Bin Zou

Chengfei Lv, Chaoyue Niu, Renjie Gu et al.

To break the bottlenecks of mainstream cloud-based machine learning (ML) paradigm, we adopt device-cloud collaborative ML and build the first end-to-end and general-purpose system, called Walle, as the foundation. Walle consists of a deployment platform, distributing ML tasks to billion-scale devices in time; a data pipeline, efficiently preparing task input; and a compute container, providing a cross-platform and high-performance execution environment, while facilitating daily task iteration. Specifically, the compute container is based on Mobile Neural Network (MNN), a tensor compute engine along with the data processing and model execution libraries, which are exposed through a refined Python thread-level virtual machine (VM) to support diverse ML tasks and concurrent task execution. The core of MNN is the novel mechanisms of operator decomposition and semi-auto search, sharply reducing the workload in manually optimizing hundreds of operators for tens of hardware backends and further quickly identifying the best backend with runtime optimization for a computation graph. The data pipeline introduces an on-device stream processing framework to enable processing user behavior data at source. The deployment platform releases ML tasks with an efficient push-then-pull method and supports multi-granularity deployment policies. We evaluate Walle in practical e-commerce application scenarios to demonstrate its effectiveness, efficiency, and scalability. Extensive micro-benchmarks also highlight the superior performance of MNN and the Python thread-level VM. Walle has been in large-scale production use in Alibaba, while MNN has been open source with a broad impact in the community.

Zhefan Rao, Bin Zou, Haoxuan Che et al.

Instruction-based video editing is a natural way to control video content with text, but adapting a video generation model into an editor usually appears data-hungry. At the same time, high-quality video editing data remains scarce. In this paper, we show that a video generation backbone can become a strong video editor without large scale video editing data. We present InsEdit, an instruction-based editing model built on HunyuanVideo-1.5. InsEdit combines a visual editing architecture with a video data pipeline based on Mutual Context Attention (MCA), which creates aligned video pairs where edits can begin in the middle of a clip rather than only from the first frame. With only O(100)K video editing data, InsEdit achieves state-of-the-art results among open-source methods on our video instruction editing benchmarks. In addition, because our training recipe also includes image editing data, the final model supports image editing without any modification.

Yiran Du, Jinlong Li, Huimin He et al.

This study investigates Chinese primary school students' acceptance of a social robot for English-as-a-foreign-language (EFL) speaking practice through a sequential explanatory mixed-methods design. Integrating the Technology Acceptance Model (TAM) and the Computers Are Social Actors (CASA) paradigm, the research explores both functional and social factors influencing learners' behavioural intention to use the robot. Quantitative data from 436 students were analysed using structural equation modelling, followed by qualitative interviews with twelve students to interpret the findings. Results show that perceived enjoyment and ease of use are the strongest predictors of acceptance, while social attributes such as warmth, anthropomorphism, and social presence significantly enhance enjoyment. Perceived intelligence affects usefulness but not ease of use. The findings suggest that emotional and social engagement are central to young learners' acceptance of educational robots, highlighting the importance of designing socially intelligent technologies that promote motivation and speaking confidence in EFL learning contexts.

Bin Zou, Yijia Yuan, Chenghao Wang et al.

This study examined whether AI-mediated speaking practice can reduce acculturative stress among Chinese international students in UK universities. Using a sequential explanatory mixed-methods design, 126 participants were randomly assigned to an experimental group, which completed a four-week intervention using EAP Talk, an AI-assisted English for Academic Purposes speaking platform offering role play, scenario-based practice, free talk, and automated feedback, or a control group, which continued usual academic and English-learning activities. Pre- and post-test questionnaires measured perceived language insufficiency, social isolation, and academic pressure, while semi-structured interviews with 20 experimental-group participants contextualised the quantitative findings. Linear mixed-effects models showed that the experimental group experienced significantly greater reductions than the control group across all three outcomes, with the strongest effect on perceived language insufficiency. Interview findings suggested that EAP Talk supported low-stakes rehearsal, communicative confidence, academic speaking preparation, and greater willingness to initiate social interaction. However, participants also noted that AI-mediated practice could not fully reproduce authentic human interaction, disciplinary feedback, or broader institutional support. The findings suggest that AI-mediated speaking practice can function as a supplementary scaffold for reducing communication-related dimensions of acculturative stress, but should be integrated with peer interaction, teacher feedback, and wider support services.

Hui Liu, Bin Zou, Suiyun Zhang et al.

Instruction-guided image editing enables users to specify modifications using natural language, offering more flexibility and control. Among existing frameworks, Diffusion Transformers (DiTs) outperform U-Net-based diffusion models in scalability and performance. However, while real-world scenarios often require concurrent execution of multiple instructions, step-by-step editing suffers from accumulated errors and degraded quality, and integrating multiple instructions with a single prompt usually results in incomplete edits due to instruction conflicts. We propose Instruction Influence Disentanglement (IID), a novel framework enabling parallel execution of multiple instructions in a single denoising process, designed for DiT-based models. By analyzing self-attention mechanisms in DiTs, we identify distinctive attention patterns in multi-instruction settings and derive instruction-specific attention masks to disentangle each instruction's influence. These masks guide the editing process to ensure localized modifications while preserving consistency in non-edited regions. Extensive experiments on open-source and custom datasets demonstrate that IID reduces diffusion steps while improving fidelity and instruction completion compared to existing baselines. The codes will be publicly released upon the acceptance of the paper.

