IEEE Transactions on Emerging Topics in Computational Intelligence最新文献_第4页

Comparative Analysis of Inference Performance of Pre-Trained Convolutional Neural Networks in Analog Accelerators 预训练卷积神经网络在模拟加速器中的推理性能比较分析

IF 6 3区计算机科学

IEEE Transactions on Emerging Topics in Computational Intelligence Pub Date : 2026-04-01 Epub Date: 2025-12-17 DOI: 10.1109/TETCI.2025.3641703

Mafizur Rahman;Lijun Qian

{"title":"Comparative Analysis of Inference Performance of Pre-Trained Convolutional Neural Networks in Analog Accelerators","authors":"Mafizur Rahman;Lijun Qian","doi":"10.1109/TETCI.2025.3641703","DOIUrl":"https://doi.org/10.1109/TETCI.2025.3641703","url":null,"abstract":"Resistive crossbars using non-volatile memory devices have become promising components for implementing deep learning models in hardware. However, crossbar-based computations encounter a notable challenge due to device and circuit-level non-idealities. In this study, we explore three recent crossbar simulation tools, namely, AIHWKIT, CrossSim, and MemTorch, and evaluate four pre-trained convolutional neural networks (CNNs) for image classification on CIFAR-100, MNIST, and SVHN datasets using these tools. To the best of our knowledge, this is the first study using all three simulation tools of analog accelerators, representing three distinctive hardware settings, to evaluate CNN-based model robustness under analog noise and nonlinearities. We first test the robustness of four models (VGG19, InceptionV3, ResNet50, & MobilenetV2) using different levels of white Gaussian noise as baselines. Then we evaluate their inference performance on the three tools to determine their resilience to analog noise and nonlinearities in the hardware environment. Results show that while all CNNs' suffer performance degradation as expected, insights are obtained such as ResNet50 outperforms others in two out of three simulators despite real-world hardware imperfections due to its deep structure. Furthermore, we explore the impact of three key factors: cell bits, ADC resolution, and tile size on CNN performance across various hardware configurations, providing recommendations for the efficient deployment of CNN on hardware. Our results suggest that increasing cell bits and ADC resolution improves CNNs' performance. It is also observed that smaller tiles are suitable for lightweight models, while for complex networks, we should choose larger shape of tiles.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 2","pages":"1324-1339"},"PeriodicalIF":6.0,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148511353","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

Toward Temporal Efficient Spiking Neural Network via Dynamic Training and Inference 基于动态训练和推理的时间高效脉冲神经网络

IF 6 3区计算机科学

IEEE Transactions on Emerging Topics in Computational Intelligence Pub Date : 2026-04-01 Epub Date: 2025-11-17 DOI: 10.1109/TETCI.2025.3628761

Na Xia;Xu Liu;Sizhou Wei;Lei Chen;Haoyu Ping

{"title":"Toward Temporal Efficient Spiking Neural Network via Dynamic Training and Inference","authors":"Na Xia;Xu Liu;Sizhou Wei;Lei Chen;Haoyu Ping","doi":"10.1109/TETCI.2025.3628761","DOIUrl":"https://doi.org/10.1109/TETCI.2025.3628761","url":null,"abstract":"Spiking Neural Networks (SNNs) have emerged as a promising alternative to traditional Artificial Neural Networks (ANNs) due to their biological interpretability. SNNs process input information in both spatial and temporal dimensions using binary spikes, offering advantages such as low energy consumption and compatibility with neural devices. Previous works have demonstrated that directly training SNNs from scratch has great potential. It can achieve performance comparable to traditional ANNs while significantly reducing energy consumption and the number of required time steps. Moreover, it outperforms ANN-SNN conversion methods. However, prior studies typically employ a fixed timestep during both training and inference, leading to two main problems: 1) During training, it results in complex gradient propagation and gradient estimation errors. 2) During inference, it causes energy waste and performance bottlenecks. To address these issues, we propose a novel strategy called Momentum Performance Queue Training and Confidence Score Inference (<bold>MPQ-CS). Firstly, MPQ-CS dynamically adjusts the timestep during the training stage based on the current training performance of the SNN using a Momentum Performance Queue (MPQ). Additionally, MPQ-CS dynamically adjusts the timestep during the inference stage using a Confidence Score (CS) based on the difficulty of input samples. Extensive experiments on multiple datasets and different SNN architectures demonstrate that MPQ-CS significantly enhances the overall performance of SNNs during both training and inference stages.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 2","pages":"1999-2008"},"PeriodicalIF":6.0,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148511746","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

