Haonan Cheng
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Haonan Cheng has a total of 77 co-authors. Here are the top co-authors by number of shared publications:
- Long Ye
Communication University of China, Beijing, China
19 shared publications - Hengyan Huang
Communication University of China, Beijing, China
4 shared publications - Qingbao Huang
School of Electrical Engineering, Guangxi University, Nanning, Guangxi, China
4 shared publications - Juanjuan Cai
Communication University of China
3 shared publications - Qin Zhang
Communication University of China
3 shared publications - Shulin Liu
School of information and communication engineering, Communication University of China, Beijing, China
3 shared publications
Latest Publications
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 15021-15030
MP-ADD: Multi-Prototype Authenticity Modeling with Decoupled Learning for Generalizable Audio Deepfake Detection
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 15021-15030
Open AccessMP-ADD: Multi-Prototype Authenticity Modeling with Decoupled Learning for Generalizable Audio Deepfake Detection
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 17730-17734
AT-ADD: All-Type Audio Deepfake Detection Challenge Summary
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 17730-17734
Open AccessAT-ADD: All-Type Audio Deepfake Detection Challenge Summary
- 8 November 2026
- Published by Springer Nature
- p. 559-572
CAGE: Component-Aware Gated Evidence Fusion for Environmental Speech Deepfake Detection
- 8 November 2026
- Published by Springer Nature
- p. 559-572
CAGE: Component-Aware Gated Evidence Fusion for Environmental Speech Deepfake Detection
- Open Access
- 24 June 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Multimedia Computing, Communications, and Applications
- Vol. 22 (7) , 1-17
VGL-DPO: Vision-Guided Lexical Direct Preference Optimization for Mitigating Hallucination in Multimodal Large Language Models
Multimodal large language models (MLLMs) have achieved significant advancements in multimodal understanding, reasoning, and interaction. However, they still suffer from hallucination, where the generated text often deviates from the factual content of the input image. To mitigate this issue, prior studies have primarily employed direct preference optimization (DPO) for human preference alignment. However, these approaches treat all textual words equally, neglecting the varying significance of individual words in grounding text generation to image content. This limitation hinders fine-grained semantic alignment and consequently constrains their effectiveness in hallucination suppression. To address this limitation, we propose a vision-guided lexical DPO method, called VGL-DPO. Specifically, we quantify the significance of words in positive preference data based on their relevance to the visual input and dynamically assign different weights to different words during training. This facilitates more precise optimization by emphasizing critical words that contribute to factual grounding. Additionally, we leverage the importance differences between high-significance words in positive and negative preference data to adaptively adjust the weight of the negative preference loss. This dynamic reweighting mechanism further refines the model’s ability to suppress hallucinated content while reinforcing factual accuracy. Extensive experiments across various models demonstrate that our method outperforms existing state-of-the-art methods in reducing hallucination and enhancing factual accuracy.
- 24 June 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Multimedia Computing, Communications, and Applications
- Vol. 22 (7) , 1-17
Open AccessVGL-DPO: Vision-Guided Lexical Direct Preference Optimization for Mitigating Hallucination in Multimodal Large Language Models
Multimodal large language models (MLLMs) have achieved significant advancements in multimodal understanding, reasoning, and interaction. However, they still suffer from hallucination, where the generated text often deviates from the factual content of the input image. To mitigate this issue, prior studies have primarily employed direct preference optimization (DPO) for human preference alignment. However, these approaches treat all textual words equally, neglecting the varying significance of individual words in grounding text generation to image content. This limitation hinders fine-grained semantic alignment and consequently constrains their effectiveness in hallucination suppression. To address this limitation, we propose a vision-guided lexical DPO method, called VGL-DPO. Specifically, we quantify the significance of words in positive preference data based on their relevance to the visual input and dynamically assign different weights to different words during training. This facilitates more precise optimization by emphasizing critical words that contribute to factual grounding. Additionally, we leverage the importance differences between high-significance words in positive and negative preference data to adaptively adjust the weight of the negative preference loss. This dynamic reweighting mechanism further refines the model’s ability to suppress hallucinated content while reinforcing factual accuracy. Extensive experiments across various models demonstrate that our method outperforms existing state-of-the-art methods in reducing hallucination and enhancing factual accuracy.
