Large Language Models (LLMs) have demonstrated remarkable performance in code generation, offering new possibilities for translating natural language into executable programs. To further enhance LLMs’ code generation capabilities, Retrieval-Augmented Generation (RAG) has emerged as a promising strategy by retrieving code examples aligned with the generation intent to guide the process. However, existing RAG-based methods often suffer from unnecessary augmentation, preference misalignment, and surface-level mimicry, which undermine the effectiveness of retrieved examples in guiding LLMs toward accurate code generation. To address these challenges, we propose SRACG, a Selective Retrieval-Augmented Code Generation framework. SRACG begins with a necessity-aware selection mechanism to identify generation intents that genuinely require retrieval support, thereby avoiding degradation from indiscriminate augmentation. For intents identified as needing enhancement, it first employs a multi-objective retrieval strategy to select examples that are semantically aligned with the intent. These candidates are then further filtered by assessing their consistency with the LLM’s inherent generation preferences, ensuring alignment in both style and structure. Finally, it extracts execution plans from the filtered examples to uncover their underlying logic, guiding the LLM to better comprehend the examples instead of merely mimicking surface-level content. Experimental results on widely used benchmarks show that SRACG significantly improves the success rate of LLM-generated code and outperforms existing approaches.

Repository-level code generation aims to generate target code conditioned on the context of the specified repository. Existing approaches generally adopt the Retrieval-Augmented Generation (RAG) framework to retrieve relevant contextual information from the repository, thereby mitigating the challenge of long context windows. However, these methods typically struggle to identify truly relevant information for the current generation requirement, and they generally overlook the imperative need for understanding the repository-level code structure during the retrieval process. To address these issues, we propose RepoMind, a novel framework that advances repository-level code generation by leveraging a repository documentation library and LLM reasoning capabilities for API retrieval. Specifically, RepoMind first introduces the RepoDocs Agent. This component constructs multi-granularity, hierarchical structural documentation in a bottom-up manner, providing LLMs with a high-quality knowledge base for understanding repository functionality and structure. Built upon this hierarchical knowledge base, RepoMind further integrates the Reasoning-Retrieval Agent. This agent mimics human developer patterns by utilizing the repository documentation for layer-by-layer exploration, and ultimately precisely localizes the relevant API sets. Finally, the framework combines the functionally relevant API set retrieved via LLM reasoning with the semantically similar API set retrieved by the vector retriever, utilizing this hybrid context to boost code generation performance. Evaluation on widely-used repository-level code generation benchmarks, CoderEval and DevEval, demonstrates that RepoMind surpasses state-of-the-art methods, achieving up to a 13.50% relative improvement in Pass@1 scores. The open-source code will be released at https://github.com/ABigTrouble/RepoMind.

Code search aims to quickly locate target code snippets from databases using natural language queries, which promotes code reusability. Existing methods can effectively obtain aligned token-level and query word-level features. However, these studies usually represent the semantics of code and query by averaging the features of each token and word respectively, which makes it difficult to accurately capture the code details that are closely related to the query. To address this issue, we propose a fine-grained code search model that consists of a cross-modal encoder, a mapping layer, and a classification layer. Specifically, we utilize a pre-trained model, GraphCodeBERT, in the cross-modal encoder to align features. In the mapping layer, we introduce a co-attention network to capture the fine-grained interactions between code and query, ensuring a model can precisely identify key code segments relevant to the query. Finally, in the classification layer, we incorporate instruction learning techniques that leverage contextual reasoning to improve the accuracy of query-code matching. Experimental results show that our proposed model significantly outperforms existing methods across multiple programming language datasets.

Model Diagram-to-Code Generation aims to translate model diagrams from research papers into implementation code that reconstructs the model's architecture. This task plays a crucial role in accelerating scientific workflows and enhancing the efficiency of industrial model deployment. While recent studies have explored various Image-to-Code Generation tasks using Multimodal Large Language Models (MLLMs), these efforts have primarily focused on reconstructing the visual appearance depicted in input images, leaving this task largely underexplored. The complex structural elements and implicit relationships in model diagrams present greater challenges for MLLMs, particularly in terms of visual reasoning and semantic interpretation. To support this task, we introduce MDCDataset, a dataset designed to evaluate the ability of MLLMs to generate code from model diagrams. It comprises 1,008 instances spanning 16 research domains, each with a model diagram, structured textual content, and the ground-truth code implementation. Furthermore, to address the inherent challenges of this task, we propose MDCAgent, a collaborative multi-agent framework composed of Parsing, Generation, and Check Agents. These agents work in coordination to analyze, extract, and verify complex elements and implicit relationships within model diagrams, thereby enhancing the visual architecture-aware reasoning capabilities of MLLMs. Our extensive experiments confirm the effectiveness of the framework.