Image Super-Resolution Reconstruction Network Based on Enhanced Swin Transformer via Alternating Aggregation of Local-Global Features 基于局部-全局特征交替聚合的增强型Swin变压器图像超分辨率重建网络

IF 6 3区计算机科学

IEEE Transactions on Emerging Topics in Computational Intelligence Pub Date : 2026-04-01 Epub Date: 2025-10-27 DOI: 10.1109/TETCI.2025.3619579

Yuming Huang;Yingpin Chen;Changhui Wu;Binhui Song;Hui Wang

{"title":"Image Super-Resolution Reconstruction Network Based on Enhanced Swin Transformer via Alternating Aggregation of Local-Global Features","authors":"Yuming Huang;Yingpin Chen;Changhui Wu;Binhui Song;Hui Wang","doi":"10.1109/TETCI.2025.3619579","DOIUrl":"https://doi.org/10.1109/TETCI.2025.3619579","url":null,"abstract":"The Swin Transformer image super-resolution (SR) reconstruction network primarily depends on the long-range relationship of the window and shifted window attention to explore features. However, this approach focuses only on global features, ignoring local ones, and considers only spatial interactions, disregarding channel and spatial-channel feature interactions, limiting its nonlinear mapping capability. Therefore, this study proposes an enhanced Swin Transformer network (ESTN) that alternately aggregates local and global features. During local feature aggregation, shift convolution facilitates the interaction between local spatial and channel information. During global feature aggregation, a block sparse global perception module is introduced, wherein spatial information is reorganized and the recombined features are then processed by a dense layer to achieve global perception. Additionally, multiscale self-attention and low-parameter residual channel attention modules are introduced to aggregate information across different scales. Finally, the effectiveness of ESTN on five public datasets and a local attribution map (LAM) are analyzed. Experimental results demonstrate that the proposed ESTN achieves higher average PSNR, surpassing SRCNN, ELAN-light, SwinIR-light, and SMFANER+ models by 2.17 dB, 0.13 dB, 0.12 dB, and 0.1 dB, respectively, with LAM further confirming its larger receptive field. ESTN delivers improved quality of SR images.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 2","pages":"1550-1564"},"PeriodicalIF":6.0,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148511701","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

Not Only Detection But Also Explain: An Interpretable Multimodal System for Anomaly Detection 既要检测又要解释：一个可解释的多模态异常检测系统

IF 6 3区计算机科学

IEEE Transactions on Emerging Topics in Computational Intelligence Pub Date : 2026-04-01 Epub Date: 2026-01-28 DOI: 10.1109/TETCI.2026.3654456

Chih-Yung Chang;Wen-Dong Jiang;I-Hsiung Chang;Diptendu Sinha Roy

{"title":"Not Only Detection But Also Explain: An Interpretable Multimodal System for Anomaly Detection","authors":"Chih-Yung Chang;Wen-Dong Jiang;I-Hsiung Chang;Diptendu Sinha Roy","doi":"10.1109/TETCI.2026.3654456","DOIUrl":"https://doi.org/10.1109/TETCI.2026.3654456","url":null,"abstract":"Anomaly detection is often formulated as a Multiple Instance Learning (MIL) problem, where an anomaly detector learns to generate frame-level anomaly scores under multi-modal supervision. However, most previous studies have two drawbacks: 1) the feature fusion between attributes overlooks the model’s interpretability; and 2) the training strategy based on pre-trained models leads to excessively long training times. To address these issues, this paper proposes an <underline>Interpretable <underline>Multi-modal <underline>Anomaly <underline>Detection <underline>System called IMADS. The proposed IMADS abandons the traditional multi-modal fusion approach and adopts single-modal training with late-stage decision fusion optimized by Bayesian techniques to resolve the interpretability problem caused by traditional feature fusion. Moreover, the proposed IMADS introduces a transfer learning method from frame-level to video-level, which essentially adds a temporal dimension to the 2D-to-3D conversion by randomizing parameters along the time dimension during training. This approach not only improves training efficiency but also ensures the model’s interpretability. Experimental results demonstrate that the proposed IMADS significantly outperforms existing methods in terms of performance, proving its innovativeness and effectiveness.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 2","pages":"1863-1877"},"PeriodicalIF":6.0,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148512088","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

Transitive Vision-Language Prompt Learning for Domain Generalization 领域泛化的传递性视觉语言提示学习

