Yaning Jia, Chunhui Zhang, Xingjian Diao, Xiangchi Yuan, Zhongyu Ouyang, Chiyu Ma, Soroush Vosoughi. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Whole slide images, with their gigapixel-scale panoramas of tissue samples, are pivotal for precise disease diagnosis. However, their analysis is hindered by immense data size and scarce annotations. Existing MIL methods face challenges due to the fundamental imbalance where a single bag-level label must guide the learning of numerous patch-level features. This sparse supervision makes it difficult to reliably identify discriminative patches during training, leading to unstable optimization and suboptimal solutions. We propose a spatially regularized MIL framework that leverages inherent spatial relationships among patch features as label-independent regularization signals. Our approach learns a shared representation space by jointly optimizing feature-induced spatial reconstruction and label-guided classification objectives, enforcing consistency between intrinsic structural patterns and supervisory signals. Experimental results on multiple public datasets demonstrate significant improvements over state-of-the-art methods, offering a promising direction. The code is available at https://github.com/wwyi1828/SRMIL.

Visual Question Answering (VQA) is increasingly used in diverse applications ranging from general visual reasoning to safety-critical domains such as medical imaging and autonomous systems, where models must provide not only accurate answers but also explanations that humans can easily understand and verify.Prototype-based modeling has shown promise for interpretability by grounding predictions in semantically meaningful regions for purely visual reasoning tasks, yet remains underexplored in the context of VQA.We present ProtoVQA, a unified prototypical framework that (i) learns question-aware prototypes that serve as reasoning anchors, connecting answers to discriminative image regions, (ii) applies spatially constrained matching to ensure that the selected evidence is coherent and semantically relevant, and (iii) supports both answering and grounding tasks through a shared prototype backbone.To assess explanation quality, we propose the Visual-Linguistic Alignment Score (VLAS), which measures how well the model's attended regions align with ground-truth evidence.Experiments on Visual7W show that ProtoVQA yields faithful, fine-grained explanations while maintaining competitive accuracy, advancing the development of transparent and trustworthy VQA systems.

Xingjian Diao, Chunhui Zhang, Keyi Kong, Weiyi Wu, Chiyu Ma, Zhongyu Ouyang, Peijun Qing, Soroush Vosoughi, Jiang Gui. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Talking face generation has gained immense popularity in the computer vision community, with various applications including AR, VR, teleconferencing, digital assistants, and avatars. Traditional methods are mainly audio-driven, which have to deal with the inevitable resource-intensive nature of audio storage and processing. To address such a challenge, we propose FT2TF - First-Person Statement Text- To -Talking Face Generation, a novel one-stage end-to-end pipeline for talking face generation driven by first-person statement text. Different from previous work, our model only leverages visual and textual information without any other sources (e.g., audio/landmark/pose) during inference. Extensive experiments are conducted on LRS2 and LRS3 datasets, and results on multidimensional evaluation metrics are reported. Both quantitative and qual-itative results showcase that FT2TF outperforms existing relevant methods and reaches the state-of-the-art. This achievement highlights our model's capability to bridge first-person statements and dynamic face generation, providing insightful guidance for future work.

Parameter-efficient fine-tuning methods, such as Low-Rank Adaptation (LoRA), are known to enhance training efficiency in Large Language Models (LLMs).Due to the limited parameters of LoRA, recent studies seek to combine LoRA with Mixture-of-Experts (MoE) to boost performance across various tasks.However, inspired by the observed redundancy in traditional MoE structures, prior studies find that LoRA experts within the MoE architecture also exhibit redundancy, suggesting a need to vary the allocation of LoRA experts across different layers.In this paper, we leverage Heavy-Tailed Self-Regularization (HT-SR) Theory to design a fine-grained allocation strategy.Our analysis reveals that the number of experts per layer correlates with layer training quality, which exhibits significant variability across layers.Based on this, we introduce AlphaLoRA, a theoretically principled and training-free method for allocating LoRA experts to reduce redundancy further.Experiments on three models across ten language processing and reasoning benchmarks demonstrate that AlphaLoRA achieves comparable or superior performance over all baselines.Our code is available at https://github.com/morelife2017/alphalora.

Learning high-quality video representation has shown significant applications in computer vision and remains challenging. Previous work based on mask autoencoders such as ImageMAE [10] and VideoMAE [23] has proven the effectiveness of learning representations in images and videos through reconstruction strategy in the visual modality. However, these models exhibit inherent limitations, particularly in scenarios where extracting features solely from the visual modality proves challenging, such as when dealing with low-resolution and blurry original videos. Based on this, we propose AV-MaskEnhancer for learning high-quality video representation by combining visual and audio information. Our approach addresses the challenge by demonstrating the complementary nature of audio and video features in cross-modality content. Moreover, our result of the video classification task on the UCF101 [21] dataset outperforms the existing work and reaches the state-of-the-art, with a top-1 accuracy of 98.8% and a top-5 accuracy of 99.9%.

Automated detection of intake gestures with wearable sensors has been a critical area of research for advancing our understanding and ability to intervene in people's eating behavior. Numerous algorithms have been developed and evaluated in terms of accuracy. However, ensuring the system is not only accurate in making predictions but also efficient in doing so is critical for real-world deployment. Despite the growing research on accurate detection of intake gestures using wearables, many of these algorithms are often energy inefficient, impeding on-device deployment for continuous and real-time monitoring of diet. This article presents a template-based optimized multicenter classifier that enables accurate intake gesture detection while maintaining low-inference time and energy consumption using a wrist-worn accelerometer and gyroscope. We designed an Intake Gesture Counter smartphone application (CountING) and validated the practicality of our algorithm against seven state-of-the-art approaches on three public datasets (In-lab FIC, Clemson, and OREBA). Compared with other methods, we achieved optimal accuracy (81.60% F1 score) and very low inference time (15.97 msec per 2.20-sec data sample) on the Clemson dataset, and among the top performing algorithms, we achieve comparable accuracy (83.0% F1 score compared with 85.6% in the top performing algorithm) but superior inference time (13.8x faster, 33.14 msec per 2.20-sec data sample) on the In-lab FIC dataset and comparable accuracy (83.40% F1 score compared with 88.10% in the top-performing algorithm) but superior inference time (33.9x faster, 16.71 msec inference time per 2.20-sec data sample) on the OREBA dataset. On average, our approach achieved a 25-hour battery lifetime (44% to 52% improvement over state-of-the-art approaches) when tested on a commercial smartwatch for continuous real-time detection. Our approach demonstrates an effective and efficient method, enabling real-time intake gesture detection using wrist-worn devices in longitudinal studies.