Liang Gao

Liang Gao, Huazhu Fu, Li Li et al.

Federated learning (FL) allows multiple clients to collectively train a high-performance global model without sharing their private data. However, the key challenge in federated learning is that the clients have significant statistical heterogeneity among their local data distributions, which would cause inconsistent optimized local models on the client-side. To address this fundamental dilemma, we propose a novel federated learning algorithm with local drift decoupling and correction (FedDC). Our FedDC only introduces lightweight modifications in the local training phase, in which each client utilizes an auxiliary local drift variable to track the gap between the local model parameter and the global model parameters. The key idea of FedDC is to utilize this learned local drift variable to bridge the gap, i.e., conducting consistency in parameter-level. The experiment results and analysis demonstrate that FedDC yields expediting convergence and better performance on various image classification tasks, robust in partial participation settings, non-iid data, and heterogeneous clients.

Chen Sun, Liang Gao, Xinyu Li et al.

Surface defect detection is one of the most essential processes for industrial quality inspection. Deep learning-based surface defect detection methods have shown great potential. However, the well-performed models usually require large training data and can only detect defects that appeared in the training stage. When facing incremental few-shot data, defect detection models inevitably suffer from catastrophic forgetting and misclassification problem. To solve these problems, this paper proposes a new knowledge distillation network, called Dual Knowledge Align Network (DKAN). The proposed DKAN method follows a pretraining-finetuning transfer learning paradigm and a knowledge distillation framework is designed for fine-tuning. Specifically, an Incremental RCNN is proposed to achieve decoupled stable feature representation of different categories. Under this framework, a Feature Knowledge Align (FKA) loss is designed between class-agnostic feature maps to deal with catastrophic forgetting problems, and a Logit Knowledge Align (LKA) loss is deployed between logit distributions to tackle misclassification problems. Experiments have been conducted on the incremental Few-shot NEU-DET dataset and results show that DKAN outperforms other methods on various few-shot scenes, up to 6.65% on the mean Average Precision metric, which proves the effectiveness of the proposed method.

Yunkang Cao, Xiaohao Xu, Chen Sun et al.

This technical report introduces the winning solution of the team Segment Any Anomaly for the CVPR2023 Visual Anomaly and Novelty Detection (VAND) challenge. Going beyond uni-modal prompt, e.g., language prompt, we present a novel framework, i.e., Segment Any Anomaly + (SAA$+$), for zero-shot anomaly segmentation with multi-modal prompts for the regularization of cascaded modern foundation models. Inspired by the great zero-shot generalization ability of foundation models like Segment Anything, we first explore their assembly (SAA) to leverage diverse multi-modal prior knowledge for anomaly localization. Subsequently, we further introduce multimodal prompts (SAA$+$) derived from domain expert knowledge and target image context to enable the non-parameter adaptation of foundation models to anomaly segmentation. The proposed SAA$+$ model achieves state-of-the-art performance on several anomaly segmentation benchmarks, including VisA and MVTec-AD, in the zero-shot setting. We will release the code of our winning solution for the CVPR2023 VAN.

Jin Huang, Jie Yang, XinLei Zhou et al.

In dynamic manufacturing environments, disruptions such as machine breakdowns and new order arrivals continuously shift the optimal dispatching strategy, making adaptive rule selection essential. Existing LLM-powered Automatic Heuristic Design (AHD) frameworks evolve toward a single elite rule that cannot meet this adaptability demand. To address this, we present DSevolve, an industrial scheduling framework that evolves a quality-diverse portfolio of dispatching rules offline and adaptively deploys them online with second-level response time. Multi-persona seeding and topology-aware evolutionary operators produce a behaviorally diverse rule archive indexed by a MAP-Elites feature space. Upon each disruption event, a probe-based fingerprinting mechanism characterizes the current shop floor state, retrieves high-quality candidate rules from an offline knowledge base, and selects the best one via rapid look-ahead simulation. Evaluated on 500 dynamic flexible job shop instances derived from real industrial data, DSevolve outperforms state-of-the-art AHD frameworks, classical dispatching rules, genetic programming, and deep reinforcement learning, offering a practical and deployable solution for intelligent shop floor scheduling.

Meituan LongCat Team, Bayan, Bei Li et al.

