Engineering Applications of Artificial Intelligence最新文献_第6页

Implicit local–global feature extraction for diffusion sequence recommendation 用于扩散序列推荐的隐式局部-全局特征提取

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-10-28 DOI: 10.1016/j.engappai.2024.109471

The existing research using diffusion model for item distribution modeling is a novel and effective recommendation method. However, the user interaction sequences contain multiple implicit features that reflect user preferences, and how to use implicit features to guide the diffusion process remains to be studied. Therefore, considering the dynamics of user preferences, we conduct fine-grained modeling of diffusion recommendation process. Specifically, we firstly define a sequence feature extraction layer that utilizes multi-scale convolutional neural networks and residual long short-term memory networks to learn local–global implicit features, and obtains implicit features through a weighted fusion strategy. Subsequently, the extracted output features are used as conditional inputs for the diffusion recommendation model to guide the denoising process. Finally, the items that meet user preferences are generated through the sampling and inference process to realize the personalized recommendation task. Through experiments on three publicly available datasets, the results show that the proposed model outperforms the strong baseline model in terms of performance. In addition, we conduct hyperparameter analysis and ablation experiments to verify the impact of model components on overall performance.

现有研究利用扩散模型对项目分布建模，是一种新颖有效的推荐方法。然而，用户交互序列包含多种反映用户偏好的隐含特征，如何利用隐含特征引导扩散过程仍有待研究。因此，考虑到用户偏好的动态变化，我们对扩散推荐过程进行了细粒度建模。具体来说，我们首先定义了一个序列特征提取层，利用多尺度卷积神经网络和残差长短期记忆网络学习局部-全局隐含特征，并通过加权融合策略获得隐含特征。随后，提取的输出特性被用作扩散推荐模型的条件输入，以指导去噪过程。最后，通过采样和推理过程生成符合用户偏好的项目，从而实现个性化推荐任务。通过在三个公开数据集上的实验，结果表明所提出的模型在性能上优于强基线模型。此外，我们还进行了超参数分析和消融实验，以验证模型组件对整体性能的影响。

{"title":"Implicit local–global feature extraction for diffusion sequence recommendation","authors":"","doi":"10.1016/j.engappai.2024.109471","DOIUrl":"10.1016/j.engappai.2024.109471","url":null,"abstract":"<div><div>The existing research using diffusion model for item distribution modeling is a novel and effective recommendation method. However, the user interaction sequences contain multiple implicit features that reflect user preferences, and how to use implicit features to guide the diffusion process remains to be studied. Therefore, considering the dynamics of user preferences, we conduct fine-grained modeling of diffusion recommendation process. Specifically, we firstly define a sequence feature extraction layer that utilizes multi-scale convolutional neural networks and residual long short-term memory networks to learn local–global implicit features, and obtains implicit features through a weighted fusion strategy. Subsequently, the extracted output features are used as conditional inputs for the diffusion recommendation model to guide the denoising process. Finally, the items that meet user preferences are generated through the sampling and inference process to realize the personalized recommendation task. Through experiments on three publicly available datasets, the results show that the proposed model outperforms the strong baseline model in terms of performance. In addition, we conduct hyperparameter analysis and ablation experiments to verify the impact of model components on overall performance.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":null,"pages":null},"PeriodicalIF":7.5,"publicationDate":"2024-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142533415","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

GFML: Gravity function for metric learning GFML：用于度量学习的重力函数

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-10-28 DOI: 10.1016/j.engappai.2024.109463

Diverse machine learning algorithms rely on the distance metric to compare and aggregate the information. A metric learning algorithm that captures the relevance between two vectors plays a critical role in machine learning. Metric learning may become biased toward the major classes and not be robust to the minor ones, i.e., metric learning may be vulnerable in an imbalanced dataset. We propose a gravity function-based metric learning (GFML) that captures the relationship between vectors based on the gravity function. We formulate GFML with two terms, (1) mass of the given vectors and (2) distance between the query and key vector. Mass learns the importance of the object itself, enabling robust metric learning on imbalanced datasets. GFML is simple and scalable; therefore, it can be adopted in diverse tasks. We validate that GFML improves the recommender system and image classification.

各种机器学习算法都依赖于距离度量来比较和汇总信息。能捕捉两个向量之间相关性的度量学习算法在机器学习中起着至关重要的作用。公因子学习可能会偏向于主要类别，而对次要类别不具有鲁棒性，也就是说，公因子学习在不平衡的数据集中可能很脆弱。我们提出了基于重力函数的度量学习（GFML），它能根据重力函数捕捉向量之间的关系。我们用两个术语来表述 GFML：(1) 给定向量的质量；(2) 查询与关键向量之间的距离。质量学习对象本身的重要性，从而在不平衡数据集上实现稳健的度量学习。GFML 简单且可扩展，因此可用于各种任务。我们验证了 GFML 能改进推荐系统和图像分类。

