This research proposed an end-to-end object detection network based on Kolmogorov–Arnold Networks (KANs)-Detection Transformer (DETR). KANs block was introduced into encoder–decoder structure instead of the full connection layer to dynamically learn the activation function and improve the robustness and accuracy of the model. Experiments showed that the detection capability of KANs-DETR on multicategory object detection was better than that of HGNetv2 and Swin Transformer as backbone. Furthermore, in order to solve the problem of insensitivity to small objects, the Squeeze-and-Excitation module was applied for feature fusion and presented better performance. The KANs-DETR achieved high detection accuracy and efficiency in handling small objects in complex scenes, providing a new perspective for network optimization.

Federated learning aims to collaboratively train a shared global model from multiple clients while preserving data privacy. However, real-world applications often involve clients learning from limited and dynamically arriving data, requiring the global model to classify all encountered classes. This paper introduces federated few-shot class-incremental learning, enabling effective learning of new classes from scarce samples within a decentralized framework. Existing methods suffer from new class bias, where new classes are often misclassified as previously learned ones. Additionally, they face local bias due to non-IID data distribution, which leads client models to focus excessively on their specific local data characteristics. We propose a Decoupled Multi-Prototype Collaboration (DMPC) method to mitigate both biases. First, we introduce a Global Consistency Aggregation mechanism (GCA) that re-weights local prototypes based on their consistency, resulting in more representative global prototypes and effectively eliminating local bias. Second, we design a Multi-Prototype Testing strategy (MPT) that enhances classification accuracy by leveraging both local and global prototypes, thereby mitigating new class bias. More importantly, GCA and MPT exhibit significant synergistic effects. Extensive experiments on three widely used datasets demonstrate the robustness and superiority of our method in bias reduction.

Skeleton-based action recognition technology has gained significant attention and made great progress in recent years. However, the performance of existing methods declines significantly when the quality of skeleton data extracted by pose estimation algorithms varies. To address this issue, this study proposes an instance-specific learning method aimed at enhancing the model’s ability to learn discriminative features when handling skeleton data of varying quality. We introduce a Dynamic Instance Discriminability Assessment (DIDA) mechanism and a Staged Instance Weighting (SIW) strategy. The DIDA mechanism dynamically evaluates the discriminability of instances by combining prior knowledge with feedback from the model during the training process. The SIW strategy adjusts the weights of instances at different training stages based on their discriminability. Notably, our method requires only a minimal increase in computational cost during training and incurs no additional computational overhead during testing compared to baseline models. We utilized Pifpaf and HR-Net pose estimation methods to extract skeleton data of varying quality from the NTU60, NTU120, and HMDB51 video datasets and conducted extensive experimental validation. The results indicate that the proposed method significantly enhances the action recognition performance while maintaining computational efficiency.

Vulnerability root cause analysis (RCA) is a crucial step following the discovery of vulnerabilities. When faced with a multitude of crashes resulting from fuzzing, effective RCA results can assist developers in swiftly identifying and rectifying the root causes of vulnerabilities. Recently, some methods that rely on statistical behavioral differences to analyze the root causes of vulnerabilities have been introduced. However, they suffer from issues such as high time costs, strong randomness, and imprecise results, rendering them impractical for real-world applications. In this paper, we propose an efficient and accurate statistical analysis-based vulnerability RCA approach named RCLocator. We introduce an enhanced crash information tuple extraction tool based on sanitizer to ensure crash consistency during the mutation process of original files. This approach reduces the time cost of the data augmentation stage and enhances the accuracy of RCA. Furthermore, it provides developers with effective explanations for root cause predicates. We evaluate our approach on RCABench and real-world vulnerabilities. The results indicate that RCLocator significantly outperforms state-of-the-art methods, the probability of obtaining correct root cause analysis results increased by 46.7%, and 9.0 times faster in terms of time.

Binary Code Similarity Detection (BCSD), which calculates the similarity between binary code snippets, plays a vital role in various security fields. Since binary functions have complete semantics, the main research objects in BCSD are binary functions. Current approaches face challenges in effectively capturing the semantic information of assembly instructions and the structural information of control flow graphs (CFG) in binary functions. This paper proposes a graph attention network (GAT) for BCSD, called BSGAT, to detect similarity between binary functions. Our contribution is twofold: first, we propose a strategy to generate rich representations of basic blocks in CFG; second, we introduce GAT, which assigns different weights to basic blocks in CFG, enabling the generation of more discriminative embeddings for target binary functions. We conduct experiments on a binary function similarity detection task and a real vulnerability detection task. The results show that our proposed model BSGAT outperforms existing models in both tasks. In the binary function similarity detection task, BSGAT achieved the highest average AUC value of 0.872. In the real vulnerability detection task, BSGAT achieves the highest average recall@10 0.378, surpassing the best-performing model Gemini (0.337) in the comparison models, with a significant improvement of 12.2%. Our code is available at https://github.com/quchao777/BSGAT.git.

