检索结果-内蒙古大学图书馆

FATCNet : Feature Adaptive Transformer and CNN for Infrared Small Target Detection

IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS 2024年第6期60卷 9231-9246页

作者： Yang, Jing Deng, Shen Zhang, Fuqing Pan, Anning Yang, Yang Yunnan Nor mal Univ Sch Informat Sci & Technol Lab Pattern Recognit & Artificial Intelligence Kunming 650500 Peoples R China Yunnan Normal Univ Sch Phys & Elect Informat Kunming 650500 Peoples R China Baoshan Univ Sch Big Data Baoshan 678000 Peoples R China

Infrared small target detection is the process of separating targets from complex and blurry backgrounds, with wide applications in rescue, maritime guidance, and other fields. However, traditional deep learning methods that rely solely on convolutional neural network (CNN)-based methods are prone to losing targets as the convolution deepens, resulting in missed detections when segmenting small infrared targets. Similarly, because of the unique characteristics of infrared small targets, Transformer-based methods lack the supplement of convolutional local detail features, making it difficult to achieve good results. To address these issues, we propose the feature adaptive Transformer and CNN network (FATCNet): a dual branch decoder structure network that utilizes CNN and Transformer to obtain local and global features in parallel. Meanwhile, we propose the attention guided adaptive fusion block to better integrate global and local feature information, eliminate semantic gaps, and dynamically fuse features based on network learning. The network encoder incorporates the hierarchical attention module, which leverages the distinct characteristics of various hierarchical feature maps to enhance features, thereby mitigating the risk of losing crucial details pertaining to infrared small targets. In addition, we have constructed a bird infrared dataset, named BSIRST_v2, which comprises 1000 authentic images with diverse target shapes, sizes, and cluttered backgrounds, and the dataset includes meticulous pixel-level annotations. Our approach's efficacy has been validated through experiments conducted on both public datasets and BSIRST_v2. In comparison to other state-of-the-art methods, our FATCNet outperforms in terms of probability of detection, F1 score, and intersection over union.

关键词： Transformers Object detection Shape Feature extraction Semantics Convolution Aerospace and electronic systems Attention guided adaptive fusion block (AGAF) dual encoder hierarchical attention module (HAM) infrared small target detection (IRSTD)

来源：评论

学校读者我要写书评

暂无评论

Data-Driven full waveform inversion for ultrasonic bone quantitative imaging

引用

NEURAL COMPUTING & APPLICATIONS 2023年第36期35卷 25027-25043页

作者： Suo, Meng Zhang, Dong Yang, Haiqi Yang, Yan Wuhan Univ Sch Phys & Technol Wuhan 430072 Hubei Peoples R China

Full waveform inversion (FWI) has potential for quantitative ultrasound bone imaging, which can provide detailed estimation of bone internal structure. The existing ultrasound FWI methods suffer from cycle skipping and local minima and high computational costs, which limits their clinical application potential. In recent years, data-driven method for solving the problem of inversion has become a very novel approach, but we haven't seen it in the application of ultrasonic bone imaging studies. Herein, we develop an improved dual-encoder-based Unet with high-frequency feature enhancement (DEFE-Unet) for ultrasonic bone quantitative imaging to obtain more detailed information by inserting high-frequency feature extractors related to bone microstructure into the network. This method can obtain more detailed and fine results than FWI and avoid the extremely high computational cost of FWI dual-parameter modeling. We designed two types of bone model datasets based on Micro_CT images to train and test the proposed network architecture. Experimental results show that the DEFE-Unet algorithm proposed in this paper obtains competitive results with less computational cost compared to the FWI algorithm. Compared with other baseline machine learning architectures, such as Unet and attention-Unet, the proposed algorithm achieves significant performance improvement with a small computational cost increment. The results show that the DEFE-Unet has better potential in clinical detection of bone disease.

关键词： Full waveform inversion (FWI) Bone quantitative imaging Data-driven method dual encoder

来源：评论

学校读者我要写书评

暂无评论

Strategy of the Negative Sampling for Training Retrieval-Based Dialogue Systems

Strategy of the Negative Sampling for Training Retrieval-Bas...

引用

IEEE International Conference on Pervasive Computing and Communications (PerCom)

作者： Nugmanova, Aigul Smirnov, Andrei Lavrentyeva, Galina Chernykh, Irina ITMO Univ Speech Technol Ctr St Petersburg Russia Speech Technol Ctr St Petersburg Russia ITMO Univ STC Innovat Ltd St Petersburg Russia

ISBN: (纸本)9781538691519

The article describes the new approach for quality improvement of automated dialogue systems for customer support service. The analysis produced in the paper demonstrates the dependency of the quality of the retrieval-based dialogue system on the choice of negative responses required for training the system. The proposed approach implies choosing the negative samples according to the distribution of responses in the training set. In this implementation, the negative samples are randomly chosen from the original response distribution and from the "artificial" distribution of negative responses, such as uniform distribution or the distribution obtained by transformation of the original one. The results obtained for the implemented systems and reported in this paper confirm the significant improvement of automated dialogue systems quality in case of using the negative responses from transformed distribution.