- 12 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. PP (15209210) , 1-15
RD-VTA: Rule-Data Guided Video-to-Audio Generation for Fine-Grained Footstep Sound
It is challenging to implement visually guided fine-grained footstep sounds based on a limited number of samples in complex scenes. This is due to the interference of redundant information in the complex background of the visual scene for audiovisual mapping. As well as the complex coupling of sound features makes audiovisual fine-grained linear mapping difficult. To address the mentioned problems, we propose an automated video-to-audio generation method (RD-VTA) for footstep sound that incorporates data-driven and rule-based modelling approaches. First, we design a data-driven masked footstep sound generation network (DM-AFSG) to acquire audiovisual temporal concordance. The network is capable of separating visual sound objects, reducing background redundant interference, and generating initial target sounds that capture temporal cues. Secondly, a rule-based fine-grained footstep sound adjustment method (RT-AFSG) is designed based on visual guides such as material, motion type and displacement distance. The proposed RT-AFSG effectively achieves diverse sounds with a limited number of sound samples through sound texture analysis and modification. Moreover, it constructs the mapping relationship between different visual cues and footstep sounds, and realizes the fine variation of footstep sounds. To adequately validate the effectiveness of the method in terms of audiovisual temporal consistency and content granularity, we perform objective synchronization metrics and subjective human evaluation on the footsteps audiovisual dataset VAFoot. The experimental results show that the method obtains an average of 5% improvement in sound synchronization performance and significantly outperforms several existing methods in terms of sound content granularity. We encourage readers to watch and listen to the footstep sound results on our demo website: https://quinntt.github.io/RD-VTA/.
- 12 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. PP (15209210) , 1-15
RD-VTA: Rule-Data Guided Video-to-Audio Generation for Fine-Grained Footstep Sound
It is challenging to implement visually guided fine-grained footstep sounds based on a limited number of samples in complex scenes. This is due to the interference of redundant information in the complex background of the visual scene for audiovisual mapping. As well as the complex coupling of sound features makes audiovisual fine-grained linear mapping difficult. To address the mentioned problems, we propose an automated video-to-audio generation method (RD-VTA) for footstep sound that incorporates data-driven and rule-based modelling approaches. First, we design a data-driven masked footstep sound generation network (DM-AFSG) to acquire audiovisual temporal concordance. The network is capable of separating visual sound objects, reducing background redundant interference, and generating initial target sounds that capture temporal cues. Secondly, a rule-based fine-grained footstep sound adjustment method (RT-AFSG) is designed based on visual guides such as material, motion type and displacement distance. The proposed RT-AFSG effectively achieves diverse sounds with a limited number of sound samples through sound texture analysis and modification. Moreover, it constructs the mapping relationship between different visual cues and footstep sounds, and realizes the fine variation of footstep sounds. To adequately validate the effectiveness of the method in terms of audiovisual temporal consistency and content granularity, we perform objective synchronization metrics and subjective human evaluation on the footsteps audiovisual dataset VAFoot. The experimental results show that the method obtains an average of 5% improvement in sound synchronization performance and significantly outperforms several existing methods in terms of sound content granularity. We encourage readers to watch and listen to the footstep sound results on our demo website: https://quinntt.github.io/RD-VTA/.