IF 6 3区计算机科学

IEEE Transactions on Emerging Topics in Computational Intelligence Pub Date : 2026-04-01 Epub Date: 2025-12-01 DOI: 10.1109/TETCI.2025.3628755

Liyuan Wang;Yan Jin;Zhen Chen;Jinlin Wu;Mengke Li;Yang Lu;Hanzi Wang

{"title":"Transitive Vision-Language Prompt Learning for Domain Generalization","authors":"Liyuan Wang;Yan Jin;Zhen Chen;Jinlin Wu;Mengke Li;Yang Lu;Hanzi Wang","doi":"10.1109/TETCI.2025.3628755","DOIUrl":"https://doi.org/10.1109/TETCI.2025.3628755","url":null,"abstract":"The vision-language pre-training has enabled deep models to make a huge step forward in generalizing across unseen domains. The recent learning method based on the vision-language pre-training model is a great tool for domain generalization and can solve this problem to a large extent. However, there are still some issues that an advancement still suffers from trading-off between domain invariance and class separability, which are crucial in current domain generalization (DG) problems. In this paper, we introduce a novel multi-modal prompt learning strategy called Transitive vision-language Prompt Learning (TPL) that leverages deep vision prompts to address domain invariance while utilizing language prompts to ensure class separability, coupled with adaptive weighting mechanisms to balance domain invariance and class separability. Unlike existing CLIP-based methods that either use fixed prompts or adapt only single modalities, TPL jointly optimizes both vision and language prompts through a carefully designed transitive learning mechanism in a multi-modal framework. This approach directly addresses the fundamental problem of domain generalization by ensuring that the model effectively captures necessary domain-invariant features, greatly enhancing its ability to adapt and perform robustly in unfamiliar or new domains. Extensive experiments demonstrate that deep vision prompts effectively extract domain-invariant features, significantly improving the generalization ability of deep models and achieving state-of-the-art performance on three datasets.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 2","pages":"1241-1255"},"PeriodicalIF":6.0,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148512258","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

Revolutionizing Cardiovascular Risk Prediction: IoT-Enhanced Retinal Imaging With Deep Learning 革命性的心血管风险预测：物联网增强视网膜成像与深度学习

IF 6 3区计算机科学

IEEE Transactions on Emerging Topics in Computational Intelligence Pub Date : 2026-04-01 Epub Date: 2025-12-23 DOI: 10.1109/TETCI.2025.3641666

Youssef Siyah;Younes El Belghiti;Khalid Minaoui;Hanae Belmajdoub;Samir Saoudi

{"title":"Revolutionizing Cardiovascular Risk Prediction: IoT-Enhanced Retinal Imaging With Deep Learning","authors":"Youssef Siyah;Younes El Belghiti;Khalid Minaoui;Hanae Belmajdoub;Samir Saoudi","doi":"10.1109/TETCI.2025.3641666","DOIUrl":"https://doi.org/10.1109/TETCI.2025.3641666","url":null,"abstract":"Cardiovascular diseases (CVDs) remain the leading cause of global mortality, emphasizing the need for accurate and accessible predictive tools for early risk assessment. This study introduces the Hierarchical Residual Fusion Network (HRFN), an advanced deep learning framework designed to estimate 10-year CVD risk using paired retinal images. The proposed architecture leverages residual learning and dual-pooling mechanisms to extract complex vascular features that reflect systemic cardiovascular health. Trained and validated on the ODIR-5 K dataset, HRFN achieves a coefficient of determination of 0.7169, surpassing traditional models in predictive accuracy. Designed for clinical applicability, the framework ensures precise and interpretable risk estimations to support early medical interventions. Additionally, this study envisions future integration with Internet of Things (IoT)-enabled medical systems, enabling real-time data acquisition and analysis for scalable deployment. By combining AI-driven predictive modeling with emerging healthcare technologies, this approach contributes to advancing personalized medicine and improving global cardiovascular health outcomes.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 2","pages":"1918-1931"},"PeriodicalIF":6.0,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148504282","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

Semi-Supervised Hyperspectral Adaptive Band Selection Based on Evolutionary Multitasking 基于进化多任务的半监督高光谱自适应波段选择

IF 6 3区计算机科学

IEEE Transactions on Emerging Topics in Computational Intelligence Pub Date : 2026-04-01 Epub Date: 2025-07-10 DOI: 10.1109/TETCI.2025.3582842