In this paper, an improved Transformer network is proposed, which reduces the overall computation of the network by nearly 50°/o while still maintaining good network performance while only retaining K and V values. Experiments show that with Swin Transformer as the backbone network, the improved method proposed in this paper can reduce the training time and testing time while maintaining high accuracy.

In this paper, we propose a lightweight mask detection algorithm and implement an intelligent vehicle system. The algorithm uses YOLOv5s as the backbone network, and at the same time incorporates the SE attention mechanism to optimize the timeliness, and is finally deployed on an intelligent vehicle system with BCM2711 as the control platform. Experiments prove that the algorithm proposed in this paper reduces the detection time by 30% while ensuring a higher MAP, which has certain value for promotion.

This paper addresses the problem of skeleton-based action prediction, which aims to predict the action label when the skeleton sequence is partially observed. The action prediction task is more challenging compared to the after-the-fact action recognition since it needs to make decisions according to the beginning part of action executions. The existing methods improve the action prediction performance by taking advantage of the global action knowledge in full sequences, and some of them require the correspondence between a partial sequence and its associated full sequence. In this paper, we step towards a new direction by exploiting the discriminative information in early observations of actions as much as possible. We propose a Graph Convolutional Network with Early Attention Module (GCN-EAM), which employs a series of spatial-temporal graph convolution blocks to extract features from skeletons. In order to infer the action category as fast as possible, we introduce an early attention module to adaptively emphasize discriminative observations at the beginning stage of actions. The proposed method is evaluated on the large-scale NTU-RGB+D dataset and achieves excellent performance for action prediction.

This letter addresses the challenging problem of action prediction with partially observed sequences of skeletons. Towards this goal, we propose a novel two-stage knowledge distillation framework, which transfers prior knowledge to assist the early prediction of ongoing actions. In the first stage, the action prediction model (also referred to as the student) learns from a couple of teachers to adaptively distill action knowledge at different progress levels for partial sequences. Then the learned student acts as a teacher in the next stage, with the objective of optimizing a better action prediction model in a self-training manner. We design an adaptive self-training strategy from the perspective of undermining the supervision from the annotated labels, since this hard supervision is actually too strict for partial sequences without enough discriminative information. Finally, the action prediction models trained in the two stages jointly constitute a two-stream architecture for action prediction. Extensive experiments on the large-scale NTU RGB+D dataset validate the effectiveness of the proposed method.

This paper aims to address the problem of fall detection from RGB-D image sequences. Towards this goal, we propose a novel Key Point Trajectory Model which represents a fall action as a series of trajectory descriptors. In the proposed model, 16 key points including 14 skeleton points and 2 centers of body parts are extracted from each pair of RGB and depth images. Then a global trajectory descriptor is constructed on 16 trajectories that are obtained by connecting the key points across several frames in the RGB-D sequence. The trajectory descriptor incorporates the spatial, depth, and temporal context of key points and characterizes the global motion of human over a short period of time. A random forest is employed to learn the classifier of trajectory descriptors, and an integration rule is developed for detecting falls according to the classification results of all trajectory descriptors within a video. Experiments conducted on two fall detection datasets demonstrate that our method achieves better performance in comparison with state-of-the-art methods.

A lightweight intelligent vehicle system was developed to realize autonomous driving, face recognition, face anti-spoofing, remote control, infrared obstacle avoidance and other functions to improve the security of contactless delivery. In this system, BCM2711 was used as kernel control chip, and it was equipped with deep network learning models such as LaneNet, ResNet and LSTM. It had been proved that this system could realize the above functions and achieve real-time effects, thus gaining great economic value and market space in contactless delivery service.

An improved NS traffic flow model was built in this paper to simulate two safety factors of vehicle-pedestrian avoidance and vehicle-vehicle avoidance under different weather conditions. Then the regulations of changes on lanes and vehicle speed under two-lane conditions were optimized as well as the improved NS model based on cellular automata. Results showed that the improved NS model can predict road conditions effectively, thereby improving the safety of roads.