Tianle Liang, Yifu Chen, Shengpeng Ji, Yijun Chen, Zhiyang Jia, Jingyu Lu, Fan Zhuo, Xueyi Pu, Yangzhuo Li, Zhou Zhao. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Jingyu Lu, Yuhan Wang, Fan Zhuo, Xize Cheng, Changhao Pan, Xueyi Pu, Yifu Chen, Chenyuhao Wen, Tianle Liang, Zhou Zhao. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Achieving robust generalization across diverse scenarios is crucial for advancing the practical application of robotics. Existing approaches typically rely on interaction object masks as visual inputs to predict subsequent actions, gaining a certain degree of generalization capability. However, these methods primarily map visual inputs and task-relevant object masks to expert actions, overlooking the environmental dynamics that govern physical interactions among objects during manipulation. To overcome these limitations, we introduce ABM++, a novel framework that leverages pre-trained VLMs to build a mask-guided world model (MGWM) within an imitation learning paradigm for generalized robotic manipulation. Specifically, we extend a world model into a coarse-to-fine imitation learning framework to reconstruct future mask-dominated visual features, which enables the model to capture state transitions between the current and next states based on predicted actions, effectively modeling environmental dynamics. Comprehensive experiments demonstrate that ABM++ significantly surpasses established baselines in both simulation and real-world environments, achieving a relative improvement of 12.2% across 8 complex tasks, which underscores the superiority of our method.

Current lifelong person re-identification (LReID) methods predominantly rely on fully labeled data streams. However, in real-world scenarios where annotation resources are limited, a vast amount of unlabeled data coexists with scarce labeled samples, leading to the Semi-Supervised LReID (Semi-LReID) problem where LReID methods suffer severe performance degradation. Existing LReID methods, even when combined with semi-supervised strategies, suffer from limited long-term adaptation performance due to struggling with the noisy knowledge occurring during unlabeled data utilization. In this paper, we pioneer the investigation of Semi-LReID, introducing a novel Self-Reinforcing Prototype Evolution with Dual-Knowledge Cooperation framework (SPRED). Our key innovation lies in establishing a self-reinforcing cycle between dynamic prototype-guided pseudo-label generation and new-old knowledge collaborative purification to enhance the utilization of unlabeled data. Specifically, learnable identity prototypes are introduced to dynamically capture the identity distributions and generate high-quality pseudo-labels. Then, the dual-knowledge cooperation scheme integrates current model specialization and historical model generalization, refining noisy pseudo-labels. Through this cyclic design, reliable pseudo-labels are progressively mined to improve current-stage learning and ensure positive knowledge propagation over long-term learning. Experiments on the established Semi-LReID benchmarks show that our SPRED achieves state-of-the-art performance. Our source code is available at https://github.com/zhoujiahuan1991/ICCV2025-SPRED

Lifelong person re-identification (LReID) suffers from the catastrophic forgetting problem when learning from non-stationary data streams. Existing exemplar-based and knowledge distillation-based LReID methods encounter data privacy and limited acquisition capacity, respectively. In this paper, we introduce the prototype, which is under-investigated in LReID, to better balance knowledge retention and acquisition. Previous prototype-based works primarily focused on the classification task, where prototypes were modeled as discrete points or statistical distributions. However, they either discarded the distribution information or omitted instance-level diversity, which are crucial fine-grained clues for LReID. Furthermore, the domain shifts between data sources result in a feature gap between the new and old data, which restricts the utilization of the fine-grained information in prototypes. To address these challenges, we propose Distribution-aware Knowledge Aligning and Prototyping (DKP++), a novel framework for modeling and leveraging prototypes in LReID. First, an Instance-level Distribution Modeling network is introduced to capture the local diversity of each instance. Next, a Distribution-oriented Prototype Generation algorithm transforms the instance-level diversity into identity-level distributions which are stored as prototypes. Then, a Prototype-based Knowledge Transfer module distills the knowledge within the prototypes to the new model. To mitigate the impact of domain shifts during knowledge transfer, we introduce a privacy-friendly Distribution Aligning module that transforms new input data to fit the historical distribution, which is incorporated with feature-level alignment constraints to enhance the coherence between new and old knowledge, effectively improving historical prototype utilization. Extensive experiments demonstrate that our method achieves a superior balance between plasticity and stability, outperforming state-of-the-art LReID methods by a large margin.

The recent success of audio language models (LMs) has revolutionized the field of neural music generation. Among all audio LM approaches, MusicGen has demonstrated the success of a single-stage LMs based music generation framework, without needing to train multiple LMs. Despite its promising performance in generating monophonic (mono) music, directly generating stereophonic (stereo) music following the previous framework has resulted in perceptible quality degradation. In this paper, we first discuss the difficulty of directly encoding stereo music with neural codec, and then provide a stable and practical solution based on a dual encoding approach. To utilize the dually encoded tokens in single-stage LMs, we also propose two forms of token sequence patterns. An extensive evaluation has been conducted using various aspects of stereo music audios to examine the performance of stereo neural codec approaches and the generation quality of single-stage LMs. Finally, our experimental results suggest that (i) our proposed dual encoding approach for neural codec is significantly better than the typical joint encoding approach in terms of reconstruction quality, and (ii) the stereo single-stage LMs trained with our proposed token sequence patterns substantially improved the perceptual quality of the state-of-the-art music generation model (i.e. MusicGen) in subjective tests.

In 3D object manipulation, collecting expert data for end-to-end imitation learning becomes a mainstream method. Though successful, previous works neglect the guiding role of language in action execution. These methods lack the understanding of action semantics, in which multiple action sequences are guided by a category of instructions, resulting in overlearned object semantics and vague action semantics. To address the above limitation, we introduce a novel framework named LLaKey, which breaks down skill commands into more detailed action instructions based on key states for fine-grained action control. Specifically, LLaKey first leverages the knowledge encoded in pre-trained large-scale models to fine-tune an action instruction conductor. Then, these instructions are executed by a downstream action model. Comprehensive experiments show that LLaKey significantly surpasses baselines with a relative improvement of 15% in nine complex and varied skill tasks, demonstrating the superiority of our method.