Jiao Shi;Nan Zhang;Yu Lei;A. K. Qin;Maoguo Gong

{"title":"Semi-Supervised Hyperspectral Adaptive Band Selection Based on Evolutionary Multitasking","authors":"Jiao Shi;Nan Zhang;Yu Lei;A. K. Qin;Maoguo Gong","doi":"10.1109/TETCI.2025.3582842","DOIUrl":"https://doi.org/10.1109/TETCI.2025.3582842","url":null,"abstract":"Band selection (BS) is commonly utilized to reduce the dimensionality of hyperspectral images (HSI). Some semi-supervised BS methods achieve great classification performance using limited labeled samples. However, most methods rely on a unified discriminant learner to find the band subset, which lacks consideration for the complex feature representation of HSI data. This paper proposes a semi-supervised adaptive BS method based on evolutionary multitasking to effectively explore the diverse intrinsic characteristics. The BS problem is modeled as a multitask optimization problem where multiple BS tasks jointly select optimal band subset based on different discriminant learners. To achieve the collaborative search of these BS tasks, a structure-aware evolutionary multitasking approach is proposed, which generates the individuals with various number of bands by considering the low redundancy and complete spectral structure information, and leverages the specially designed genetic operations to facilitate the population evolution under the structure preservation constraint. A dynamic interaction mechanism is devised to adjust task interaction intensity adaptively to enhance effective knowledge sharing between tasks. Additionally, a discriminability evaluation criterion is proposed to mitigate noisy sample interference by integrating sample reliability and class separability. Experiments have demonstrated that the proposed method effectively explores the complex feature representation of HSI data through joint discriminant learners.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 2","pages":"1878-1891"},"PeriodicalIF":6.0,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148504300","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

ProtoNAS: Accelerating Differentiable Neural Architecture Search via Prototypical Data Pruning ProtoNAS：基于原型数据剪枝的加速可微神经结构搜索

IF 6 3区计算机科学

IEEE Transactions on Emerging Topics in Computational Intelligence Pub Date : 2026-04-01 Epub Date: 2025-11-24 DOI: 10.1109/TETCI.2025.3631626

Liming Shi;Lan Deng;Gaochang Wu;Jingxin Zhang;Jinliang Ding

{"title":"ProtoNAS: Accelerating Differentiable Neural Architecture Search via Prototypical Data Pruning","authors":"Liming Shi;Lan Deng;Gaochang Wu;Jingxin Zhang;Jinliang Ding","doi":"10.1109/TETCI.2025.3631626","DOIUrl":"https://doi.org/10.1109/TETCI.2025.3631626","url":null,"abstract":"In this paper, we take a close look at the efficient differentiable Neural Architecture Search (NAS) problem through the lens of data pruning. A key insight is that the core samples in the pruned dataset can be interpreted as support vectors for networks to make informed decisions. Therefore, we introduce ProtoNAS, an innovative prototypical data pruning method that dynamically and explicitly measures the prototypical distance of each data sample to its class-specific clustering centroid (i.e., prototype). These prototypes are dynamically updated on-the-fly throughout the search phase, enabling flexible data pruning that adjusts with the architectures and network parameters. Compared to conventional cross-entropy-based or Euclidean-based distances, the prototypical distance not only considers the distance between a sample and its class-specific prototype, but also the relative distances to the prototypes of other classes in the embedding space, facilitating more precise score assignment. Furthermore, beyond sample pruning, we introduce a class pruning policy aimed at masking the classes with the lowest average scores. This additional policy is designed to further expedite the NAS without compromising the performance of the final searched architecture. Extensive experiments conducted on NAS-Bench-201 and NAS-Bench-1Shot1 search spaces demonstrate that ProtoNAS can achieve a significant 90% pruning rate with minimal performance trade-offs or even improved performance when applied to DARTS and its variants, such as PC-DARTS and <inline-formula><tex-math>$beta$</tex-math></inline-formula>-DARTS. Compared to State-Of-The-Art (SOTA) data pruning techniques, ProtoNAS achieves a superior balance between search performance and search cost.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 2","pages":"1170-1184"},"PeriodicalIF":6.0,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148508200","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

Heterogeneous Modality Dynamic Trustworthy Fusion Network for Cross-Subject Sleep Stage Classification 跨主体睡眠阶段分类的异构模态动态可信融合网络

