Code summarization (CS) and code generation (CG) are two crucial tasks in the field of automatic software development. Various neural network-based approaches proposed to solve these separately. However, there exists a specific intuitive correlation between CS CG, which have not been exploited previous work. In this paper, we apply relations improve performance both tasks. other words, exploiting duality tasks, propose dual training framework train simultaneously. framework, consider...

Parkinson's disease (PD) is a prevalent and challenging neurodegenerative disorder, may involve impaired autophagy. Nuclear factor erythroid-2-related 2 (Nrf2) crucial for regulating autophagy-related genes, maintaining cellular homeostasis. Electroacupuncture (EA), complementary alternative therapy PD, has gained widespread clinical application. In this study, we investigate whether EA at Baihui (GV20) Taichong (LR3) acupoints modulates autophagy through the Nrf2 pathway, providing...

Code completion, one of the most useful features in Integrated Development Environments (IDEs), can accelerate software development by suggesting libraries, APIs, and method names real-time. Recent studies have shown that statistical language models improve performance code completion tools through learning from large-scale repositories. However, these suffer three major drawbacks: a) The hierarchical structural information programs is not fully utilized program's representation; b) In...

Tables are widely used with various structures to organize and present data. Recent attempts on table understanding mainly focus relational tables, yet overlook other common structures. In this paper, we propose TUTA, a unified pre-training architecture for generally structured tables. Noticing that requires spatial, hierarchical, semantic information, enhance transformers three novel structure-aware mechanisms. First, devise tree-based structure, called bi-dimensional coordinate tree,...

In recent years, larger and deeper models are springing up continuously pushing state-of-the-art (SOTA) results across various fields like natural language processing (NLP) computer vision (CV). However, despite promising results, it needs to be noted that the computations required by SOTA have been increased at an exponential rate. Massive not only a surprisingly large carbon footprint but also negative effects on research inclusiveness deployment real-world applications. Green deep...

Deep learning (DL) has recently been widely applied to diverse source code processing tasks in the software engineering (SE) community, which achieves competitive performance (e.g., accuracy). However, robustness, requires model produce consistent decisions given minorly perturbed inputs, still lacks systematic investigation as an important quality indicator. This article initiates early step and proposes a framework CARROT for robustness detection, measurement, enhancement of DL models...

Autophagy is confirmed to be involved in the onset and development of depression, some antidepressants took effect by influencing autophagic process. Electroacupuncture (EA), as a common complementary treatment for may share mechanism autophagy hippocampus like antidepressants. To investigate that, sixty Sprague-Dawley rats firstly went through chronic unpredictable mild stress (CUMS) model establishment, 15 were assigned control group. After modeling, 45 successfully CUMS-induced randomly...

In programming, the names for program entities, especially methods, are intuitive characteristic understanding functionality of code. To ensure readability and maintainability programs, method should be named properly. Specifically, meaningful consistent with other used in related contexts their codebase. recent years, many automated approaches proposed to suggest among which neural machine translation (NMT) based models widely have achieved state-of-the-art results. However, these NMT-based...

Developers often perform repetitive code editing activities (up to 70%) for various reasons (e.g., refactoring) during software development. Many deep learning (DL) models have been proposed automate by from the history. Among DL-based models, pre-trained achieved state-of-the-art (SOTA) results. Pre-trained are first with pre-training tasks and fine-tuned task. Existing mainly infilling masked language modeling), which derived natural processing field not designed automatic editing. In this...

Life-long user behavior modeling, i.e., extracting a user's hidden interests from rich historical behaviors in months or even years, plays central role modern CTR prediction systems. Conventional algorithms mostly follow two cascading stages: simple General Search Unit (GSU) for fast and coarse search over tens of thousands long-term an Exact (ESU) effective Target Attention (TA) the small number finalists GSU. Although efficient, existing suffer crucial limitation: inconsistent...

Software developers frequently use code completion tools to accelerate software development by suggesting the following elements. Researchers usually employ AutoRegressive (AR) decoders complete sequences in a left-to-right, token-by-token fashion. To improve accuracy and efficiency of completion, we argue that tokens within statement have potential be predicted concurrently. In this article, first conduct an empirical study analyze dependency among target line-level completion. The results...

Currently, masked language modeling (e.g., BERT) is the prime choice to learn contextualized representations. Due pervasiveness, it naturally raises an interesting question: how do models (MLMs) contextual representations? In this work, we analyze learning dynamics of MLMs and find that adopts sampled embeddings as anchors estimate inject semantics representations, which limits efficiency effectiveness MLMs. To address these problems, propose TACO, a simple yet effective representation...

Growing evidence suggests that gut microbiota can affect depression-like behavior, and electroacupuncture (EA) regulate the composition abundance of microbiota. At same time, not a lot research has been done on how EA affects to behavior. The objective this study was associated mechanisms by which exerts antidepressant effects modulating microbiota.Twenty-four C57BL/6 male mice were randomly divided into three groups, one group (n = 8) normal control (NC). And other two groups chronic...

Abstract The ongoing decline of sea ice in the Arctic has heightened need for accurate sea‐ice forecasts to support environmental protection and resource development region beyond. While deep learning shown promise seasonal forecasting, most existing models overlook crucial influence atmospheric factors, thereby limiting their ability capture intricate characteristics system improve forecast accuracy. To address this deficiency, we propose an attention convolutional long short‐term memory...

Abstract Variations in the Pacific decadal oscillation (PDO) can influence marine ecosystems and regional climate phenomena. Accurate long‐term forecasts of PDO are therefore crucial for governance. This paper presents a novel seasonal gated recurrent unit (SGRU) model, based on deep learning, forecasting at multiple time scales. The model first decomposes complex nonlinear index series into three separate components, each retaining distinct pattern PDO. Next, three‐pathway GRU is...

Developers often perform repetitive code editing activities for various reasons (e.g., refactoring) during software development. Pre-trained models have achieved the state-of-the-art (SOTA) results. are first pre-trained with pre-training tasks and fine-tuned task. Existing mainly infilling masked language modeling), which derived from natural processing field not designed automatic editing. This paper proposes a novel task specialized in presents an effective model named CodeEditor. Our...

Abstract Volunteered geographic information (VGI) is an emerging phenomenon where anyone can create and share it with others. Compared traditional authoritative geospatial data, has several advantages, such as enriched instant updates, low cost. The object matching method widely used in VGI quality assessment data updates. However, faces certain challenges, the levels of detail that vary from to object, uneven distribution quality, automated requirement. To resolve these problems, this...

Land cover mapping plays a pivotal role in global resource monitoring, sustainable development research, and effective management. However, the complexity of process, coupled with significant computational data storage requirements, often leads to delays between processing product publication, thereby bringing challenges creating multi-timesteps large-area products for monitoring dynamic land cover. Therefore, improving efficiency each stage automating process is currently an urgent issue be...

To determine the susceptibilities of Mycoplasma homonis (M. hominis) to Chinese medicinal herbs in vitro, 30 clinical strains M. hominis were isolated and identified from specimen. The 19 determined by serial dilution methods vitro. results showed that was susceptible Radix Isatidis, Angelicae Dahuricae, Cortex Phellodendri, et Rhizoma Rhei, Fructus Kochiae Herba Houttuyniae. These findings laid a foundation treating infection with herbs.