We introduce LongCat-Flash, a 560-billion-parameter Mixture-of-Experts (MoE) language model designed for both computational efficiency and advanced agentic capabilities. Stemming from the need for scalable efficiency, LongCat-Flash adopts two novel designs: (a) Zero-computation Experts, which enables dynamic computational budget allocation and activates 18.6B-31.3B (27B on average) per token depending on contextual demands, optimizing resource usage. (b) Shortcut-connected MoE, which enlarges the computation-communication overlap window, demonstrating notable gains in inference efficiency and throughput compared to models of a comparable scale. We develop a comprehensive scaling framework for large models that combines hyperparameter transfer, model-growth initialization, a multi-pronged stability suite, and deterministic computation to achieve stable and reproducible training. Notably, leveraging the synergy among scalable architectural design and infrastructure efforts, we complete model training on more than 20 trillion tokens within 30 days, while achieving over 100 tokens per second (TPS) for inference at a cost of \$0.70 per million output tokens. To cultivate LongCat-Flash towards agentic intelligence, we conduct a large-scale pre-training on optimized mixtures, followed by targeted mid- and post-training on reasoning, code, and instructions, with further augmentation from synthetic data and tool use tasks. Comprehensive evaluations demonstrate that, as a non-thinking foundation model, LongCat-Flash delivers highly competitive performance among other leading models, with exceptional strengths in agentic tasks. The model checkpoint of LongCat-Flash is open-sourced to foster community research. LongCat Chat: https://longcat.ai Hugging Face: https://huggingface.co/meituan-longcat GitHub: https://github.com/meituan-longcat

Yunkang Cao, Xiaohao Xu, Chen Sun et al.

We present a novel framework, i.e., Segment Any Anomaly + (SAA+), for zero-shot anomaly segmentation with hybrid prompt regularization to improve the adaptability of modern foundation models. Existing anomaly segmentation models typically rely on domain-specific fine-tuning, limiting their generalization across countless anomaly patterns. In this work, inspired by the great zero-shot generalization ability of foundation models like Segment Anything, we first explore their assembly to leverage diverse multi-modal prior knowledge for anomaly localization. For non-parameter foundation model adaptation to anomaly segmentation, we further introduce hybrid prompts derived from domain expert knowledge and target image context as regularization. Our proposed SAA+ model achieves state-of-the-art performance on several anomaly segmentation benchmarks, including VisA, MVTec-AD, MTD, and KSDD2, in the zero-shot setting. We will release the code at \href{https://github.com/caoyunkang/Segment-Any-Anomaly}{https://github.com/caoyunkang/Segment-Any-Anomaly}.

Fuxiang Tan, YuTing Kong, Yingying Fan et al.

Rain streaks degrade the image quality and seriously affect the performance of subsequent computer vision tasks, such as autonomous driving, social security, etc. Therefore, removing rain streaks from a given rainy images is of great significance. Convolutional neural networks(CNN) have been widely used in image deraining tasks, however, the local computational characteristics of convolutional operations limit the development of image deraining tasks. Recently, the popular transformer has global computational features that can further facilitate the development of image deraining tasks. In this paper, we introduce Swin-transformer into the field of image deraining for the first time to study the performance and potential of Swin-transformer in the field of image deraining. Specifically, we improve the basic module of Swin-transformer and design a three-branch model to implement single-image rain removal. The former implements the basic rain pattern feature extraction, while the latter fuses different features to further extract and process the image features. In addition, we employ a jump connection to fuse deep features and shallow features. In terms of experiments, the existing public dataset suffers from image duplication and relatively homogeneous background. So we propose a new dataset Rain3000 to validate our model. Therefore, we propose a new dataset Rain3000 for validating our model. Experimental results on the publicly available datasets Rain100L, Rain100H and our dataset Rain3000 show that our proposed method has performance and inference speed advantages over the current mainstream single-image rain streaks removal models.The source code will be available at https://github.com/H-tfx/SDNet.

Yanhao Wu, Haoyang Zhang, Fei He et al.

End-to-end autonomous driving has rapidly progressed, enabling joint perception and planning in complex environments. In the planning stage, state-of-the-art (SOTA) end-to-end autonomous driving models decouple planning into parallel lateral and longitudinal predictions. While effective, this parallel design can lead to i) coordination failures between the planned path and speed, and ii) underutilization of the drive path as a prior for longitudinal planning, thus redundantly encoding static information. To address this, we propose a novel cascaded framework that explicitly conditions longitudinal planning on the drive path, enabling coordinated and collision-aware lateral and longitudinal planning. Specifically, we introduce a path-conditioned formulation that explicitly incorporates the drive path into longitudinal planning. Building on this, the model predicts longitudinal displacements along the drive path rather than full 2D trajectory waypoints. This design simplifies longitudinal reasoning and more tightly couples it with lateral planning. Additionally, we introduce a planning-oriented data augmentation strategy that simulates rare safety-critical events, such as vehicle cut-ins, by adding agents and relabeling longitudinal targets to avoid collision. Evaluated on the challenging Bench2Drive benchmark, our method sets a new SOTA, achieving a driving score of 89.07 and a success rate of 73.18%, demonstrating significantly improved coordination and safety

Mingming Peng, Zhendong Chen, Jie Yang et al.