关键词： dialogue systems negative sampling dual encoder retrieval-based dialogue systems

来源：评论

学校读者我要写书评

暂无评论

Surprise: Result List Truncation via Extreme Value Theory 23

Surprise: Result List Truncation via Extreme Value Theory

引用

46th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR)

作者： Bahri, Dara Zheng, Che Tay, Yi Metzler, Donald Tomkins, Andrew Google Res Mountain View CA 94043 USA

ISBN: (纸本)9781450394086

Work in information retrieval has largely been centered around ranking and relevance: given a query, return some number of results ordered by relevance to the user. The problem of result list truncation, or where to truncate the ranked list of results, however, has received less attention despite being crucial in a variety of applications. Such truncation is a balancing act between the overall relevance, or usefulness of the results, with the user cost of processing more results. Result list truncation can be challenging because relevance scores are often not well-calibrated. This is particularly true in large-scale IR systems where documents and queries are embedded in the same metric space and a query's nearest document neighbors are returned during inference. Here, relevance is inversely proportional to the distance between the query and candidate document, but what distance constitutes relevance varies from query to query and changes dynamically as more documents are added to the index. In this work, we propose Surprise scoring, a statistical method that leverages the Generalized Pareto Distribution that arises in extreme value theory to produce interpretable and calibrated relevance scores at query time using nothing more than the ranked scores. We demonstrate its effectiveness on the result list truncation task across image, text, and IR datasets and compare it to both classical and recent baselines. We draw connections to hypothesis testing and p-values.

关键词： nearest neighbor dual encoder result list truncation extreme value theory

来源：评论

学校读者我要写书评

暂无评论

Explicit and Implicit Semantic Ranking Framework

Explicit and Implicit Semantic Ranking Framework

引用

32nd World Wide Web Conference (WWW)

作者： Zhu, Xiaofeng Lin, Thomas Anand, Vishal Calderwood, Matthew Clausen-Brown, Eric Lueck, Gord Yim, Wen-wai Wu, Cheng Microsoft Redmond WA 98052 USA Nuance Commun Burlington MA USA

ISBN: (纸本)9781450394161

The core challenge in numerous real-world applications is to match an inquiry to the best document from a mutable and fnite set of candidates. Existing industry solutions, especially latency-constrained services, often rely on similarity algorithms that sacrifce quality for speed. In this paper we introduce a generic semantic learning-to-rank framework, Self-training Semantic Cross-attention Ranking (sRank). This transformer-based framework uses linear pairwise loss with mutable training batch sizes and achieves quality gains and high efciency, and has been applied efectively to show gains on two industry tasks at Microsoft over real-world large-scale data sets: Smart Reply (SR) and Ambient Clinical Intelligence (ACI). In Smart Reply, sRank assists live customers with technical support by selecting the best reply from predefned solutions based on consumer and support agent messages. It achieves 11.7% gain in ofine top-one accuracy on the SR task over the previous system, and has enabled 38.7% time reduction in composing messages in telemetry recorded since its general release in January 2021. In the ACI task, sRank selects relevant historical physician templates that serve as guidance for a text summarization model to generate higher quality medical notes. It achieves 35.5% top-one accuracy gain, along with 46% relative ROUGE-L gain in generated medical notes.

关键词： semantic search pairwise learning to rank smart reply text summarization transformer dual encoder

来源：评论

学校读者我要写书评

暂无评论

Learning Semantics-Grounded Vocabulary Representation for Video-Text Retrieval 23

Learning Semantics-Grounded Vocabulary Representation for Vi...

引用

31st ACM International Conference on Multimedia (MM)

作者： Shi, Yaya Liu, Haowei Xu, Haiyang Ma, Zongyang Ye, Qinghao Hu, Anwen Yan, Ming Zhang, Ji Huang, Fei Yuan, Chunfeng Li, Bing Hu, Weiming Zha, Zheng-Jun Univ Sci & Technol China Hefei Peoples R China Chinese Acad Sci Inst Automat MAIS Beijing Peoples R China Univ Chinese Acad Sci Sch AI Beijing Peoples R China Alibaba Grp DAMO Acad Hangzhou Peoples R China ShanghaiTech Univ Sch Informat Sci & Technol Beijing Peoples R China

ISBN: (纸本)9798400701085

Previous dual-encoder pre-training methods for video-text retrieval employ contrastive learning for cross-modal alignment in a latent space. However, such learned latent spaces often result in modality gap problem [26]. In this paper, we introduce a novel SemVTR framework designed to learn semantics-grounded video-text representations in a vocabulary space, in which each dimension corresponds to a semantic concept represented by a word. The representation is obtained by grounding video and text into semantically-related dimensions with high activation values. As video-text pairs share grounded dimensions, their vocabulary representations are expected to cluster together and thus alleviate modality gap problem. So, the crux of our method lies in grounding video and text into vocabulary space. Specifically, we propose a Multi-Granularity Video Semantics Grounding approach and a Textual Semantics Preserving training strategy. The visualization illustrates that SemVTR obtains semantics-gronded vocabulary representation and also alleviates the modality gap problem. SemVTR significantly outperforms existing methods on four video-text retrieval benchmarks.