引用次数: 0

Search task extraction using k-contour based recurrent deep graph clustering 使用基于 k-轮廓的递归深度图聚类提取搜索任务

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-10-28 DOI: 10.1016/j.engappai.2024.109501

Search engines must accurately predict the implicit intent of users to effectively guide their online search experience and assist them in completing their tasks. Users create time-ordered query logs by performing various queries on search engines to access desired information. Search task extraction groups queries with the same intent into unique clusters, whether these queries come from different tasks within the same session or from the same task across different sessions. Accurate identification of user intent improves the performance of search-guiding processes, including query suggestion, personalized search, and advertisement retrieval. Many existing methods focus on creating graphs that show relationships between queries. However, these methods typically cluster the graph using simple threshold-based techniques rather than leveraging graph topological structure features. Recent studies have introduced deep clustering layers to prevent the model size from growing as the number of queries increases. However, these models rely on labeled data and overlook modern embeddings from language models. We propose a novel k-contour-based graph convolutional network connective proximity clustering layer (CoGCN-C-CL) architecture that clusters graphs without requiring labeled data by leveraging graph topological properties. CoGCN-C-CL simultaneously learns query representations and search tasks. The k-contours identify distinct regions of the graph, while the graph convolutional network (GCN) exploits interactions between nodes within these regions. Experimental results demonstrate that CoGCN-C-CL outperforms existing state-of-the-art search task clustering methods on frequently used search task datasets.

搜索引擎必须准确预测用户的隐含意图，才能有效指导用户的在线搜索体验，并协助他们完成任务。用户通过在搜索引擎上执行各种查询来获取所需信息，从而创建了有时间顺序的查询日志。无论这些查询是来自同一会话中的不同任务，还是来自不同会话中的同一任务，搜索任务提取都会将具有相同意图的查询归入独特的群组。准确识别用户意图可以提高搜索引导过程的性能，包括查询建议、个性化搜索和广告检索。许多现有方法都侧重于创建显示查询之间关系的图表。但是，这些方法通常使用简单的基于阈值的技术对图进行聚类，而不是利用图的拓扑结构特征。最近的研究引入了深度聚类层，以防止模型规模随着查询次数的增加而扩大。然而，这些模型依赖于标注数据，忽略了语言模型的现代嵌入。我们提出了一种新颖的基于 k-contour 的图卷积网络连接邻近聚类层（CoGCN-C-CL）架构，它利用图的拓扑特性，无需标注数据即可对图进行聚类。CoGCN-CL 可同时学习查询表示和搜索任务。k-contours 标识了图的不同区域，而图卷积网络 (GCN) 则利用了这些区域内节点之间的交互作用。实验结果表明，CoGCN-CL 在常用搜索任务数据集上的表现优于现有的最先进搜索任务聚类方法。

{"title":"Search task extraction using k-contour based recurrent deep graph clustering","authors":"","doi":"10.1016/j.engappai.2024.109501","DOIUrl":"10.1016/j.engappai.2024.109501","url":null,"abstract":"<div><div>Search engines must accurately predict the implicit intent of users to effectively guide their online search experience and assist them in completing their tasks. Users create time-ordered query logs by performing various queries on search engines to access desired information. Search task extraction groups queries with the same intent into unique clusters, whether these queries come from different tasks within the same session or from the same task across different sessions. Accurate identification of user intent improves the performance of search-guiding processes, including query suggestion, personalized search, and advertisement retrieval. Many existing methods focus on creating graphs that show relationships between queries. However, these methods typically cluster the graph using simple threshold-based techniques rather than leveraging graph topological structure features. Recent studies have introduced deep clustering layers to prevent the model size from growing as the number of queries increases. However, these models rely on labeled data and overlook modern embeddings from language models. We propose a novel k-contour-based graph convolutional network connective proximity clustering layer (CoGCN-C-CL) architecture that clusters graphs without requiring labeled data by leveraging graph topological properties. CoGCN-C-CL simultaneously learns query representations and search tasks. The k-contours identify distinct regions of the graph, while the graph convolutional network (GCN) exploits interactions between nodes within these regions. Experimental results demonstrate that CoGCN-C-CL outperforms existing state-of-the-art search task clustering methods on frequently used search task datasets.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":null,"pages":null},"PeriodicalIF":7.5,"publicationDate":"2024-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142533406","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

High-order correlation preserved multi-view unsupervised feature selection 保留高阶相关性的多视角无监督特征选择

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-10-28 DOI: 10.1016/j.engappai.2024.109507

Multi-view unsupervised feature selection (MUFS) has attracted considerable attention as an efficient dimensionality reduction technique. Data usually exhibit certain correlations, and in multi-view data there are more complex high-order correlations. However, some MUFS methods neglect to explore the high-order correlations. In addition, existing methods focus only on the high-order correlation between views or between samples. To tackle these shortcomings, this paper proposes a high-order correlation preserved MUFS (HCFS) method, which fully preserves both the high-order correlation between views and between samples. Specifically, HCFS embeds the energy preservation into the self-representation learning for multi-view data, which preserves the global structure while performing feature selection. Meanwhile, HCFS uses the adaptive weighting strategy to fuse the self-representation matrices of each view into a consistent graph, and constructs a hypergraph based on it to maintain the high-order correlation in the consistent information. Furthermore, the high-order correlation between views is preserved by low-rank tensor learning, and the local structure of data is preserved by using the hyper-Laplacian regularization. Extensive experimental results on eight public datasets demonstrate that the proposed method outperforms several existing state-of-the-art methods, which validates the effectiveness of the proposed method.

