Bugs in compilers, which are critical infrastructure today, can have outsized negative impacts. Mutational fuzzers aid compiler bug detection by systematically mutating compiler inputs, i.e., programs. Their effectiveness depends on the quality of the mutators used. Yet, no prior work used compiler bug histories as a source of mutators.

With the rapid development and application of deep learning technologies, the performance of image denoising has been significantly enhanced. In this paper, we propose a novel network architecture based on the boosting strategy to improve the image recovery quality. Firstly, we incorporate the advantages of traditional iterative methods by introducing the plug-and-play prior technology, originally applied to image restoration. By transforming the denoising task into an optimization problem and solving it using the half-quadratic splitting algorithm, we obtain the recursive function for the boosting strategy. Subsequently, this iterative step is unfolded into a deep learning network, thereby enhancing the performance of the original denoising module through a recurrent network structure. Experimental results show that by modifying image denoising algorithms based on supervised and self-supervised deep learning, the improved network can effectively enhance the performance of the original algorithms, demonstrating excellent performance in both additive white Gaussian noise (AWGN) and real noise denoising.

Inference scaling empowers LLMs with unprecedented reasoning ability, with reinforcement learning as the core technique to elicit complex reasoning. However, key technical details of state-of-the-art reasoning LLMs are concealed (such as in OpenAI o1 blog and DeepSeek R1 technical report), thus the community still struggles to reproduce their RL training results. We propose the **D**ecoupled Clip and **D**ynamic s**A**mpling **P**olicy **O**ptimization (**DAPO**) algorithm, and fully open-source a state-of-the-art large-scale RL system that achieves 50 points on AIME 2024 using Qwen2.5-32B base model. Unlike previous works that withhold training details, we introduce four key techniques of our algorithm that make large-scale LLM RL a success. In addition, we open-source our training code, which is built on the verl framework, along with a carefully curated and processed dataset. These components of our open-source system enhance reproducibility and support future research in large-scale LLM RL.

Mixture-of-Experts (MoE) showcases tremendous potential to scale large language models (LLMs) with enhanced performance and reduced computational complexity. However, its sparsely activated architecture shifts feed-forward networks (FFNs) from being compute-intensive to memory-intensive during inference, leading to substantially lower GPU utilization and increased operational costs.

Abstract Pedestrian re‐identification (re‐ID) is an important research direction in computer vision, with extensive applications in pattern recognition and monitoring systems. Due to uneven data distribution, and the need to solve clustering standards and similarity evaluation problems, the performance of unsupervised methods is limited. To address these issues, an improved unsupervised re‐ID method, called Enhanced Feature Representation and Robust Clustering (EFRRC), which combines EFRRC is proposed. First, a relation network that considers the relations between each part of the pedestrian's body and other parts is introduced, thereby obtaining more discriminative feature representations. The network makes the feature at the single‐part level also contain partial information of other body parts, making it more discriminative. A global contrastive pooling (GCP) module is introduced to obtain the global features of the image. Second, a dispersion‐based clustering method, which can effectively evaluate the quality of clustering and discover potential patterns in the data is designed. This approach considers a wider context of sample‐level pairwise relationships for robust cluster affinity assessment. It effectively addresses challenges posed by imbalanced data distributions in complex situations. The above structures are connected through a clustering contrastive learning framework, which not only improves the discriminative power of features and the accuracy of clustering, but also solves the problem of inconsistent clustering updates. Experimental results on three public datasets demonstrate the superiority of our method over existing unsupervised re‐ID methods.

Programmable Logic Controllers (PLCs) are the core unit of the production system, which frequently need to implement new processes to address customer needs. These changes must be fully tested to ensure the reliability of the PLC code, which is commonly programmed through Functional Block Diagrams (FBDs). This is a tedious task that requires considerable time and effort given the manual nature of the process involved in PLC testing. Hence, we present a cost-effective test selection approach to test FBD programs in dynamic environments. The proposed method uses a search-based multi-objective test case selection algorithm as a regression technique to test recently modified FBD programs. Specifically, we derived a total of 7 fitness function combinations, by combining different cost and quality-based fitness functions. We carried out an empirical evaluation, by employing fitness metrics in the wellknown NSGA-II algorithm to determine the best configuration setup for testing FBD programs. Furthermore, we benchmarked the performance of the NSGA-II with the baseline Random Search (RS). The study was carried out with three case studies of a reactor protection system, and evaluated with two sets of mutants. The results demonstrated that the proposed approach significantly reduces time, while keeping high the overall fault detection capability.