IF 6 3区计算机科学

IEEE Transactions on Emerging Topics in Computational Intelligence Pub Date : 2026-04-01 Epub Date: 2026-01-12 DOI: 10.1109/TETCI.2025.3647679

Kun Wang;Qi Zhu;Junyong Zhao;Chuhang Zheng;Wei Shao;Daoqiang Zhang

{"title":"Heterogeneous Modality Dynamic Trustworthy Fusion Network for Cross-Subject Sleep Stage Classification","authors":"Kun Wang;Qi Zhu;Junyong Zhao;Chuhang Zheng;Wei Shao;Daoqiang Zhang","doi":"10.1109/TETCI.2025.3647679","DOIUrl":"https://doi.org/10.1109/TETCI.2025.3647679","url":null,"abstract":"Sleep stage classification plays a pivotal role in clinical practice, significantly enhancing the efficiency of sleep assessment and assisting in the diagnosis of sleep disorders. However, existing sleep stage classification methods are susceptible to heterogeneous multi-modal sleep data and the distribution drift among subjects. Additionally, sleep is a dynamic process, inherently encompassing uncertainty at both temporal and sample levels, which presents challenges for achieving highly trustworthy identification of sleep stages. To address these issues, we propose a Heterogeneous Modality Dynamic Trustworthy fusion network, called HMDT-Net, for cross-subject sleep stage classification. Specifically, we first develop a heterogeneous modality dynamic trustworthy fusion strategy with self-attention and cross-attention mechanisms. It effectively integrates heterogeneous multi-modal data and adaptively estimates uncertainty across different time windows. Second, our approach incorporates adversarial learning to extract a common representation of features across subjects while enhancing the discriminative power of these features. This mitigates the influence of inter-subject variations on the model. Additionally, we introduce a self-paced learning (SPL) mechanism to mitigate the influence of challenging training samples, identified through alignment loss and domain discrimination loss, on the classification model. Finally, we utilize a multi-layer perceptron (MLP) to classify the fused features extracted. Extensive experiments conducted on three multi-modal sleeping datasets containing EEG and EOG data, namely Sleep-EDF-20, Sleep-EDF-78, and SHHS, demonstrate that our proposed method surpasses several state-of-the-art sleep stage classification techniques.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 2","pages":"1537-1549"},"PeriodicalIF":6.0,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148508737","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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

GaussFocus RPN for Remote Sensing Object Detection 用于遥感目标检测的GaussFocus RPN

IF 6 3区计算机科学

IEEE Transactions on Emerging Topics in Computational Intelligence Pub Date : 2026-04-01 Epub Date: 2026-02-16 DOI: 10.1109/TETCI.2026.3654459

Zilong Wang;Zishan Xu;Wei Yang;Wei Chen;Yunyue Li;Hongxian Tian;Jueting Liu;Tingting Xu;Zehua Wang

{"title":"GaussFocus RPN for Remote Sensing Object Detection","authors":"Zilong Wang;Zishan Xu;Wei Yang;Wei Chen;Yunyue Li;Hongxian Tian;Jueting Liu;Tingting Xu;Zehua Wang","doi":"10.1109/TETCI.2026.3654459","DOIUrl":"https://doi.org/10.1109/TETCI.2026.3654459","url":null,"abstract":"Objects in remotesensing imagery are often unevenly distributed and densely clustered, making foreground selection particularly challenging. In two-stage detectors, the Region Proposal Network (RPN) is pivotal, yet proposals from conventional RPNs can be imprecise and redundant, hurting recall and inflating computation.We introduce GaussFocus, a density-guided two-stage detector that builds a unified Gaussian foreground prior from rotated annotations and injects it into proposal generation, feature sampling, and RoI score calibration. Specifically, we derive an image-level Gaussian foreground density map and train a lightweight density branch inside the RPN, where MDeformConv uses regression-guided offsets and density-aware gating to produce features that better align with object geometry and foreground regions. A Gauss Guided Selector further leverages the predicted density map to retain proposals concentrated in foreground-dense areas, reducing background clutter and second-stage computation. In addition, an inference-only Gauss-Guided RoI logit shift calibrates RoI classification scores using the same density prior without retraining, leading to more reliable foreground–background separation and ranking. Across four challenging datasets (including DOTA-v1.0, DOTA-v2.0, and HRSC2016), GaussFocus consistently improves proposal quality and detection accuracy over strong baselines while reducing redundant RoIs and computation.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 2","pages":"1512-1524"},"PeriodicalIF":6.0,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148509743","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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