Author name cluster

Tong Wu

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.

51 papers

2 author rows

EAAI Journal 2026 Journal Article

A double auto-weighted strategy for multi-view clustering

Tong Wu
Gui-Fu Lu

Multi-view clustering (MVC) excels by combining insights from various viewpoints, thereby improving both the precision and stability of clustering outcomes. However, existing MVC algorithms are inevitably affected by noise, leading to a decrease in clustering performance. Furthermore, the high-order correlations between multiple views and the underlying structural information are not effectively utilized. To solve these issues, we propose a double auto-weighted strategy for multi-view clustering (DAWS). Specifically, first, to mitigate the impact of outliers, we design an auto-weighted strategy on data reconstruction errors of each view, which can adaptively assign smaller weight to the feature with larger reconstruction error and assign larger weight to the important feature. Second, we stack these similarity graphs into a tensor and design a novel automatically weighted exponential tensor nuclear norm (AWETNN) to constrain it, which serves as a better alternative to tensor rank. AWETNN adequately considers the physical differences between singular values through a non-convex penalty function, thus more accurately utilizing the high-order correlations between multiple views and the intrinsic structural information. Ultimately, the two stages are merged into a cohesive framework using the augmented Lagrange multiplier method. Compared with ten different algorithms on six different datasets, DAWS consistently takes the lead. Moreover, DAWS converges very quickly, generally within 15–35 iterations.

Details DOI

AAAI Conference 2026 Conference Paper

DySy-Det: A Synergistic Framework with Dynamic Reconstruction-Path Consistency for AI-Generated Image Detection

Fanli Jin
Feng Lin
Gaojian Wang
Tong Wu
Zhisheng Yan

Advanced image generative models have led to concerns about malicious use, underscoring the necessity for generalizable detection methods. However, existing approaches tend to overfit to domain-specific forgery patterns, while overlooking complementary cues from different domains. Therefore, we introduce DySy-Det (Dynamic Synergy Detector), a novel framework that mines collaborative and robust forgery artifacts from multiple evidence domains. First, DySy-Det fine-tunes a CLIP vision transformer to extract high-level semantics for identifying conceptual inconsistencies, while generating attention maps that pinpoint key discriminative regions. Then, this semantic guidance, in the form of a mask, directs a targeted reconstruction process. By focusing on these salient areas, our approach effectively extracts localized reconstruction errors, thereby filtering out irrelevant background noise. Furthermore, inspired by the intrinsic generative mechanics of diffusion models, we introduce the concept of Reconstruction-Path Consistency (RPC), which quantifies the temporal stability of the denoising trajectory to expose dynamic generative artifacts. We capture this by computing noise alignment scores across multiple timesteps and encode them via a lightweight network. Extensive evaluations on GenImage and UniversalFakeDetect benchmarks demonstrate that DySy-Det outperforms the state-of-the-art detector by 6.14% and 1.57% in mean accuracy, respectively.

PDF Details DOI

AAAI Conference 2026 Conference Paper

Eguard: Defending LLM Embeddings Against Inversion Attacks via Text Mutual Information Optimization

Tiantian Liu
Hongwei Yao
Feng Lin
Tong Wu
Zhan Qin
Kui Ren

While text embeddings enable efficient semantic processing in LLMs, they remain vulnerable to inversion attacks that reconstruct sensitive original text. However, current defense methods typically treat text embeddings from the feature level independently, ignoring the exploitation of the mutual relation among the embedding construction pipeline. To address this limitation, we propose Eguard, a framework that effectively disrupts chains of relationships between the original semantic space and defended functional space. Our improvements manifest at two levels, i.e., the global-level and local-level mutual information. At the global level, we propose to minimize the statistical dependency between protected embeddings and their original inputs, effectively decoupling sensitive content from the semantic space accessible to adversaries. At the local level, we apply keyword-antonym contrastive learning to enforce semantic discriminability within the space of downstream utility. This synergy of global privacy control and local semantic alignment allows Eguard to achieve a superior privacy-utility trade-off than traditional defenses. Our approach significantly reduces privacy risks, protecting over 95 percent of tokens from inversion while maintaining high performance across downstream tasks consistent with original embeddings.