多视图无监督特征选择（MUFS）作为一种高效的降维技术，已经引起了广泛关注。数据通常表现出一定的相关性，而在多视图数据中，高阶相关性更为复杂。然而，一些 MUFS 方法忽视了对高阶相关性的探索。此外，现有方法只关注视图之间或样本之间的高阶相关性。针对这些不足，本文提出了一种保留高阶相关性的 MUFS（HCFS）方法，它既能完全保留视图之间的高阶相关性，也能完全保留样本之间的高阶相关性。具体来说，HCFS 将能量保存嵌入多视图数据的自表示学习中，在进行特征选择的同时保留了全局结构。同时，HCFS 利用自适应加权策略将各视图的自表示矩阵融合为一致图，并在此基础上构建超图，以保持一致信息中的高阶相关性。此外，还通过低阶张量学习保留了视图之间的高阶相关性，并利用超拉普拉奇正则化保留了数据的局部结构。在八个公共数据集上的大量实验结果表明，所提出的方法优于现有的几种最先进的方法，这验证了所提出方法的有效性。

{"title":"High-order correlation preserved multi-view unsupervised feature selection","authors":"","doi":"10.1016/j.engappai.2024.109507","DOIUrl":"10.1016/j.engappai.2024.109507","url":null,"abstract":"<div><div>Multi-view unsupervised feature selection (MUFS) has attracted considerable attention as an efficient dimensionality reduction technique. Data usually exhibit certain correlations, and in multi-view data there are more complex high-order correlations. However, some MUFS methods neglect to explore the high-order correlations. In addition, existing methods focus only on the high-order correlation between views or between samples. To tackle these shortcomings, this paper proposes a high-order correlation preserved MUFS (HCFS) method, which fully preserves both the high-order correlation between views and between samples. Specifically, HCFS embeds the energy preservation into the self-representation learning for multi-view data, which preserves the global structure while performing feature selection. Meanwhile, HCFS uses the adaptive weighting strategy to fuse the self-representation matrices of each view into a consistent graph, and constructs a hypergraph based on it to maintain the high-order correlation in the consistent information. Furthermore, the high-order correlation between views is preserved by low-rank tensor learning, and the local structure of data is preserved by using the hyper-Laplacian regularization. Extensive experimental results on eight public datasets demonstrate that the proposed method outperforms several existing state-of-the-art methods, which validates the effectiveness of the proposed method.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":null,"pages":null},"PeriodicalIF":7.5,"publicationDate":"2024-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142533414","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Learning to generate and evaluate fact-checking explanations with transformers 学习用转换器生成和评估事实核查说明

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-10-28 DOI: 10.1016/j.engappai.2024.109492

In an era increasingly dominated by digital platforms, the spread of misinformation poses a significant challenge, highlighting the need for solutions capable of assessing information veracity. Our research contributes to the field of Explainable Artificial Antelligence (XAI) by developing transformer-based fact-checking models that contextualise and justify their decisions by generating human-accessible explanations. Importantly, we also develop models for automatic evaluation of explanations for fact-checking verdicts across different dimensions such as (self)-contradiction, hallucination, convincingness and overall quality. By introducing human-centred evaluation methods and developing specialised datasets, we emphasise the need for aligning Artificial Intelligence (AI)-generated explanations with human judgements. This approach not only advances theoretical knowledge in XAI but also holds practical implications by enhancing the transparency, reliability and users’ trust in AI-driven fact-checking systems. Furthermore, the development of our metric learning models is a first step towards potentially increasing efficiency and reducing reliance on extensive manual assessment. Based on experimental results, our best performing generative model achieved a Recall-Oriented Understudy for Gisting Evaluation-1 (ROUGE-1) score of 47.77 demonstrating superior performance in generating fact-checking explanations, particularly when provided with high-quality evidence. Additionally, the best performing metric learning model showed a moderately strong correlation with human judgements on objective dimensions such as (self)-contradiction and hallucination, achieving a Matthews Correlation Coefficient (MCC) of around 0.7.

