D-PCA) was used to extract feature and reduce dimensionality of matrix-formed data efficiently. Then linear discriminant analysis (LDA) was performed on these extracted features to find the best projection direction to separate patients with RNMO from RRMS. Finally, a minimum distance classifier was generated on the basis of projection scores. The correct recognition rate of our method reached 85. 7%, validated by the leave-one-out method. This result was much higher than that using feature of ADC or FA separately (59. 5% for ADC, 76. 2% for FA). In conclusion, the proposed method on the basis of combined features is more effective for classification than those merely using the features separately, and it may be helpful in differentiating RNMO from RRMS patients.

Details DOI

YNIMG Journal 2006 Journal Article

Quantitative analysis along the pyramidal tract by length-normalized parameterization based on diffusion tensor tractography: Application to patients with relapsing neuromyelitis optica

Fuchun Lin
Chunshui Yu
Tianzi Jiang
Kuncheng Li
Xiaobo Li
Wen Qin
Hong Sun
Piu Chan

In this study, we introduced a length-normalized parameterization method to establish anatomical correspondence of white matter fiber tracts across subjects and applied this method to investigate the presence of abnormal diffusion along the pyramidal tract (PYT) of relapsing neuromyelitis optica (RNMO) patients without visible brain lesions. In this approach, the part of the PYT between the lowest slice of the cerebral peduncle and the uppermost slice of the lateral ventricle was reconstructed to establish the anatomical correspondence across subjects using diffusion tensor tractography. Then it was parameterized by normalizing its length and dividing equally the normalized length into a certain number of segments, so that the comparability of each segment across subjects along the PYT was established. Tract-specific diffusion indices, including directionally averaged diffusivity (D av), fractional anisotropy (FA), primary diffusivity (λ 1) and transverse diffusivity (λ 23), were obtained from each segment. Thus, the distribution maps of these indices along the PYT were obtained. The distribution maps of D av, FA, and λ 23 of RNMO patients were significantly different from those of healthy controls, especially in the lower part of the PYT. The differences may be caused by secondary degeneration to lesions in the spinal cord. In conclusion, a length-normalized parameterization method is proposed to establish anatomical correspondence for the PYT. Compared with existed methods, a major merit of our method is to provide comparability across subjects along the PYT on the basis of diffusion tensor tractography and to make it possible for the quantitative analysis along the fiber tract. This method can also be used to quantitatively analyze other white matter fiber tracts between two definite anatomic landmarks in many neurological or psychiatric diseases.

Details DOI