With the accelerated development of Industry 4.0, intelligent manufacturing systems increasingly require efficient task allocation and scheduling in multi-robot systems. However, existing methods rely on domain expertise and face challenges in adapting to dynamic production constraints. Additionally, enterprises have high privacy requirements for production scheduling data, which prevents the use of cloud-based large language models (LLMs) for solution development. To address these challenges, there is an urgent need for an automated modeling solution that meets data privacy requirements. This study proposes a knowledge-augmented mixed integer linear programming (MILP) automated formulation framework, integrating local LLMs with domain-specific knowledge bases to generate executable code from natural language descriptions automatically. The framework employs a knowledge-guided DeepSeek-R1-Distill-Qwen-32B model to extract complex spatiotemporal constraints (82% average accuracy) and leverages a supervised fine-tuned Qwen2.5-Coder-7B-Instruct model for efficient MILP code generation (90% average accuracy). Experimental results demonstrate that the framework successfully achieves automatic modeling in the aircraft skin manufacturing case while ensuring data privacy and computational efficiency. This research provides a low-barrier and highly reliable technical path for modeling in complex industrial scenarios.

Wenze Xu, Chun Wang, Jiazhen Yu et al.

Spoken Language Models (SLMs), which extend Large Language Models (LLMs) to perceive speech inputs, have gained increasing attention for their potential to advance speech understanding tasks. However, despite recent progress, studies show that SLMs often struggle to generalize across datasets, even for trained languages and tasks, raising concerns about whether they process speech in a text-like manner as intended. A key challenge underlying this limitation is the modality gap between speech and text representations. The high variability in speech embeddings may allow SLMs to achieve strong in-domain performance by exploiting unintended speech variations, ultimately hindering generalization. To mitigate this modality gap, we introduce Optimal Transport Regularization (OTReg), a method that formulates speech-text alignment as an optimal transport problem and derives a regularization loss to improve SLM training. In each training iteration, OTReg first establishes a structured correspondence between speech and transcript embeddings by determining the optimal transport plan, then incorporates the regularization loss based on this transport plan to optimize SLMs in generating speech embeddings that align more effectively with transcript embeddings. OTReg is lightweight, requiring no additional labels or learnable parameters, and integrates seamlessly into existing SLM training procedures. Extensive multilingual ASR experiments demonstrate that OTReg enhances speech-text alignment, mitigates the modality gap, and consequently improves SLM generalization across diverse datasets.

Jin Xie, Xinyu Li, Liang Gao et al.

Job shop scheduling problem (JSP) is a widely studied NP-complete combinatorial optimization problem. Neighborhood structures play a critical role in solving JSP. At present, there are three state-of-the-art neighborhood structures, i.e., N5, N6, and N7. Improving the upper bounds of some famous benchmarks is inseparable from the role of these neighborhood structures. However, these existing neighborhood structures only consider the movement of critical operations within a critical block. According to our experiments, it is also possible to improve the makespan of a scheduling scheme by moving a critical operation outside its critical block. According to the above finding, this paper proposes a new N8 neighborhood structure considering the movement of critical operations within a critical block and the movement of critical operations outside the critical block. Besides, a neighborhood clipping method is designed to avoid invalid movement, reducing the computational time. Tabu search (TS) is a commonly used algorithm framework combined with neighborhood structures. This paper uses this framework to compare the N8 neighborhood structure with N5, N6, and N7 neighborhood structures on four famous benchmarks. The experimental results verify that the N8 neighborhood structure is more effective and efficient in solving JSP than the other state-of-the-art neighborhood structures.

Liang Gao, Kele Xu, Huaimin Wang et al.

As an important component of multimedia analysis tasks, audio classification aims to discriminate between different audio signal types and has received intensive attention due to its wide applications. Generally speaking, the raw signal can be transformed into various representations (such as Short Time Fourier Transform and Mel Frequency Cepstral Coefficients), and information implied in different representations can be complementary. Ensembling the models trained on different representations can greatly boost the classification performance, however, making inference using a large number of models is cumbersome and computationally expensive. In this paper, we propose a novel end-to-end collaborative learning framework for the audio classification task. The framework takes multiple representations as the input to train the models in parallel. The complementary information provided by different representations is shared by knowledge distillation. Consequently, the performance of each model can be significantly promoted without increasing the computational overhead in the inference stage. Extensive experimental results demonstrate that the proposed approach can improve the classification performance and achieve state-of-the-art results on both acoustic scene classification tasks and general audio tagging tasks.