在数字平台日益占主导地位的时代，错误信息的传播构成了巨大的挑战，凸显了对能够评估信息真实性的解决方案的需求。我们的研究为可解释人工智能（XAI）领域做出了贡献，我们开发了基于变压器的事实核查模型，通过生成人类可理解的解释，对其决策进行上下文说明和论证。重要的是，我们还开发了模型，用于自动评估不同维度的事实核查判决解释，如（自我）矛盾、幻觉、说服力和整体质量。通过引入以人为本的评估方法和开发专用数据集，我们强调了将人工智能（AI）生成的解释与人类判断相统一的必要性。这种方法不仅推动了 XAI 理论知识的发展，还通过提高人工智能驱动的事实核查系统的透明度、可靠性和用户信任度，产生了实际影响。此外，我们的度量学习模型的开发是提高效率和减少对大量人工评估依赖的第一步。根据实验结果，我们性能最好的生成模型获得了以召回为导向的 Gisting 评估-1（ROUGE-1）得分 47.77，这表明我们在生成事实核查解释方面表现出色，尤其是在提供高质量证据的情况下。此外，在（自我）矛盾和幻觉等客观维度上，表现最佳的度量学习模型与人类判断显示出适度的强相关性，马修斯相关系数（MCC）达到 0.7 左右。

{"title":"Learning to generate and evaluate fact-checking explanations with transformers","authors":"","doi":"10.1016/j.engappai.2024.109492","DOIUrl":"10.1016/j.engappai.2024.109492","url":null,"abstract":"<div><div>In an era increasingly dominated by digital platforms, the spread of misinformation poses a significant challenge, highlighting the need for solutions capable of assessing information veracity. Our research contributes to the field of Explainable Artificial Antelligence (XAI) by developing transformer-based fact-checking models that contextualise and justify their decisions by generating human-accessible explanations. Importantly, we also develop models for automatic evaluation of explanations for fact-checking verdicts across different dimensions such as <span>(self)-contradiction</span>, <span>hallucination</span>, <span>convincingness</span> and <span>overall quality</span>. By introducing human-centred evaluation methods and developing specialised datasets, we emphasise the need for aligning Artificial Intelligence (AI)-generated explanations with human judgements. This approach not only advances theoretical knowledge in XAI but also holds practical implications by enhancing the transparency, reliability and users’ trust in AI-driven fact-checking systems. Furthermore, the development of our metric learning models is a first step towards potentially increasing efficiency and reducing reliance on extensive manual assessment. Based on experimental results, our best performing generative model achieved a Recall-Oriented Understudy for Gisting Evaluation-1 (<span>ROUGE-1</span>) score of 47.77 demonstrating superior performance in generating fact-checking explanations, particularly when provided with high-quality evidence. Additionally, the best performing metric learning model showed a moderately strong correlation with human judgements on objective dimensions such as <span>(self)-contradiction</span> and <span>hallucination</span>, achieving a Matthews Correlation Coefficient (MCC) of around 0.7.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":null,"pages":null},"PeriodicalIF":7.5,"publicationDate":"2024-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142533408","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Reference-based image super-resolution of hyperspectral and red-green-blue image for determination of wheat kernel quality using deep learning networks 利用深度学习网络对高光谱和红绿蓝图像进行基于参考的图像超分辨率处理，以确定小麦籽粒质量

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-10-25 DOI: 10.1016/j.engappai.2024.109513

In the process of cultivation and harvest, wheat kernel quality is highly susceptible to various factors, such as disease, mildew, atrophy and impurities, and detection of kernel quality is essential to avoid hazard proliferation, facilitate product grading, and ensure food safety. Possessing abundant image and spectral characteristics, hyperspectral imaging (HSI) has gained impressive achievements in kernel quality analysis, but its low spatial resolution limits its detection accuracy. In this study, reference-based image super-resolution (RefSR) of HSI and Red-Green-Blue image was adopted to improve resolution to determine wheat kernel quality using deep learning networks. Firstly, RefSR was conducted by the improved transformer network with dual-branch feature extraction and weighted fusion operation and achieved excellent RefSR with significant resolution improvement, peak signal to noise ratio of 35.521 and structural similarity index of 0.97, outweighing the existing state-of-the-art networks. Then, the reflectance images (RIs) of effective wavelengths (EWs) from generated HSI images were combined with the residual network with a spatial, channel attention and multi-scale residual to determine wheat kernel quality. Precise analysis was achieved with the accuracy in calibration, validation and prediction sets of 100.00%, 95.26% and 92.78%. RefSR provides a novel and efficient approach for obtaining HSI images of high spatial resolution and facilitates the application of HSI in analysis of crop kernels. RIs of several sporadic EWs can be easily acquired and processed, achieving field and rapid kernel detection. Therefore, the proposed method furnishes the efficient, accurate and applicable determination of wheat kernel quality.

