Shiting Wen

Code-to-code translation aims to automatically convert a program written in one programming language into a semantically equivalent, executable program in another language, and is a key enabling technology for legacy-system modernization, cross-platform migration and software supply-chain security. The field has evolved from handwritten rules and statistical machine translation, through pre-trained model fine-tuning and unsupervised back-translation, to LLM prompting, execution-feedback repair and multi-agent collaboration. This survey reviews work from 2013 to 2026. Using a literature search and selection process reported according to PRISMA 2020, we identify 47 included method entries and organize them in a two-level taxonomy of three families and ten subclasses. We compare these entries under an explicit coding scheme, summarize CodeXGLUE results together with their comparability limits, and organize datasets, benchmarks, and multilevel evaluation signals. Using four review questions on datasets, evaluation, open gaps, and trade-offs as a framework, we argue that the central challenge of the LLM era is practical reliability: whether a translation can be compiled, tested, reproduced, and integrated, rather than merely resemble a reference. We accordingly propose complementary evaluation criteria and discuss executable repository-level evaluation, transparent error taxonomies, security, licensing, and compliance risks, and low-resource, ecosystem-mismatched language pairs.

To address growing concerns about data privacy on mobile devices, the federated learning (FL) paradigm enables clients to collaboratively train models while sharing only local model updates. However, privacy risks remain in FL, as adversaries can still infer sensitive information from these updates. To enhance secure aggregation in FL, various protection mechanisms combining encryption and multi-party computation (MPC) have been proposed. These approaches, however, often introduce substantial communication and computational overhead, making secure aggregation impractical on resource-constrained devices, e.g., smart phones. To tackle these efficiency challenges, we are among the first to propose the integration of differential privacy (DP) with encryption and MPC for secure aggregation. Our proposed protocol, Federated Learning with Noise-based Secure Aggregation (FedNSA), injects noise through DP to obfuscate individual model updates. Encryption is employed to correlate the noise across different clients, while MPC ensures perfect noise cancellation at the server side. Finally, we theoretically analyze its advantages and conduct extensive experiments on public datasets to demonstrate the superiority of our approach across multiple dimensions in comparison with the state-of-the-art baselines.

Federated learning (FL) is revolutionizing machine learning by enabling multiple decentralized clients to collaboratively train a shared model. In mobile scenarios, client devices exchange model parameters with a central server via wireless channels. However, designing efficient wireless FL (WFL) is challenging due to limited energy and channel capacity. To fully utilize communication resources, over-the-air (OTA) computation has been introduced, allowing direct aggregation of analog parameter signals. However, it conceals clients’ information from the server, making advanced client selection strategies, e.g., cluster-based client selection, and sparsification compression algorithms like TopK inapplicable. To address these limitations, we propose the WFL with Voting-based Clustering (WFL-VC) algorithm, which can integrate advanced client selection and the TopK model sparsification algorithm with OTA computation. WFL-VC consists of two phases: 1) Phase 1: clients vote on significant parameters based on local models, allowing the server to select clients and identify the global Topk parameters; and 2) Phase 2: the selected clients upload model update parameters with globally aligned indices for over-the-air computation at the server. By combining OTA computation with cluster-based client selection and TopK sparsification, WFL-VC substantially reduces the energy consumption of WFL. Extensive experiments on real-world datasets show that WFL-VC outperforms competitive baselines while consuming considerably less energy.

Unmanned aerial vehicle (UAV) remote sensing technology has been widely applied in pine wilt disease (PWD) detection. UAV-collected PWD images are characterized by complex backgrounds, small target sizes, and uneven spatial distribution. State-of-the-art generic detection methods typically use Transformers or Mamba-based backbones to mitigate background interference by enhancing the global receptive field. However, directly applying these generic detectors to detect PWD results in suboptimal performance due to insufficient utilization of local features. Although current PWD detection methods use the attention mechanisms to improve the recognition of infected targets, the limitations of Convolutional Neural Networks (CNNs) in capturing long-range dependencies hinder their ability to separate targets from the background. To overcome these challenges, a novel PWD detection method called feature interaction and dependency capturing-YOLO (FIDCYOLO) is proposed in this article. Firstly, to effectively extract the discriminative features of PWD targets, the shuffle efficient layer aggregation network (SELAN) is proposed to promote information interaction between features, improving the model's learning capability. Then, to address the interference of complex backgrounds, the bidirectional extension of Mamba (Hydra) is improved by employing the partial dependency capture (PDC) module, which captures both global and local dependencies for suppressing background information. Additionally, the feature aggregation-spatial pyramid pooling fast (FA-SPPF) module is introduced in front of PDC, which aggregates cross-level features to enhance detection performance. Finally, we use ADown as the downsampling module to avoid the loss of target information. Extensive experiments on our self-constructed PWD dataset and the public DIOR dataset demonstrate the excellent performance of FIDC-YOLO.

Proof of Authority (PoA) plays a pivotal role in blockchains for reaching consensus. Clique, which selects consensus nodes to generate blocks with a pre-determined order, is the most popular implementation of PoA due to its low communication overhead and energy consumption. However, our study unveils that the speed to process transactions by Clique is severely restricted by 1) the long communication delay of full blocks (each containing a certain number of transactions) between consensus nodes; and 2) occurrences of no-turn blocks, generated by no-turn nodes if an in-turn block generation fails. Consequently, Clique struggles to support distributed applications requiring a high transaction processing speed, e.g., online gaming. To overcome this deficiency, we propose an Express Clique (ExClique) algorithm by improving Clique from two perspectives: compacting blocks for broadcasting to shorten communication delay and prohibiting the occurrences of no-turn blocks. For performance evaluation, we implement ExClique by modifying Geth of Ethereum, the software implementing Clique, and deploy a permissioned blockchain network by using container technology. The experimental results show that ExClique achieves a substantial enhancement in transactions per second (TPS). Specifically, it boosts TPS by 2.25X in a typical network with 21 consensus nodes and an impressive 7.01X in a large-scale network with 101 consensus nodes when compared to Clique.