Intelligent Vehicle Systems (IVSs) devote to integrating the data sensing, processing, and transmission in the Vehicle to Everything (V2X) scenarios, where the Unnamed Aircraft Vehicle (UAV)-aided traffic monitoring network is one of the most significant applications. Moreover, since the central premise to support the IVS is timely and effectively sensing data processing, Age of Information (AoI) can precisely reflect the timeliness and effectiveness of the communication process in the UAV-aided traffic monitoring network. However, recent researches pay little attention to AoI minimization issue, especially when the malicious attacker attempts to deteriorate the network performance. The accurately modelling of the adversarial relationship between legitimate UAVs and attacker is not fully investigated. To make up this research gap, we start from the Stackelberg game viewpoint to investigate the AoI optimization problem in the UAV-aided traffic monitoring network under attack. Firstly, the system model and three-layer Stackelberg game-based optimization goal are established. Secondly, based on the Backward Induction (BI) analysis, the follower’s data sensing rate, transmission power, and the leader’s attacking power are determined by the Lagrange duality optimization technology successively. Moreover, the sub-gradient update-based optimization technology is used to achieve the Stackelberg Equilibrium (SE). Finally, simulations are performed under various parameters. The evaluation results present better performance of our proposed approach when compared with the typical baselines.

Efficient grant-free uplink transmission is critical in minimizing Age of Information (AoI) in multichannel Internet of Things (IoT) networks. But less attention has been paid to this topic especially when dynamic channel access attacks (DCAAs) exist. To bridge this gap, this article formulates the distributed channel access problem in AoI-oriented IoT networks, and then a reinforcement learning-based solution is put forward based on the theoretical results of the game theory. First, a utility maximization problem is formulated for each sensor node based on its average AoI under DCAAs with probabilistic ACK feedback. Second, the problem is transformed into two ordinary potential game (OPG) models, which are both proved to have at least one nash equilibrium (NE); and a distributed learning algorithm is proposed to reach the NE. Finally, extensive simulations are conducted to evaluate the proposal’s performance. Simulation results verify the effectiveness of the proposed algorithm in various parameters settings.

Boosting algorithms have demonstrated their effectiveness in improving the restoration quality of existing image denoising methods by extracting the residual signal or removing the noise leftover iteratively. Unlike existing boosting algorithms that focus on designing an ingenious recursive step by making use of the residual signal or the noise leftover, in this paper, we propose a novel model-guided boosting framework. Specifically, we derive the recursive step from an overall restoration model constructed with the technique of Regularization by Denoising (RED) towards an interpretable, extensible and flexible boosting mechanism. By using the RED, we can apply explicit regularization equipped with powerful image denoising engine to establish the global minimization problem , making the obtained model is clearly defined and well optimized. The framework enjoys the advantage of easily extending to the case of composite denoising via superadding a regularization term. As such, we develop a simultaneous model through the joint use of deep neural network and low-rank regularization to fully utilize both external and internal image properties. The resulting restoration models are capable of being flexibly solved with fixed-point strategy and steepest-descent method, leading to two types of denoising boosters. It is shown that the proposed schemes have promise results due to the improvement in signal-to-noise ratio of input signal, and are guaranteed to converge. Experiments verify the validity of the boosters for several denoising algorithms, and show that combining the power of internal and external denoising based on our framework achieves enhancement in denoising performance.

Summary Function block diagram (FBD) is a standard programming language for programmable logic controllers (PLCs). PLCs have been widely used to develop safety‐critical systems such as nuclear reactor protection systems. It is crucial to test FBD programs for such systems effectively. This paper presents an automated test sequence generation approach using mutation testing techniques for FBD programs and the developed tool, MuFBDTester. Given an FBD program, MuFBDTester analyses the program and generates mutated programs based on mutation operators. MuFBDTester translates the given program and mutants into the input language of a satisfiability modulo theories (SMT) solver to derive a set of test sequences. The primary objective is to find the test data that can distinguish between the results of the given program and mutants. We conducted experiments with several examples including real industrial cases to evaluate the effectiveness and efficiency of our approach. With the control of test size, the results indicated that the mutation‐based test suites were statistically more effective at revealing artificial faults than structural coverage‐based test suites. Furthermore, the mutation‐based test suites detected more reproduced faults, found in industrial programs, than structural coverage‐based test suites. Compared to structural coverage‐based test generation time, the time required by MuFBDTester to generate one test sequence from industrial programs is approximately 1.3 times longer; however, it is considered to be worth paying the price for high effectiveness. Using MuFBDTester, the manual effort of creating test suites was significantly reduced from days to minutes due to automated test generation. MuFBDTester can provide highly effective test suites for FBD engineers.

Circular RNA (circRNA) is a group of RNA family generated by RNA circularization, which was discovered ubiquitously across different species and tissues. However, there is no global view of tissue specificity for circRNAs to date. Here we performed the comprehensive analysis to characterize the features of human and mouse tissue-specific (TS) circRNAs. We identified in total 302 853 TS circRNAs in the human and mouse genome, and showed that the brain has the highest abundance of TS circRNAs. We further confirmed the existence of circRNAs by reverse transcription polymerase chain reaction (RT-PCR). We also characterized the genomic location and conservation of these TS circRNAs and showed that the majority of TS circRNAs are generated from exonic regions. To further understand the potential functions of TS circRNAs, we identified microRNAs and RNA binding protein, which might bind to TS circRNAs. This process suggested their involvement in development and organ differentiation. Finally, we constructed an integrated database TSCD (Tissue-Specific CircRNA Database: http://gb.whu.edu.cn/TSCD) to deposit the features of TS circRNAs. This study is the first comprehensive view of TS circRNAs in human and mouse, which shed light on circRNA functions in organ development and disorders.