在种植和收获过程中，小麦籽粒质量极易受到病害、霉变、萎缩和杂质等各种因素的影响，而籽粒质量检测对于避免危害扩散、促进产品分级和确保食品安全至关重要。高光谱成像（HSI）具有丰富的图像和光谱特性，在果仁质量分析方面取得了令人瞩目的成就，但其较低的空间分辨率限制了其检测精度。本研究采用基于参考的高光谱成像和红绿蓝图像超分辨率（RefSR）技术，利用深度学习网络提高分辨率以确定小麦籽粒质量。首先，通过改进的变压器网络进行双分支特征提取和加权融合操作，实现了出色的 RefSR，分辨率显著提高，峰值信噪比为 35.521，结构相似度指数为 0.97，超过了现有的先进网络。然后，将生成的 HSI 图像中有效波长（EW）的反射率图像（RIs）与具有空间、通道关注和多尺度残差的残差网络相结合，以确定小麦籽粒质量。在校准、验证和预测组中，精确分析的准确率分别达到 100.00%、95.26% 和 92.78%。RefSR 为获取高空间分辨率的 HSI 图像提供了一种新颖、高效的方法，并促进了 HSI 在作物籽粒分析中的应用。多个零星 EW 的 RIs 可以轻松获取和处理，从而实现田间和快速的果核检测。因此，所提出的方法可高效、准确、适用地测定小麦籽粒质量。

{"title":"Reference-based image super-resolution of hyperspectral and red-green-blue image for determination of wheat kernel quality using deep learning networks","authors":"","doi":"10.1016/j.engappai.2024.109513","DOIUrl":"10.1016/j.engappai.2024.109513","url":null,"abstract":"<div><div>In the process of cultivation and harvest, wheat kernel quality is highly susceptible to various factors, such as disease, mildew, atrophy and impurities, and detection of kernel quality is essential to avoid hazard proliferation, facilitate product grading, and ensure food safety. Possessing abundant image and spectral characteristics, hyperspectral imaging (HSI) has gained impressive achievements in kernel quality analysis, but its low spatial resolution limits its detection accuracy. In this study, reference-based image super-resolution (RefSR) of HSI and Red-Green-Blue image was adopted to improve resolution to determine wheat kernel quality using deep learning networks. Firstly, RefSR was conducted by the improved transformer network with dual-branch feature extraction and weighted fusion operation and achieved excellent RefSR with significant resolution improvement, peak signal to noise ratio of 35.521 and structural similarity index of 0.97, outweighing the existing state-of-the-art networks. Then, the reflectance images (RIs) of effective wavelengths (EWs) from generated HSI images were combined with the residual network with a spatial, channel attention and multi-scale residual to determine wheat kernel quality. Precise analysis was achieved with the accuracy in calibration, validation and prediction sets of 100.00%, 95.26% and 92.78%. RefSR provides a novel and efficient approach for obtaining HSI images of high spatial resolution and facilitates the application of HSI in analysis of crop kernels. RIs of several sporadic EWs can be easily acquired and processed, achieving field and rapid kernel detection. Therefore, the proposed method furnishes the efficient, accurate and applicable determination of wheat kernel quality.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":null,"pages":null},"PeriodicalIF":7.5,"publicationDate":"2024-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142533153","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Consistency-guided Multi-Source-Free Domain Adaptation 一致性指导的多源自由领域适应性

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-10-25 DOI: 10.1016/j.engappai.2024.109497

Deep neural networks suffer from severe performance degradation when facing a distribution shift between the labeled source domain and unlabeled target domain. Domain adaptation addresses this issue by aligning the feature distributions of both domains. Conventional methods assume that the labeled source samples are drawn from a single data distribution (domain) and can be fully accessed during training. However, in real applications, multiple source domains with different distributions often exist, and source samples may be unavailable due to privacy and storage constraints. To address multi-source and data-free challenges, Multi-Source-Free Domain Adaptation (MSFDA) uses only diverse pre-trained source models without requiring any source data. Most existing MSFDA methods adapt each source model to the target domain individually, making them ineffective in leveraging the complementary transferable knowledge from different source models. In this paper, we propose a novel COnsistency-guided multi-source-free Domain Adaptation (CODA) method, which leverages the label consistency criterion as a bridge to facilitate the cooperation among source models. CODA applies consistency regularization on the soft labels of weakly- and strongly-augmented target samples from each pair of source models, allowing them to supervise each other. To achieve high-quality pseudo-labels, CODA also performs a consistency-based denoising to unify the pseudo-labels from different source models. Finally, CODA optimally combines different source models by maximizing the mutual information of the predictions of the resulting target model. Extensive experiments on four benchmark datasets demonstrate the effectiveness of CODA compared to the state-of-the-art methods.

