Author name cluster

Hao Cheng

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.

36 papers

2 author rows

AAAI Conference 2026 Conference Paper

4DSTR: Advancing Generative 4D Gaussians with Spatial-Temporal Rectification for High-Quality and Consistent 4D Generation

Mengmeng Liu
Jiuming Liu
Yunpeng Zhang
Jiangtao Li
Michael Ying Yang
Francesco Nex
Hao Cheng

Remarkable advances in recent 2D image and 3D shape generation have induced a significant focus on dynamic 4D content generation. However, previous 4D generation methods commonly struggle to maintain spatial-temporal consistency and adapt poorly to rapid temporal variations, due to the lack of effective spatial-temporal modeling. To address these problems, we propose a novel 4D generation network called 4DSTR, which modulates generative 4D Gaussian Splatting with spatial-temporal rectification. Specifically, temporal correlation across generated 4D sequences is designed to rectify deformable scales and rotations and guarantee temporal consistency. Furthermore, an adaptive spatial densification and pruning strategy is proposed to address significant temporal variations by dynamically adding or deleting Gaussian points with the awareness of their pre-frame movements. Extensive experiments demonstrate that our 4DSTR achieves state-of-the-art performance in video-to-4D generation, excelling in reconstruction quality, spatial-temporal consistency, and adaptation to rapid temporal movements.

PDF Details DOI

AAAI Conference 2026 Conference Paper

Modeling Uncertainty Trends for Timely Retrieval in Dynamic RAG

Bo Li
Tian Tian
Zhenghua Xu
Hao Cheng
Shikun Zhang
Wei Ye

Dynamic retrieval-augmented generation (RAG) allows large language models (LLMs) to fetch external knowledge on demand, offering greater adaptability than static RAG. A central challenge in this setting lies in determining the optimal timing for retrieval. Existing methods often trigger retrieval based on low token-level confidence, which may lead to delayed intervention after errors have already propagated. We introduce Entropy-Trend Constraint (ETC), a training-free method that determines optimal retrieval timing by modeling the dynamics of token-level uncertainty. Specifically, ETC utilizes first- and second-order differences of the entropy sequence to detect emerging uncertainty trends, enabling earlier and more precise retrieval. Experiments on six QA benchmarks with three LLM backbones demonstrate that ETC consistently outperforms strong baselines while reducing retrieval frequency. ETC is particularly effective in domain-specific scenarios, exhibiting robust generalization capabilities. Ablation studies and qualitative analyses further confirm that trend-aware uncertainty modeling yields more effective retrieval timing. The method is plug-and-play, model-agnostic, and readily integrable into existing decoding pipelines. Implementation code is included in the supplementary materials.

PDF Details DOI

AAAI Conference 2025 Conference Paper

Adaptive Decision Boundary for Few-Shot Class-Incremental Learning

Linhao Li
Yongzhang Tan
Siyuan Yang
Hao Cheng
Yongfeng Dong
Liang Yang

Few-Shot Class-Incremental Learning (FSCIL) aims to continuously learn new classes from a limited set of training samples without forgetting knowledge of previously learned classes. Conventional FSCIL methods typically build a robust feature extractor during the base training session with abundant training samples and subsequently freeze this extractor, only fine-tuning the classifier in subsequent incremental phases. However, current strategies primarily focus on preventing catastrophic forgetting, considering only the relationship between novel and base classes, without paying attention to the specific decision spaces of each class. To address this challenge, we propose a plug-and-play Adaptive Decision Boundary Strategy (ADBS), which is compatible with most FSCIL methods. Specifically, we assign a specific decision boundary to each class and adaptively adjust these boundaries during training to optimally refine the decision spaces for the classes in each session. Furthermore, to amplify the distinctiveness between classes, we employ a novel inter-class constraint loss that optimizes the decision boundaries and prototypes for each class. Extensive experiments on three benchmarks, namely CIFAR100, miniImageNet, and CUB200, demonstrate that incorporating our ADBS method with existing FSCIL techniques significantly improves performance, achieving overall state-of-the-art results.

PDF Details DOI

TIST Journal 2025 Journal Article

Aspect-Enhanced Explainable Recommendation with Multi-modal Contrastive Learning

Hao Liao
Shuo Wang
Hao Cheng
Wei Zhang
Jiwei Zhang
Mingyang Zhou
Kezhong Lu
Rui Mao

Explainable recommender systems ( ERS ) aim to enhance users’ trust in the systems by offering personalized recommendations with transparent explanations. This transparency provides users with a clear understanding of the rationale behind the recommendations, fostering a sense of confidence and reliability in the system’s outputs. Generally, the explanations are presented in a familiar and intuitive way, which is in the form of natural language, thus enhancing their accessibility to users. Recently, there has been an increasing focus on leveraging reviews as a valuable source of rich information in both modeling user-item preferences and generating textual interpretations, which can be performed simultaneously in a multi-task framework. Despite the progress made in these review-based recommendation systems, the integration of implicit feedback derived from user-item interactions and user-written text reviews has yet to be fully explored. To fill this gap, we propose a model named SERMON (A s pect-enhanced E xplainable R ecommendation with M ulti-modal C o ntrast Lear n ing). Our model explores the application of multimodal contrastive learning to facilitate reciprocal learning across two modalities, thereby enhancing the modeling of user preferences. Moreover, our model incorporates the aspect information extracted from the review, which provides two significant enhancements to our tasks. Firstly, the quality of the generated explanations is improved by incorporating the aspect characteristics into the explanations generated by a pre-trained model with controlled textual generation ability. Secondly, the commonly used user-item interactions are transformed into user-item-aspect interactions, which we refer to as interaction triple, resulting in a more nuanced representation of user preference. To validate the effectiveness of our model, we conduct extensive experiments on three real-world datasets. The experimental results show that our model outperforms state-of-the-art baselines, with a 2.0% improvement in prediction accuracy and a substantial 24.5% enhancement in explanation quality for the TripAdvisor dataset.

Details DOI

NeurIPS Conference 2025 Conference Paper

Decoder-Hybrid-Decoder Architecture for Efficient Reasoning with Long Generation

Liliang Ren
Congcong Chen
Haoran Xu
Young Jin Kim
Adam Atkinson
Zheng Zhan
Jiankai Sun
Baolin Peng

Recent advances in language modeling have demonstrated the effectiveness of State Space Models (SSMs) for efficient sequence modeling. While hybrid architectures such as Samba and the decoder-decoder architecture, YOCO, have shown promising performance gains over Transformers, prior works have not investigated the efficiency potential of representation sharing between SSM layers. In this paper, we introduce the Gated Memory Unit (GMU), a simple yet effective mechanism for efficient memory sharing across layers. We apply it to create SambaY, a decoder-hybrid-decoder architecture that incorporates GMUs in the cross-decoder to share memory readout states from a Samba-based self-decoder. SambaY significantly enhances decoding efficiency, preserves linear pre-filling time complexity, and boosts long-context performance, all while eliminating the need for explicit positional encoding. Through extensive scaling experiments, we demonstrate that our model exhibits a significantly lower irreducible loss compared to a strong YOCO baseline, indicating superior performance scalability under large-scale compute regimes. Our largest model enhanced with Differential Attention, Phi4-mini-Flash-Reasoning, achieves significantly better performance than Phi4-mini-Reasoning on reasoning tasks such as Math500, AIME24/25, and GPQA Diamond without any reinforcement learning, while delivering up to 10× higher decoding throughput on 2K-length prompts with 32K generation length under the vLLM inference framework. We release our training codebase on open-source data at https: //github. com/microsoft/ArchScale.