PDF Details DOI

JMLR Journal 2026 Journal Article

Guaranteed Nonconvex Low-Rank Tensor Estimation via Scaled Gradient Descent

Tong Wu

Tensors, which give a faithful and effective representation to deliver the intrinsic structure of multi-dimensional data, play a crucial role in an increasing number of signal processing and machine learning problems. However, tensor data are often accompanied by arbitrary signal corruptions, including missing entries and sparse noise. A fundamental challenge is to reliably extract the meaningful information from corrupted tensor data in a statistically and computationally efficient manner. This paper develops a scaled gradient descent (ScaledGD) algorithm to directly estimate the tensor factors with tailored spectral initializations under the tensor-tensor product (t-product) and tensor singular value decomposition (t-SVD) framework. With tailored variants for tensor robust principal component analysis, (robust) tensor completion and tensor regression, we theoretically show that ScaledGD achieves linear convergence at a constant rate that is independent of the condition number of the ground truth low-rank tensor, while maintaining the low per-iteration cost of gradient descent. To the best of our knowledge, ScaledGD is the first algorithm that provably has such properties for low-rank tensor estimation with the t-SVD. Finally, numerical examples are provided to demonstrate the efficacy of ScaledGD in accelerating the convergence rate of ill-conditioned low-rank tensor estimation in a number of applications. [abs] [ pdf ][ bib ] [ code ] &copy JMLR 2026. ( edit, beta )

PDF Details

EAAI Journal 2026 Journal Article

Improving Transformer-based image captioning via deep collaborative cross-fusion

Junbo Hu
Tong Wu
Zhixin Li

Image captioning is a pivotal interdisciplinary task at the intersection of computer vision and natural language processing, aiming to generate accurate and descriptive sentences for visual content. Research indicates that enhancing the correlation between multimodal features can enhance image captioning generation’s accuracy. However, existing methods based on the encoder–decoder framework are limited to fusing multimodal features either during the encoding or decoding stage, making it difficult to effectively bridge the semantic gap. Confronted with challenges such as processing complex image features, filtering redundant information, and fusing multimodal features, this paper proposes a Deep Synergistic Cross-fusion Transformer model (DSCT), which leverages a comprehensive multi-feature fusion method during the entire process to improve generation quality. Specifically, we have designed a Sparse Cross-fusion Encoder (SCFE) that pre-fuses features and leverages sparsity to effectively eliminate visual redundant information, thereby solving the problem of processing complex image features and filtering redundant information, generating refined dual visual features. Furthermore, we introduce a Dual-feature Regulation Fusion Attention (DRFA) module, which can precisely regulate the significance and input of visual and semantic representations during the process of creating captions, facilitating deep fusion of multimodal features to address the issue of inadequate multimodal feature fusion. Experimental results on the MS COCO dataset demonstrate that the DSCT model achieves performance competitive with state-of-the-art methods, attaining a CIDEr score of 141. 1% on the Karpathy test split, which validates the effectiveness of the innovative approach proposed in this study.

Details DOI

EAAI Journal 2025 Journal Article

A self-attention enhanced generative adversarial autoencoder for anomaly detection of self-discharge of lithium-ion batteries