- 23 May 2026
- Published by Springer Nature
- p. 52-64
MC-AVD: Generalizable Audio-Visual Deepfake Detection via Multi-modal Collaboration
- 23 May 2026
- Published by Springer Nature
- p. 52-64
MC-AVD: Generalizable Audio-Visual Deepfake Detection via Multi-modal Collaboration
- 31 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. PP (2162237X) , 1-15
TS-Binaural: Visually Guided Binaural Audio Generation by Temporal–Spatial Dynamic Analysis
Visually guided binaural audio generation (VGBAG) aims at estimating spatial information from visual images to reconstruct binaural audio, yet struggles to model dynamic audio-visual relationships in complex multisource scenarios. This work focuses on musical instrument performance videos, a critical case characterized by a variable number of sound sources, multiple performance forms, and diverse stage effects. These dynamic characteristics are ignored by existing methods, resulting in inaccuracy of the generated binaural audio. To overcome these limitations, in this article, we propose a novel VGBAG method (TS-Binaural) based on temporal–spatial dynamic analysis (TSDA). Different from the overall audio-visual content analysis in previous work, the TS-Binaural employs a novel “separation-later-mixing” strategy. It splits complex audio-visual scenes into separate units to improve accuracy. The TS-Binaural comprises two key components: a TSDA module and a temporal–spatial information fusion (TSIF) guided binaural audio generation module. Furthermore, in the TSDA module, we design an automatic mask generation-based performance analyzer (AMGPA) to prevent intersource interference. The AMGPA integrates novel position detection logic to identify and separate the activation region, which refers to a separate visual region consisting of a player and his/her musical instrument being played at different times in the video. Extensive quantitative and qualitative comparisons demonstrate that our TS-Binaural outperforms the state-of-the-art VGBAG methods.
- 31 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. PP (2162237X) , 1-15
TS-Binaural: Visually Guided Binaural Audio Generation by Temporal–Spatial Dynamic Analysis
Visually guided binaural audio generation (VGBAG) aims at estimating spatial information from visual images to reconstruct binaural audio, yet struggles to model dynamic audio-visual relationships in complex multisource scenarios. This work focuses on musical instrument performance videos, a critical case characterized by a variable number of sound sources, multiple performance forms, and diverse stage effects. These dynamic characteristics are ignored by existing methods, resulting in inaccuracy of the generated binaural audio. To overcome these limitations, in this article, we propose a novel VGBAG method (TS-Binaural) based on temporal–spatial dynamic analysis (TSDA). Different from the overall audio-visual content analysis in previous work, the TS-Binaural employs a novel “separation-later-mixing” strategy. It splits complex audio-visual scenes into separate units to improve accuracy. The TS-Binaural comprises two key components: a TSDA module and a temporal–spatial information fusion (TSIF) guided binaural audio generation module. Furthermore, in the TSDA module, we design an automatic mask generation-based performance analyzer (AMGPA) to prevent intersource interference. The AMGPA integrates novel position detection logic to identify and separate the activation region, which refers to a separate visual region consisting of a player and his/her musical instrument being played at different times in the video. Extensive quantitative and qualitative comparisons demonstrate that our TS-Binaural outperforms the state-of-the-art VGBAG methods.
- Open Access
- 10 February 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Multimedia Computing, Communications, and Applications
- Vol. 22 (2) , 1-24
Implement Referring Expression Comprehension by Extending Auto-focus Lens to Locked Vision Model
Referring Expression Comprehension (REC) aims to achieve fine-grained cross-modal content alignment. The traditional two-stage approaches, by decomposing REC into localization (region proposal) and comprehension (expression-based ranking), lead to the isolation of continuous image information and heavily rely on the quality of the proposals. In this article, we propose a point-based two-stage framework for REC to quickly achieve localization by inserting a language-modulated auto-focus module into the locked vision model. Specifically, we redefine REC as two processes: point-based cross-modal comprehension and point-based instance localization. For the comprehension stage, we reconstruct the raw annotations into soft masks at the feature point level as a metric of cross-modal correlation. With this indirect metric, REC can be approximated as a binary classification problem, which fundamentally avoids the impact of isolated regions. Remarkably, soft masks are shape-independent, which means our method is extremely general. By switching different vision models, different types of predictions (e.g., localization and segmentation) can be obtained. Experiments on multiple benchmarks demonstrate the feasibility and potential of our point-based paradigm. Our code will be public at https://github.com/VILAN-Lab/PBREC-AF .
