Anomaly detection (AD) is a fundamental task of critical importance across numerous domains. Current systems increasingly operate in rapidly evolving environments that generate diverse yet interconnected data modalities—such as time series, system logs, and tabular records—as exemplified by modern IT systems. Effective AD methods in such environments must therefore possess two critical capabilities: (1) the ability to handle heterogeneous data formats within a unified framework, allowing the model to process and detect multiple modalities in a consistent manner during anomalous events; (2) a strong generalization ability to quickly adapt to new scenarios without extensive retraining. However, most existing methods fall short of these requirements, as they typically focus on single modalities and lack the flexibility to generalize across domains. To address this gap, we introduce a novel paradigm: In-Context Anomaly Detection (ICAD), where anomalies are defined by their dissimilarity to a relevant reference set of normal samples. Under this paradigm, we propose ICAD-LLM, a unified AD framework leveraging Large Language Models' in-context learning abilities to process heterogeneous data within a single model. Extensive experiments demonstrate that ICAD-LLM achieves competitive performance with task-specific AD methods and exhibits strong generalization to previously unseen tasks, which substantially reduces deployment costs and enables rapid adaptation to new environments. To the best of our knowledge, ICAD-LLM is the first model capable of handling anomaly detection tasks across diverse domains and modalities.

Code generation plays a crucial role in various tasks, such as code auto-completion and mathematical reasoning.Previous work has proposed numerous methods to enhance code generation performance, including integrating feedback from the compiler.Inspired by this, we present ReflectionCoder, a novel approach that effectively leverages reflection sequences constructed by integrating compiler feedback to improve one-off code generation performance.Furthermore, we propose reflection self-distillation and dynamically masked distillation to effectively utilize these reflection sequences.Extensive experiments on three benchmarks, i.e., HumanEval (+), MBPP (+), and MultiPL-E, demonstrate that models fine-tuned with our method achieve state-of-the-art performance.Beyond the code domain, we believe this approach can benefit other domains that focus on final results and require long reasoning paths.

In infilling tasks, sub-tokens, representing instances where a complete token is segmented into two parts, often emerge at the boundaries of prefixes, middles, and suffixes.Traditional methods focused on training models at the token level, leading to sub-optimal performance in character-level infilling tasks during the inference stage.Alternately, some approaches considered character-level infilling, but they relied on predicting sub-tokens in inference, yet this strategy diminished ability in character-level infilling tasks due to the large perplexity of the model on sub-tokens.In this paper, we introduce FIM-SE, which stands for Fill-In-the-Middle with both Starting and Ending character constraints.The proposed method addresses character-level infilling tasks by utilizing a line-level format to avoid predicting any sub-token in inference.In addition, we incorporate two special tokens to signify the rest of the incomplete lines, thereby enhancing generation guidance.Extensive experiments demonstrate that our proposed approach surpasses previous methods, offering a significant advantage.Code is available at https://github.com

Vehicle detection is a challenging task in computer vision. Since the differences among different vehicle shapes are small and not easy to recognize, the common object detection models usually do not work well for vehicle recognition. To select positive samples, the existing anchor-based detection methods usually use high-quality anchor boxes which are very close to the shapes of the vehicles. Such methods often make the prediction strongly dependent on high-quality anchor boxes. At present, the ATSS (Adaptive Training Sample Selection) method has also been used to select high-quality anchor boxes. The main purpose of this method is to automatically set the shapes of the anchor boxes according to the training datasets. However, for vehicle detection, the shapes of the vehicles are usually similar while the scales of the vehicles are quite different, which may influence the performance of the ATSS based vehicle detection. To solve the above problems, we propose a vehicle detection model PSLTNet (Positives Selection with Low Threshold Network). To reduce the prediction instability due to relying on high-quality anchor boxes, PSLTNet applies the positive samples selection with low threshold to enhance the prediction ability for different quality anchor boxes. To improve the identification ability of different vehicle types, PSLTNet adopts a pyramid structure and cascades dilated convolutions at each level. In addition, in order to make the bounding box regression loss be independent of scales, PSLTNet uses GIOU (Generalized Intersection Over Union) to supervise the position regression task of candidate bounding boxes. Finally, to reduce the loss of a large number of simple negatives, PSLTNet uses Focal Loss to supervise the classification task of the background candidate bounding boxes. The Experimental results upon the BIT-Vehicle dataset show that the proposed method can obtain better performance than that of the existing methods.

Abstract Pixel‐wise classification of hyperspectral image (HSI) is a hot spot in the field of remote sensing. The classification of HSI requires the model to be more sensitive to dense features, which is quite different from the modelling requirements of traditional image classification tasks. Cycle‐Multilayer Perceptron (MLP) has achieved satisfactory results in dense feature prediction tasks because it is an expert in extracting high‐resolution features. In order to obtain a more stable receptive field and enhance the effect of feature extraction in multiple directions, we propose an MLP‐like model called DriftNet for HSI classification inspired by Cycle‐MLP and deformable convolution. Besides, the proposed DriftNet uses a unique ladder‐like fully connected structure to achieve progressive activation of neurons and facilitates the fusion of spatial and spectral information, thereby obtaining more sensitive location information for better classification results. Experimental results on three public hyperspectral datasets demonstrate the effectiveness and generalisation of DriftNet.

The existing approaches for crowd counting usually estimate a density map with deep convolutional neural network to obtain the crowd counts. Influenced by the background noises, some approaches may result in incorrect pedestrian heads recognition. Therefore, some approaches try to estimate an attention map to mask background noises. However, since the background noises are complex and stochastic, single attention is of incompetence to recognize them. Consequently, we proposed softer, and more reasonable Cascade Residual Attention Network (CRANet), which cascades several effective residual attention modules to mask background noises. Also, due to pixel-level isolation of Euclidean loss, we designed a novel Pyramid Structural Similarity Loss to train our CRANet. The proposed approach was evaluated on three crowd datasets. Experimental results demonstrated that our approach achieves the state-of-the-art.

ThispaperpresentsBigDL (adistributeddeeplearning framework for Apache Spark), which has been used by a variety of users in the industry for building deep learning applications on production big data platforms. It allows deep learning applications to run on the Apache Hadoop/Spark cluster so as to directly process the production data, and as a part of the end-to-end data analysis pipeline for deployment and management. Unlike existing deep learning frameworks, BigDL implements distributed, data parallel training directly on top of the functional compute model (with copy-on-write and coarse-grained operations) of Spark. We also share real-world experience and "war stories" of users that havead-optedBigDLtoaddresstheirchallenges(i.e., howtoeasilybuildend-to-enddataanalysisanddeep learning pipelines for their production data).