深度神经网络在面对有标签源域和无标签目标域之间的分布变化时，性能会严重下降。域适应通过调整两个域的特征分布来解决这一问题。传统方法假定有标签的源样本来自单一数据分布（域），并且在训练过程中可以完全访问。然而，在实际应用中，往往存在多个具有不同分布的源域，而且由于隐私和存储限制，可能无法获得源样本。为了应对多源和无数据的挑战，多源无域适应（MSFDA）只使用不同的预训练源模型，而不需要任何源数据。现有的大多数 MSFDA 方法都是将每个源模型单独适应目标领域，因此无法有效利用不同源模型的互补可转移知识。在本文中，我们提出了一种新颖的以一致性为指导的无源多领域适应（CODA）方法，该方法利用标签一致性准则作为桥梁，促进源模型之间的合作。CODA 对每对源模型的弱增强和强增强目标样本的软标签进行一致性正则化，使它们能够相互监督。为了获得高质量的伪标签，CODA 还执行了基于一致性的去噪处理，以统一来自不同源模型的伪标签。最后，CODA 通过最大化目标模型预测的互信息来优化组合不同的源模型。在四个基准数据集上进行的广泛实验证明，与最先进的方法相比，CODA 是有效的。

{"title":"Consistency-guided Multi-Source-Free Domain Adaptation","authors":"","doi":"10.1016/j.engappai.2024.109497","DOIUrl":"10.1016/j.engappai.2024.109497","url":null,"abstract":"<div><div>Deep neural networks suffer from severe performance degradation when facing a distribution shift between the labeled source domain and unlabeled target domain. Domain adaptation addresses this issue by aligning the feature distributions of both domains. Conventional methods assume that the labeled source samples are drawn from a single data distribution (domain) and can be fully accessed during training. However, in real applications, multiple source domains with different distributions often exist, and source samples may be unavailable due to privacy and storage constraints. To address multi-source and data-free challenges, Multi-Source-Free Domain Adaptation (MSFDA) uses only diverse pre-trained source models without requiring any source data. Most existing MSFDA methods adapt each source model to the target domain individually, making them ineffective in leveraging the complementary transferable knowledge from different source models. In this paper, we propose a novel COnsistency-guided multi-source-free Domain Adaptation (CODA) method, which leverages the label consistency criterion as a bridge to facilitate the cooperation among source models. CODA applies consistency regularization on the soft labels of weakly- and strongly-augmented target samples from each pair of source models, allowing them to supervise each other. To achieve high-quality pseudo-labels, CODA also performs a consistency-based denoising to unify the pseudo-labels from different source models. Finally, CODA optimally combines different source models by maximizing the mutual information of the predictions of the resulting target model. Extensive experiments on four benchmark datasets demonstrate the effectiveness of CODA compared to the state-of-the-art methods.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":null,"pages":null},"PeriodicalIF":7.5,"publicationDate":"2024-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142533289","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Interactive streaming feature selection based on neighborhood rough sets 基于邻域粗糙集的交互式流媒体特征选择

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-10-25 DOI: 10.1016/j.engappai.2024.109479

Feature streams refer to features that arrive continuously over time without changing the number of samples. Such data is commonly encountered in various practical application scenarios. Stream feature selection is a technique designed to select relevant features from high-dimensional stream data, thereby reducing its overall size. Feature interaction plays a crucial role in influencing the results of feature selection. Most existing methods address stream feature selection primarily by focusing on irrelevance and redundancy, often overlooking the important interactions between features. Additionally, these methods typically assume that all samples and features are known, which contradicts the fundamental nature of streaming data. This study introduces an interactive feature selection approach for stream feature selection, utilizing the neighborhood rough set. First, we provide a basic explanation of multi-neighbor entropy, which measures the amount of information related to neighborhood classes. It is used to measure how the amount of information about neighborhood classes. Next, we propose a feature evaluation method based on correlation, redundancy, and interaction analysis. Finally, we elaborate on functions for feature evaluation criteria, aiming to design streaming feature selection algorithms that integrate correlation, redundancy, and interactivity. The proposed algorithm is compared with six other representative feature selection algorithms across 14 public datasets. Experimental results demonstrate the validity of our proposed solution.

特征流指的是在不改变样本数量的情况下随时间连续到达的特征。在各种实际应用场景中，经常会遇到这样的数据。数据流特征选择是一种技术，旨在从高维数据流中选择相关特征，从而缩小数据流的总体规模。特征交互对特征选择结果的影响至关重要。大多数现有方法主要通过关注无关性和冗余性来解决流特征选择问题，但往往忽略了特征之间的重要交互作用。此外，这些方法通常假设所有样本和特征都是已知的，这与流数据的基本性质相矛盾。本研究介绍了一种利用邻域粗糙集进行流特征选择的交互式特征选择方法。首先，我们对多邻类熵进行了基本解释。它用于衡量邻域类的信息量。接下来，我们提出了一种基于相关性、冗余性和交互性分析的特征评估方法。最后，我们阐述了特征评估标准的函数，旨在设计出整合相关性、冗余性和交互性的流特征选择算法。我们将所提出的算法与其他六种具有代表性的特征选择算法在 14 个公共数据集上进行了比较。实验结果证明了我们提出的解决方案的有效性。