- 10 February 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Multimedia Computing, Communications, and Applications
- Vol. 22 (2) , 1-24
Open AccessImplement Referring Expression Comprehension by Extending Auto-focus Lens to Locked Vision Model
Referring Expression Comprehension (REC) aims to achieve fine-grained cross-modal content alignment. The traditional two-stage approaches, by decomposing REC into localization (region proposal) and comprehension (expression-based ranking), lead to the isolation of continuous image information and heavily rely on the quality of the proposals. In this article, we propose a point-based two-stage framework for REC to quickly achieve localization by inserting a language-modulated auto-focus module into the locked vision model. Specifically, we redefine REC as two processes: point-based cross-modal comprehension and point-based instance localization. For the comprehension stage, we reconstruct the raw annotations into soft masks at the feature point level as a metric of cross-modal correlation. With this indirect metric, REC can be approximated as a binary classification problem, which fundamentally avoids the impact of isolated regions. Remarkably, soft masks are shape-independent, which means our method is extremely general. By switching different vision models, different types of predictions (e.g., localization and segmentation) can be obtained. Experiments on multiple benchmarks demonstrate the feasibility and potential of our point-based paradigm. Our code will be public at https://github.com/VILAN-Lab/PBREC-AF .
- 7 February 2026
- Published by Elsevier in Information Fusion
- Vol. 132
Lightweight music recommendation via multi-physiological feature fusion
- 7 February 2026
- Published by Elsevier in Information Fusion
- Vol. 132
Lightweight music recommendation via multi-physiological feature fusion
- 28 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 21 (15566013) , 2047-2060
Anchor-Based Multimodal Verification: A Dynamic Query Framework for Fake News Forensics in Short Videos
The proliferation of maliciously altered short videos on social media platforms poses a significant threat to information security ecosystems, eroding public trust in digital media. Despite recent advancements in detecting fake video news, significant challenges remain in the forensic analysis of short videos, leading to issues of bias. First, as technology rapidly advances, fake videos are becoming increasingly semantically convincing, undermining the effectiveness of current classification methods. Second, the heterogeneous nature of video modalities (visual, textual, audio) creates critical challenges for models to learn discriminative feature representations. To address these challenges, we propose a dynamic query framework for fake news forensics in short videos, termed the Semantic Guided Adaptive Network (SGAN). Our approach is motivated by the need to utilize superficial alignment to identify suspicious manipulations through anchor-based verification and to leverage the adaptive capability of learnable queries to learn the heterogeneous boundary in each modality. Specifically, SGAN comprises a verification module and a flexible query learning module. The verification module employs text as the anchor to verify detailed context, mining fine-grained information while emphasizing key features, thereby providing candidate manipulations for downstream modules. The query learning module leverages learnable queries to map heterogeneous forensic features and integrates them through multi-level fusion for decision-making. Extensive experiments conducted on two widely used datasets demonstrate the effectiveness and generalization of the proposed method.
- 28 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 21 (15566013) , 2047-2060
Anchor-Based Multimodal Verification: A Dynamic Query Framework for Fake News Forensics in Short Videos
The proliferation of maliciously altered short videos on social media platforms poses a significant threat to information security ecosystems, eroding public trust in digital media. Despite recent advancements in detecting fake video news, significant challenges remain in the forensic analysis of short videos, leading to issues of bias. First, as technology rapidly advances, fake videos are becoming increasingly semantically convincing, undermining the effectiveness of current classification methods. Second, the heterogeneous nature of video modalities (visual, textual, audio) creates critical challenges for models to learn discriminative feature representations. To address these challenges, we propose a dynamic query framework for fake news forensics in short videos, termed the Semantic Guided Adaptive Network (SGAN). Our approach is motivated by the need to utilize superficial alignment to identify suspicious manipulations through anchor-based verification and to leverage the adaptive capability of learnable queries to learn the heterogeneous boundary in each modality. Specifically, SGAN comprises a verification module and a flexible query learning module. The verification module employs text as the anchor to verify detailed context, mining fine-grained information while emphasizing key features, thereby providing candidate manipulations for downstream modules. The query learning module leverages learnable queries to map heterogeneous forensic features and integrates them through multi-level fusion for decision-making. Extensive experiments conducted on two widely used datasets demonstrate the effectiveness and generalization of the proposed method.