Chen Sun, Ye Tian, Liang Gao et al.

Calibration models have been developed for determination of trace elements, silver for instance, in soil using laser-induced breakdown spectroscopy (LIBS). The major concern is the matrix effect. Although it affects the accuracy of LIBS measurements in a general way, the effect appears accentuated for soil because of large variation of chemical and physical properties among different soils. The purpose is to reduce its influence in such way an accurate and soil-independent calibration model can be constructed. At the same time, the developed model should efficiently reduce experimental fluctuations affecting measurement precision. A univariate model first reveals obvious influence of matrix effect and important experimental fluctuation. A multivariate model has been then developed. A key point is the introduction of generalized spectrum where variables representing the soil type are explicitly included. Machine learning has been used to develop the model. After a necessary pretreatment where a feature selection process reduces the dimension of raw spectrum accordingly to the number of available spectra, the data have been fed in to a back-propagation neuronal networks (BPNN) to train and validate the model. The resulted soilindependent calibration model allows average relative error of calibration (REC) and average relative error of prediction (REP) within the range of 5-6%.

Haozhen Dong, Liang Gao, Xinyu Li et al.

Differential evolution(DE) is a conventional algorithm with fast convergence speed. However, DE may be trapped in local optimal solution easily. Many researchers devote themselves to improving DE. In our previously work, whale swarm algorithm have shown its strong searching performance due to its niching based mutation strategy. Based on this fact, we propose a new DE algorithm called DE with Better and Nearest option (NbDE). In order to evaluate the performance of NbDE, NbDE is compared with several meta-heuristic algorithms on nine classical benchmark test functions with different dimensions. The results show that NbDE outperforms other algorithms in convergence speed and accuracy.

Bing Zeng, Xinyu Li, Liang Gao et al.

Most real-world optimization problems often come with multiple global optima or local optima. Therefore, increasing niching metaheuristic algorithms, which devote to finding multiple optima in a single run, are developed to solve these multimodal optimization problems. However, there are two difficulties urgently to be solved for most existing niching metaheuristic algorithms: how to set the optimal values of niching parameters for different optimization problems, and how to jump out of the local optima efficiently. These two difficulties limited their practicality largely. Based on Whale Swarm Algorithm (WSA) we proposed previously, this paper presents a new multimodal optimizer named WSA with Iterative Counter (WSA-IC) to address these two difficulties. In the one hand, WSA-IC improves the iteration rule of the original WSA for multimodal optimization, which removes the need of specifying different values of attenuation coefficient for different problems to form multiple subpopulations, without introducing any niching parameter. In the other hand, WSA-IC enables the identification of extreme point during iterations relying on two new parameters (i.e., stability threshold Ts and fitness threshold Tf), to jump out of the located extreme point. Moreover, the convergence of WSA-IC is proved. Finally, the proposed WSA-IC is compared with several niching metaheuristic algorithms on CEC2015 niching benchmark test functions and five additional classical multimodal functions with high dimensions. The experimental results demonstrate that WSA-IC statistically outperforms other niching metaheuristic algorithms on most test functions.

Wei Cui, Liang Gao

We present an optical mapping near-eye (OMNI) three-dimensional display method for wearable devices. By dividing a display screen into different sub-panels and optically mapping them to various depths, we create a multiplane volumetric image with correct focus cues for depth perception. The resultant system can drive the eye's accommodation to the distance that is consistent with binocular stereopsis, thereby alleviating the vergence-accommodation conflict, the primary cause for eye fatigue and discomfort. Compared with the previous methods, the OMNI display offers prominent advantages in adaptability, image dynamic range, and refresh rate.

Bing Zeng, Liang Gao, Xinyu Li

Increasing nature-inspired metaheuristic algorithms are applied to solving the real-world optimization problems, as they have some advantages over the classical methods of numerical optimization. This paper has proposed a new nature-inspired metaheuristic called Whale Swarm Algorithm for function optimization, which is inspired by the whales behavior of communicating with each other via ultrasound for hunting. The proposed Whale Swarm Algorithm has been compared with several popular metaheuristic algorithms on comprehensive performance metrics. According to the experimental results, Whale Swarm Algorithm has a quite competitive performance when compared with other algorithms.