{"title":"Interactive streaming feature selection based on neighborhood rough sets","authors":"","doi":"10.1016/j.engappai.2024.109479","DOIUrl":"10.1016/j.engappai.2024.109479","url":null,"abstract":"<div><div>Feature streams refer to features that arrive continuously over time without changing the number of samples. Such data is commonly encountered in various practical application scenarios. Stream feature selection is a technique designed to select relevant features from high-dimensional stream data, thereby reducing its overall size. Feature interaction plays a crucial role in influencing the results of feature selection. Most existing methods address stream feature selection primarily by focusing on irrelevance and redundancy, often overlooking the important interactions between features. Additionally, these methods typically assume that all samples and features are known, which contradicts the fundamental nature of streaming data. This study introduces an interactive feature selection approach for stream feature selection, utilizing the neighborhood rough set. First, we provide a basic explanation of multi-neighbor entropy, which measures the amount of information related to neighborhood classes. It is used to measure how the amount of information about neighborhood classes. Next, we propose a feature evaluation method based on correlation, redundancy, and interaction analysis. Finally, we elaborate on functions for feature evaluation criteria, aiming to design streaming feature selection algorithms that integrate correlation, redundancy, and interactivity. The proposed algorithm is compared with six other representative feature selection algorithms across 14 public datasets. Experimental results demonstrate the validity of our proposed solution.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":null,"pages":null},"PeriodicalIF":7.5,"publicationDate":"2024-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142533290","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A dual-branch convolutional neural network with domain-informed attention for arrhythmia classification of 12-lead electrocardiograms 用于 12 导联心电图心律失常分类的具有领域信息关注的双分支卷积神经网络

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-10-25 DOI: 10.1016/j.engappai.2024.109480

The automatic classification of arrhythmia is an important task in the intelligent auxiliary diagnosis of an electrocardiogram. Its efficiency and accuracy are vital for practical deployment and applications in the medical field. For the 12-lead electrocardiogram, we know that the comprehensive utilization of lead characteristics is key to enhancing diagnostic accuracy. However, existing classification methods (1) neglect the similarities and differences between the limb lead group and the precordial lead group; (2) the commonly adopted attention mechanisms struggle to capture the domain characteristics in an electrocardiogram. To address these issues, we propose a new dual-branch convolutional neural network with domain-informed attention, which is novel in two ways. First, it adopts a dual-branch network to extract intra-group similarities and inter-group differences of limb and precordial leads. Second, it proposes a domain-informed attention mechanism to embed the critical domain knowledge of electrocardiogram, multiple RR (R wave to R wave) intervals, into coordinated attention to adaptively assign attention weights to key segments, thereby effectively capturing the characteristics of the electrocardiogram domain. Experimental results show that our method achieves an F1-score of 0.905 and a macro area under the curve of 0.936 on two widely used large-scale datasets, respectively. Compared to state-of-the-art methods, our method shows significant performance improvements with a drastic reduction in model parameters.

心律失常的自动分类是心电图智能辅助诊断中的一项重要任务。其效率和准确性对于医疗领域的实际部署和应用至关重要。对于 12 导联心电图，我们知道综合利用导联特征是提高诊断准确性的关键。然而，现有的分类方法（1）忽视了肢体导联组和心前区导联组之间的异同；（2）普遍采用的注意机制难以捕捉心电图中的领域特征。为了解决这些问题，我们提出了一种新的双分支卷积神经网络，该网络具有两方面的新颖性。首先，它采用双分支网络来提取肢体导联和心前区导联的组内相似性和组间差异。其次，它提出了一种领域知情注意力机制，将心电图的关键领域知识--多个 RR（R 波到 R 波）间隔嵌入到协调注意力中，自适应地为关键片段分配注意力权重，从而有效捕捉心电图领域的特征。实验结果表明，我们的方法在两个广泛使用的大规模数据集上分别取得了 0.905 的 F1 分数和 0.936 的宏观曲线下面积。与最先进的方法相比，我们的方法在大幅减少模型参数的同时，还显著提高了性能。

{"title":"A dual-branch convolutional neural network with domain-informed attention for arrhythmia classification of 12-lead electrocardiograms","authors":"","doi":"10.1016/j.engappai.2024.109480","DOIUrl":"10.1016/j.engappai.2024.109480","url":null,"abstract":"<div><div>The automatic classification of arrhythmia is an important task in the intelligent auxiliary diagnosis of an electrocardiogram. Its efficiency and accuracy are vital for practical deployment and applications in the medical field. For the 12-lead electrocardiogram, we know that the comprehensive utilization of lead characteristics is key to enhancing diagnostic accuracy. However, existing classification methods (1) neglect the similarities and differences between the limb lead group and the precordial lead group; (2) the commonly adopted attention mechanisms struggle to capture the domain characteristics in an electrocardiogram. To address these issues, we propose a new dual-branch convolutional neural network with domain-informed attention, which is novel in two ways. First, it adopts a dual-branch network to extract intra-group similarities and inter-group differences of limb and precordial leads. Second, it proposes a domain-informed attention mechanism to embed the critical domain knowledge of electrocardiogram, multiple RR (R wave to R wave) intervals, into coordinated attention to adaptively assign attention weights to key segments, thereby effectively capturing the characteristics of the electrocardiogram domain. Experimental results show that our method achieves an F1-score of 0.905 and a macro area under the curve of 0.936 on two widely used large-scale datasets, respectively. Compared to state-of-the-art methods, our method shows significant performance improvements with a drastic reduction in model parameters.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":null,"pages":null},"PeriodicalIF":7.5,"publicationDate":"2024-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142533155","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