关键词： video-text retrieval dual encoder modality gap vocabulary space semantics-grounded representation

来源：评论

学校读者我要写书评

暂无评论

BTS-ST: Swin transformer network for segmentation and classification of multimodality breast cancer images

引用

KNOWLEDGE-BASED SYSTEMS 2023年第1期267卷

作者： Iqbal, Ahmed Sharif, Muhammad COMSATS Univ Islamabad Dept Comp Sci Wah Campus Islamabad Pakistan

Breast cancer is considered the most commonly diagnosed cancer globally and falls second to lung cancer. For the early detection of breast tumors in women, breast cancer analysis using Ultrasound, Mammography, and MRI modalities as the initial screening process. Due to the random variation, irregular shapes, and blurred boundaries of tumor regions, the accurate segmentation of breast tumors is still a tricky task. The existing convolutional neural networks (CNNs) inherit their limitation by extracting global context information and, in most cases, proved less efficient in obtaining satisfactory results. As a solution, we proposed the BTS-ST network, a novel solution for breast tumor segmentation and classification that Swin-Transformer (ST) inspires. The BTS-ST network incorporates Swin-Transformer into traditional CNNs-based U-Net to improve global modeling capabilities. To improve the feature representation capability of irregularly shaped tumors, we first introduced a Spatial Interaction block (SIB), encoding spatial knowledge in the Swin Transformer block by developing pixel-level correlation. The segmentation accuracy of small-scale tumor regions is increased by building a Feature Compression block (FCB) to prevent information loss and compress smaller-scale features in patch token down sampling of Swin-Transformer. Finally, a Relationship Aggregation block (RAB) is developed as a bridge between dual encoders to combine global dependencies from Swin-Transformer into the features from CNN hierarchically. Extensive experiments are performed on breast tumor segmentation and classification tasks using multimodality Ultrasound, Mammogram, and MRI-based datasets. The results demonstrate that our proposed solution is comparatively better than other state-of-the-art methods. (c) 2023 Elsevier B.V. All rights reserved.

关键词： dual encoder Swin transformer U -Shaped Network Breast tumor classification Breast tumor segmentation

来源：评论

学校读者我要写书评

暂无评论

DEformer: dual Embedded Transformer for Multivariate Time Series Forecasting

引用

IEEE ACCESS 2024年 12卷 153851-153858页

作者： Kim, Minje Lee, Suwon Choi, Sang-Min Gyeongsang Natl Univ Dept Comp Sci & Engn Jinju Si 52828 South Korea Gyeongsang Natl Univ Res Inst Nat Sci Jinju Si 52828 South Korea

Deep learning models have significantly addressed the challenges of multivariate time series forecasting. Recently, Transformer-based models which have primarily focused on either temporal or inter-variate (spatial) dependencies have demonstrated exceptional performance. These models decide whether to embed multivariate time series data temporally or spatially. Hence, we propose the dual embedded transformer (DEformer) which simultaneously considers both temporal and spatial dependencies. Our model enables capturing these dependencies independently through two distinct encoders. The temporal encoder, which processes the entire time series of each variate, is designed to learn temporal dependency. Conversely, the spatial encoder, which processes the multivariate data at each time step, is optimized to capture spatial dependency. Both encoders share an identical architecture, comprising an attention mechanism to model correlations and a feed-forward network to enhance feature representation. Through empirical studies on challenging real-world datasets, we not only demonstrate that our method can outperforms state-of-the-art approaches, but also prove the performance of independently embedding both dependencies through ablation study.

关键词： Deep learning multivariate time series forecasting transformer dual encoder Deep learning multivariate time series forecasting transformer dual encoder

来源：评论

学校读者我要写书评

暂无评论

Chinese Short Text Summary Generation Model Combining Global and Local Information

Chinese Short Text Summary Generation Model Combining Global...

引用

2018 International Conference on Network, Communication, Computer Engineering (NCCE 2018)

作者： Guanqin Chen School of Computer Guangdong University of Technology

ISBN: (纸本)9781510871076

Short text comprehension summary generation is currently a hot *** this paper,we improve the attention mechanism under the framework of encoder-decoder and proposes a comprehensible short text abstract generation model that integrates the global and local semantic *** model consists of a dual encoder and a *** dual encoder structure can combine the global and local semantic information and fully obtain the abstract features of the original *** the improved mechanism can adaptively combine all information of short text to provide the input with summary characteristics for the decoder,so that the decoder can more accurately focus on the core content of the source *** this paper,LCSTS dataset is used to train and test the *** experimental results show that compared with the Seq2Seq and Seq2Seq with standard attention models,the proposed method can produce high-quality summary which consists of less repetitive words and performs better evaluation value in ROUGE.

关键词： dual encoder Attention mechanism Global information Local information Text summary Seq2Seq

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：