Xiaotang Jiang, Huan Wang, Yiliu Chen et al.

Deploying deep learning models on mobile devices draws more and more attention recently. However, designing an efficient inference engine on devices is under the great challenges of model compatibility, device diversity, and resource limitation. To deal with these challenges, we propose Mobile Neural Network (MNN), a universal and efficient inference engine tailored to mobile applications. In this paper, the contributions of MNN include: (1) presenting a mechanism called pre-inference that manages to conduct runtime optimization; (2)deliveringthorough kernel optimization on operators to achieve optimal computation performance; (3) introducing backend abstraction module which enables hybrid scheduling and keeps the engine lightweight. Extensive benchmark experiments demonstrate that MNN performs favorably against other popular lightweight deep learning frameworks. MNN is available to public at: https://github.com/alibaba/MNN.

Hui Liu, Bin Zou, Kecheng Chen et al.

Large language models (LLMs) exhibit substantial variability in performance and computational cost across tasks and queries, motivating routing systems that select models to meet user-specific cost-performance trade-offs. However, existing routers generalize poorly in cold-start scenarios where in-domain training data is unavailable. We address this limitation with a multi-level task-profile-guided data synthesis framework that constructs a hierarchical task taxonomy and produces diverse question-answer pairs to approximate the test-time query distribution. Building on this, we introduce TRouter, a task-type-aware router approach that models query-conditioned cost and performance via latent task-type variables, with prior regularization derived from the synthesized task taxonomy. This design enhances TRouter's routing utility under both cold-start and in-domain settings. Across multiple benchmarks, we show that our synthesis framework alleviates cold-start issues and that TRouter delivers effective LLM routing.

Jinqi Zhang, Fangzhou Han, Di Zhuang et al.

In recent years, Deep Learning (DL) based methods have received extensive and sufficient attention in the field of PolSAR image classification, which show excellent performance. However, due to the ``black-box" nature of DL methods, the interpretation of the high-dimensional features extracted and the backtracking of the decision-making process based on the features are still unresolved problems. In this study, we first highlight this issue and attempt to achieve the interpretability analysis of DL-based PolSAR image classification technology with the help of Polarimetric Target Decomposition (PTD), a feature extraction method related to the scattering mechanism unique to the PolSAR image processing field. In our work, by constructing the polarimetric conceptual labels and a novel structure named Parallel Concept Bottleneck Networks (PaCBM), the uninterpretable high-dimensional features are transformed into human-comprehensible concepts based on physically verifiable polarimetric scattering mechanisms. Then, the Kolmogorov-Arnold Network (KAN) is used to replace Multi-Layer Perceptron (MLP) for achieving a more concise and understandable mapping process between layers and further enhanced non-linear modeling ability. The experimental results on several PolSAR datasets show that the features could be conceptualization under the premise of achieving satisfactory accuracy through the proposed pipeline, and the analytical function for predicting category labels from conceptual labels can be obtained by combining spline functions, thus promoting the research on the interpretability of the DL-based PolSAR image classification model.

Jiang Qin, Bin Zou, Haolin Li et al.

In recent years, continuous improvements in SAR resolution have significantly benefited applications such as urban monitoring and target detection. However, the improvement in resolution leads to increased discrepancies in scattering characteristics, posing challenges to the generalization ability of target detection models. While domain adaptation technology is a potential solution, the inevitable discrepancies caused by resolution differences often lead to blind feature adaptation and unreliable semantic propagation, ultimately degrading the domain adaptation performance. To address these challenges, this paper proposes a novel SAR target detection method (termed CR-Net), that incorporates structure priors and evidential learning theory into the detection model, enabling reliable domain adaptation for cross-resolution detection. To be specific, CR-Net integrates Structure-induced Hierarchical Feature Adaptation (SHFA) and Reliable Structural Adjacency Alignment (RSAA). SHFA module is introduced to establish structural correlations between targets and achieve structure-aware feature adaptation, thereby enhancing the interpretability of the feature adaptation process. Afterwards, the RSAA module is proposed to enhance reliable semantic alignment, by leveraging the secure adjacency set to transfer valuable discriminative knowledge from the source domain to the target domain. This further improves the discriminability of the detection model in the target domain. Based on experimental results from different-resolution datasets,the proposed CR-Net significantly enhances cross-resolution adaptation by preserving intra-domain structures and improving discriminability. It achieves state-of-the-art (SOTA) performance in cross-resolution SAR target detection.

Lamei Zhang, Siyu Zhang, Bin Zou et al.