TSD-DETR: A lightweight real-time detection transformer of traffic sign detection for long-range perception of autonomous driving TSD-DETR：用于自动驾驶远距离感知的轻量级交通标志检测实时检测变换器

IF 7.5 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Engineering Applications of Artificial Intelligence

Pub Date : 2024-10-25 DOI: 10.1016/j.engappai.2024.109536

The key to accurate perception and efficient decision making of autonomous driving is the long-range detection of traffic signs. Long-range detection of traffic signs has the problems of small traffic sign size and complex background. In order to solve these problems, this paper proposes a lightweight model for traffic sign detection based on real-time detection transformer (TSD-DETR). Firstly, the feature extraction module is constructed using multiple types of convolutional modules. The model extracts multi-scale features of different levels to enhance feature extraction ability. Then, small object detection module and detection head are designed to extract and detect shallow features. It can improve the detection of small traffic signs. Finally, Efficient Multi-Scale Attention is introduced to adjust the channel weights. It aggregates the output features of three parallel branches interactively. TSD-DETR achieves a mean average precision (mAp) of 96.8% on Tsinghua-Tencent 100K dataset. It is improved by 2.5% compared with real-time detection transformer. In small object detection, mAp improved by 9%. TSD-DETR achieves 99.4% mAp on the Changsha University of Science and Technology Chinese Traffic Sign Detection Benchmark dataset, with an improvement of 0.6%. The experimental results show that TSD-DETR reduces the number of parameters by 9.06M by optimizing the model structure. On the premise of ensuring the real-time performance of the model, the detection accuracy of the model is improved greatly. The results of ablation experiments show that the feature extraction module and small object detection module proposed in this paper are conducive to improving the detection accuracy.

自动驾驶的准确感知和高效决策关键在于交通标志的远距离探测。交通标志的远距离检测存在交通标志尺寸小、背景复杂等问题。为了解决这些问题，本文提出了一种基于实时检测变换器（TSD-DETR）的轻量级交通标志检测模型。首先，利用多种卷积模块构建特征提取模块。该模型提取不同层次的多尺度特征，以增强特征提取能力。然后，设计了小目标检测模块和检测头，用于提取和检测浅层特征。它可以提高对小型交通标志的检测能力。最后，引入高效多尺度关注来调整通道权重。它将三个并行分支的输出特性进行交互式聚合。在清华-腾讯 100K 数据集上，TSD-DETR 的平均精度 (mAp) 达到 96.8%。与实时检测转换器相比，提高了 2.5%。在小物体检测方面，mAp 提高了 9%。在长沙理工大学中国交通标志检测基准数据集上，TSD-DETR 的 mAp 达到 99.4%，提高了 0.6%。实验结果表明，TSD-DETR 通过优化模型结构，减少了 9.06M 的参数数量。在保证模型实时性的前提下，大大提高了模型的检测精度。烧蚀实验结果表明，本文提出的特征提取模块和小目标检测模块有利于提高检测精度。

{"title":"TSD-DETR: A lightweight real-time detection transformer of traffic sign detection for long-range perception of autonomous driving","authors":"","doi":"10.1016/j.engappai.2024.109536","DOIUrl":"10.1016/j.engappai.2024.109536","url":null,"abstract":"<div><div>The key to accurate perception and efficient decision making of autonomous driving is the long-range detection of traffic signs. Long-range detection of traffic signs has the problems of small traffic sign size and complex background. In order to solve these problems, this paper proposes a lightweight model for traffic sign detection based on real-time detection transformer (TSD-DETR). Firstly, the feature extraction module is constructed using multiple types of convolutional modules. The model extracts multi-scale features of different levels to enhance feature extraction ability. Then, small object detection module and detection head are designed to extract and detect shallow features. It can improve the detection of small traffic signs. Finally, Efficient Multi-Scale Attention is introduced to adjust the channel weights. It aggregates the output features of three parallel branches interactively. TSD-DETR achieves a mean average precision (mAp) of 96.8% on Tsinghua-Tencent 100K dataset. It is improved by 2.5% compared with real-time detection transformer. In small object detection, mAp improved by 9%. TSD-DETR achieves 99.4% mAp on the Changsha University of Science and Technology Chinese Traffic Sign Detection Benchmark dataset, with an improvement of 0.6%. The experimental results show that TSD-DETR reduces the number of parameters by 9.06M by optimizing the model structure. On the premise of ensuring the real-time performance of the model, the detection accuracy of the model is improved greatly. The results of ablation experiments show that the feature extraction module and small object detection module proposed in this paper are conducive to improving the detection accuracy.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":null,"pages":null},"PeriodicalIF":7.5,"publicationDate":"2024-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142533407","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0