Renal cell carcinoma (RCC) is the most common type of kidney cancer in adults, its early detection and clinical diagnosis are crucial for reducing mortality. Deep convolutional neural networks (DCNNs) have demonstrated outstanding performance in medical image classification. Therefore, we selected the pre-trained EfficientNet-B2 as the backbone architecture for its lightweight and efficient characteristics. However, due to the implementation of privacy protection policies, acquiring sensitive medical data such as those related to kidney cancer has become increasingly challenging. Federated learning (FL) safeguards patient privacy by sharing only model parameters instead of raw data. Nevertheless, medical image data often exhibits non-independent and identically distributed (non-IID) characteristics, data sparsity, and imbalanced distributions, which adversely affect model performance. To address these issues, we propose an optimized federated learning framework, FedMilNet, which evaluates the data quality of each client and then utilizes dynamic curriculum learning. This approach arranges the sequence of client participation from easy to difficult based on data quality, gradually expanding the candidate frontier. Additionally, we determine “commanders” and “auxiliary teachers” based on the quality score, dynamically adjusting their weights as training progresses. FedMilNet achieved an accuracy of 62.71% on the ccRCC dataset, outperforming mainstream federated baselines by approximately 10%. On the Kaggle Kidney Cancer and CT KIDNEY DATASET, it reached high accuracies of 99.83% and 95.78%. By quantifying and utilizing client data quality in federated environments, and integrating curriculum-based client selection with dynamic knowledge distillation, our approach significantly enhances model generalization and accuracy under small-sample and non-IID conditions, advancing scientific progress in privacy-preserving and practical intelligent diagnosis of kidney cancer.

Text clustering, as one of the most fundamental challenges in unsupervised learning, aims at grouping semantically similar text segments without relying on human annotations. With the rapid development of deep learning, deep clustering has achieved significant advantages over traditional clustering methods. Despite the effectiveness, most existing deep text clustering methods rely heavily on representations pre-trained in general domains, which may not be the most suitable solution for clustering in specific target domains. To address this issue, we propose CEIL, a novel Classification-Enhanced Iterative Learning framework for short text clustering, which aims at generally promoting the clustering performance by introducing a classification objective to iteratively improve feature representations. In each iteration, we first adopt a language model to retrieve the initial text representations, from which the clustering results are collected using our proposed Category Disentangled Contrastive Clustering (CDCC) algorithm. After strict data filtering and aggregation processes, samples with clean category labels are retrieved, which serve as supervision information to update the language model with the classification objective via a prompt learning approach. Finally, the updated language model with improved representation ability is used to enhance clustering in the next iteration. Extensive experiments demonstrate that the CEIL framework significantly improves the clustering performance over iterations, and is generally effective on various clustering algorithms. Moreover, by incorporating CEIL on CDCC, we achieve the state-of-the-art clustering performance on a wide range of short text clustering benchmarks outperforming other strong baseline methods.

BACKGROUND: Medication recommendation based on electronic medical record (EMR) is a research hot spot in smart healthcare. For developing computational medication recommendation methods based on EMR, an important challenge is the lack of a large number of longitudinal EMR data with time correlation. Faced with this challenge, this paper proposes a new EMR-based medication recommendation model called MR-KPA, which combines knowledge-enhanced pre-training with the deep adversarial network to improve medication recommendation from both feature representation and the fine-tuning process. Firstly, a knowledge-enhanced pre-training visit model is proposed to realize domain knowledge-based external feature fusion and pre-training-based internal feature mining for improving the feature representation. Secondly, a medication recommendation model based on the deep adversarial network is developed to optimize the fine-tuning process of pre-training visit model and alleviate over-fitting of model caused by the task gap between pre-training and recommendation. RESULT: The experimental results on EMRs from medical and health institutions in Hainan Province, China show that the proposed MR-KPA model can effectively improve the accuracy of medication recommendation on small-scale longitudinal EMR data compared with existing representative methods. CONCLUSION: The advantages of the proposed MR-KPA are mainly attributed to knowledge enhancement based on ontology embedding, the pre-training visit model and adversarial training. Each of these three optimizations is very effective for improving the capability of medication recommendation on small-scale longitudinal EMR data, and the pre-training visit model has the most significant improvement effect. These three optimizations are also complementary, and their integration makes the proposed MR-KPA model achieve the best recommendation effect.