Zhengyu Liu
Chuanqing Wu
Tong Wu
Chao Sun
Yining Liu

Self-discharge is a crucial parameter affecting the reliability and lifespan of lithium-ion batteries (LIBs). However, traditional methods for detecting self-discharge rely heavily on time-consuming experimental testing. To address this limitation, we propose a novel model, self-attention enhanced generative adversarial autoencoder (SAE-GAAE), for rapid and accurate LIB self-discharge anomaly detection. SAE-GAAE integrates recent advances in artificial intelligence into battery production scenarios by combining a dot-product self-attention mechanism within the encoder-which captures inter-feature dependencies and highlights key indicators-with a generative adversarial component in the latent space, which enhances generalization and robustness by regularizing feature representations. This end-to-end deep learning framework enables automatic extraction of informative representations from raw input data without relying on manual feature engineering. Moreover, a tree-structured Parzen estimator (TPE)-based Bayesian optimization algorithm is employed to efficiently fine-tune model hyperparameters, improving detection performance. Applied to the capacity grading stage in LIB production, the model uses 40 features – including voltage, current, capacity, and temperature – extracted from charge–discharge curves. Experimental evaluation on real-world production data demonstrates that SAE-GAAE achieves a 26. 27% improvement in the average area under the receiver operating characteristic curve (AUC-ROC) in four models based on the autoencoder, with a detection accuracy of 99. 05% and a recall rate of 100%. These results highlight the model’s practical value in enhancing battery screening efficiency while reducing reliance on long-duration standing tests.

Details DOI

NeurIPS Conference 2025 Conference Paper

Absolute Zero: Reinforced Self-play Reasoning with Zero Data

Andrew Zhao
Yiran Wu
Tong Wu
Quentin Xu
Yang Yue
Matthieu Lin
Shenzhi Wang
Qingyun Wu

Reinforcement learning with verifiable rewards (RLVR) has shown promise in enhancing the reasoning capabilities of large language models by learning directly from rule-based outcome rewards. Recent RLVR works that operate under the zero setting avoid supervision in labeling the reasoning process, but still depend on manually curated collections of questions and answers for training. The scarcity of high-quality, human-produced examples raises concerns about the long-term scalability of relying on human supervision, a challenge already evident in the domain of language model pretraining. Furthermore, in a hypothetical future where AI surpasses human intelligence, tasks provided by humans may offer limited learning potential for a superintelligent system. To address these concerns, we propose a new RLVR paradigm called Absolute Zero, in which a single model learns to propose tasks that maximize its own learning progress and improves reasoning by solving them, without relying on any external human or distillation data. Under this paradigm, we introduce the Absolute Zero Reasoner (AZR), a system that self-evolves its training curriculum and reasoning ability. AZR uses a code executor to both validate self-proposed code reasoning tasks and verify answers, serving as an unified source of verifiable feedback to guide open-ended yet grounded learning. Despite being trained entirely without external data, AZR achieves overall SOTA performance on coding and mathematical reasoning tasks, outperforming existing zero-setting models that rely on tens of thousands of in-domain human-curated examples. Furthermore, we demonstrate that AZR can be effectively applied across different model scales and is compatible with various model classes.

PDF Details

ICRA Conference 2025 Conference Paper

Chemistry3D: Robotic Interaction Toolkit for Chemistry Experiments

Shoujie Li
Yan Huang
Changqing Guo
Tong Wu
Jiawei Zhang 0012
Linrui Zhang
Wenbo Ding 0001

The advent of simulation engines has revolutionized learning and operational efficiency for robots, offering cost-effective and swift pipelines. However, the lack of a universal simulation platform tailored for chemical scenarios impedes progress in robotic manipulation and visualization of reaction processes. Addressing this void, we present Chemistry3D, an innovative toolkit that integrates extensive chemical and robotic knowledge. Chemistry3D not only enables robots to perform chemical experiments but also provides real-time visualization of temperature, color, and pH changes during reactions. Built on the NVIDIA Omniverse platform, Chemistry3D offers interfaces for robot operation, visual inspection, and liquid flow control, facilitating the simulation of special objects such as liquids and transparent entities. Leveraging this toolkit, we have devised RL tasks, object detection, and robot operation scenarios. Additionally, to discern disparities between the rendering engine and the real world, we conducted transparent object detection experiments using Sim2Real, validating the toolkit's exceptional simulation performance. The source code is available at https://github.com/huangyan28/Chemistry3D, and a related tutorial can be found at https://www.omni-chemistry.com.