Deep learning and convolutional neural networks (CNNs) have made progress in polarimetric synthetic aperture radar (PolSAR) image classification over the past few years. However, a crucial issue has not been addressed, i.e., the requirement of CNNs for abundant labeled samples versus the insufficient human annotations of PolSAR images. It is well-known that following the supervised learning paradigm may lead to the overfitting of training data, and the lack of supervision information of PolSAR images undoubtedly aggravates this problem, which greatly affects the generalization performance of CNN-based classifiers in large-scale applications. To handle this problem, in this paper, learning transferrable representations from unlabeled PolSAR data through convolutional architectures is explored for the first time. Specifically, a PolSAR-tailored contrastive learning network (PCLNet) is proposed for unsupervised deep PolSAR representation learning and few-shot classification. Different from the utilization of optical processing methods, a diversity stimulation mechanism is constructed to narrow the application gap between optics and PolSAR. Beyond the conventional supervised methods, PCLNet develops an unsupervised pre-training phase based on the proxy objective of instance discrimination to learn useful representations from unlabeled PolSAR data. The acquired representations are transferred to the downstream task, i.e., few-shot PolSAR classification. Experiments on two widely-used PolSAR benchmark datasets confirm the validity of PCLNet. Besides, this work may enlighten how to efficiently utilize the massive unlabeled PolSAR data to alleviate the greedy demands of CNN-based methods for human annotations.

Hongwei Dong, Siyu Zhang, Bin Zou et al.

Convolutional neural networks (CNNs) have shown good performance in polarimetric synthetic aperture radar (PolSAR) image classification due to the automation of feature engineering. Excellent hand-crafted architectures of CNNs incorporated the wisdom of human experts, which is an important reason for CNN's success. However, the design of the architectures is a difficult problem, which needs a lot of professional knowledge as well as computational resources. Moreover, the architecture designed by hand might be suboptimal, because it is only one of thousands of unobserved but objective existed paths. Considering that the success of deep learning is largely due to its automation of the feature engineering process, how to design automatic architecture searching methods to replace the hand-crafted ones is an interesting topic. In this paper, we explore the application of neural architecture search (NAS) in PolSAR area for the first time. Different from the utilization of existing NAS methods, we propose a differentiable architecture search (DAS) method which is customized for PolSAR classification. The proposed DAS is equipped with a PolSAR tailored search space and an improved one-shot search strategy. By DAS, the weights parameters and architecture parameters (corresponds to the hyperparameters but not the topologies) can be optimized by stochastic gradient descent method during the training. The optimized architecture parameters should be transformed into corresponding CNN architecture and re-train to achieve high-precision PolSAR classification. In addition, complex-valued DAS is developed to take into account the characteristics of PolSAR images so as to further improve the performance. Experiments on three PolSAR benchmark datasets show that the CNNs obtained by searching have better classification performance than the hand-crafted ones.

Hongwei Dong, Lamei Zhang, Bin Zou

Redundancy and noise exist in the bands of hyperspectral images (HSIs). Thus, it is a good property to be able to select suitable parts from hundreds of input bands for HSIs classification methods. In this letter, a band attention module (BAM) is proposed to implement the deep learning based HSIs classification with the capacity of band selection or weighting. The proposed BAM can be seen as a plug-and-play complementary component of the existing classification networks which fully considers the adverse effects caused by the redundancy of the bands when using convolutional neural networks (CNNs) for HSIs classification. Unlike most of deep learning methods used in HSIs, the band attention module which is customized according to the characteristics of hyperspectral images is embedded in the ordinary CNNs for better performance. At the same time, unlike classical band selection or weighting methods, the proposed method achieves the end-to-end training instead of the separated stages. Experiments are carried out on two HSI benchmark datasets. Compared to some classical and advanced deep learning methods, numerical simulations under different evaluation criteria show that the proposed method have good performance. Last but not least, some advanced CNNs are combined with the proposed BAM for better performance.

Lamei Zhang, Hongwei Dong, Bin Zou

Convolutional neural networks (CNNs) have been widely used to improve the accuracy of polarimetric synthetic aperture radar (PolSAR) image classification. However, in most studies, the difference between PolSAR images and optical images is rarely considered. Most of the existing CNNs are not tailored for the task of PolSAR image classification, in which complex-valued PolSAR data have been simply equated to real-valued data to fit the optical image processing architectures and avoid complex-valued operations. This is one of the reasons CNNs unable to perform their full capabilities in PolSAR classification. To solve the above problem, the objective of this paper is to develop a tailored CNN framework for PolSAR image classification, which can be implemented from two aspects: Seeking a better form of PolSAR data as the input of CNNs and building matched CNN architectures based on the proposed input form. In this paper, considering the properties of complex-valued numbers, amplitude and phase of complex-valued PolSAR data are extracted as the input for the first time to maintain the integrity of original information while avoiding immature complex-valued operations. Then, a multi-task CNN (MCNN) architecture is proposed to match the improved input form and achieve better classification results. Furthermore, depthwise separable convolution is introduced to the proposed architecture in order to better extract information from the phase information. Experiments on three PolSAR benchmark datasets not only prove that using amplitude and phase as the input do contribute to the improvement of PolSAR classification, but also verify the adaptability between the improved input form and the well-designed architectures.