检索结果-内蒙古大学图书馆

Bottom-Up transformer Reasoning Network for Text-Image Retrieval 1

29th International Conference on Neural Information Processing

作者： Yang, Zonghao Zhou, Yue Chen, Ao Shanghai Jiao Tong Univ Shanghai Peoples R China

ISBN: (数字)9789819916450

ISBN: (纸本)9789819916443;9789819916450

Image-text retrieval is a complicated and challenging task in the cross-modality area, and lots of experiments have made great progress. Most existing researches process images and text in one pipeline or are highly entangled, which is not practical and human-friendly in the real world. Moreover, the image regions extracted by Faster-RCNN are highly over-sampled in the image pipeline, which causes ambiguities for the extracted visual embeddings. From this point of view, we introduce the Bottom-up transformer Reasoning Network (BTRN). Our method is built upon the transformer encoders to process the image and text separately. We also embed the tag information generated by Faster-RCNN to strengthen the connection between the two modalities. Recall at K and normalized discounted cumulative gain metric (NDCG) metrics are used to evaluate our model. Through various experiments, we prove our model can reach state-of-the-art results.

关键词： Text-image retrieval transformer encoder Tag embedding Double-Stream architecture

来源：评论

学校读者我要写书评

暂无评论

SFYOLO: A Lightweight and Effective Network Based on Space-Friendly Aggregation Perception for Pear Detection 1

引用

17th International Conference on Green, Pervasive, and Cloud Computing (GPC)

作者： Li, Yipu Rao, Yuan Jin, Xiu Jiang, Zhaohui Liu, Lu Wang, Yuwei Anhui Agr Univ Coll Informat & Comp Sci Hefei 230036 Peoples R China Minist Agr & Rural Affairs Key Lab Agr Sensors Hefei 230036 Peoples R China Anhui Prov Key Lab Smart Agr Technol & Equipment Hefei 230036 Peoples R China

ISBN: (数字)9783031261183

ISBN: (纸本)9783031261176;9783031261183

It is always challenging for efficiently conducting accurate detection of small and occluded pears in modern orchards. In the past few years, the aforementioned detection tasks remained unsolved though lots of researchers attempted to optimize the adaption of background noise and viewpoints, particularly compliant models suitable for simultaneously detecting small and occluded pears with low computational cost and memory usage. In this paper, we proposed a lightweight and effective object detection network called as SFYOLO based on space-friendly aggregation perception. Specifically, a novel space-friendly attention mechanism was proposed for implementing the aggregate perception of spatial domain and channel domain. Afterwards, an improved space-friendly transformer encoder was put forward for enhancing the ability of information exchange between channels. Finally, the decoupled anchor-free detectors were used as the head to improve the adaptability of the network. The mean Average Precision (mAP) for in-field pears was 93.12% in SFYOLO, which was increased by 2.03% compared with original YOLOv5s. Additional experiments and comparison were carried out considering newly proposed YOLOv6 and YOLOv7 that aimed at optimizing the detection accuracy and speed. Results verified that small and occluded pears could be detected fast and accurately by the competitive SFYOLO network under various viewpoints for further orchard yield estimation and development of pear picking system.

关键词： YOLOv5s Object detection Visual attention mechanism transformer encoder Aggregate perception

来源：评论

学校读者我要写书评

暂无评论

TsP-Tran: Two-Stage Pure transformer for Multi-Label Image Retrieval 23

TsP-Tran: Two-Stage Pure Transformer for Multi-Label Image R...

引用

ACM International Conference on Multimedia Retrieval (ICMR)

作者： Li, Ying Guan, Chunming Gao, Jiaquan Nanjing Normal Univ Nanjing Peoples R China

ISBN: (纸本)9798400701788

Image retrieval aims to find similar images given the query. Most of existing retrieval works are based on the pre-trained model of single-label image classification. In practice, the query usually contains more than one instance, and the single label is far from enough for fully depicting the attributes of an open-world image. Due to the complicated similarity relationships between multiple semantics, the multi-label image retrieval task is not so well solved as the single-label task. In this work, we propose a two-stage pure transformer model for multi-label image retrieval, which leverages a transformer encoder to exploit the complex dependencies among visual features and labels. Except for the transformer encoder, the image feature embedding module is also based on transformer, so that the optimal model weights could be learned in an end-to-end manner. To be specific, inputs of the transformer encoder mainly consist of an Vision transformer branch and a label embedding branch, which generates suitable image features and label descriptions, respectively. Given an input set of visual features and text labels, the developed transformer encoder could be accordingly optimized in the training stage with compressed multi-label output layer. In order to obtain sufficient outputs to accurately find images containing similar semantics with the query from the database, we adjust the network by removing the last fully connected layer in the retrieval stage. Specially, images and labels are used for the training stage in a randomly masked manner to enhance the model performance, and no labels are visible in the content-based image retrieval stage. Comprehensive experiments are performed on three multi-label datasets including MS-COCO, NUS-WIDE and VOC2007, demonstrating promising results of our proposed method against the state-of-the-arts for multi-label image retrieval.

关键词： image retrieval multi-label transformer encoder

来源：评论

学校读者我要写书评

暂无评论

A comparative study on deep learning models for text classification of unstructured medical notes with various levels of class imbalance

引用

BMC MEDICAL RESEARCH METHODOLOGY 2022年第1期22卷 1-12页

作者： Lu, Hongxia Ehwerhemuepha, Louis Rakovski, Cyril Chapman Univ Schmid Coll Sci & Technol 1 Univ Dr Orange CA 92866 USA CHOC Orange CA 92868 USA

Background Discharge medical notes written by physicians contain important information about the health condition of patients. Many deep learning algorithms have been successfully applied to extract important information from unstructured medical notes data that can entail subsequent actionable results in the medical domain. This study aims to explore the model performance of various deep learning algorithms in text classification tasks on medical notes with respect to different disease class imbalance scenarios. Methods In this study, we employed seven artificial intelligence models, a CNN (Convolutional Neural Network), a transformer encoder, a pretrained BERT (Bidirectional encoder Representations from transformers), and four typical sequence neural networks models, namely, RNN (Recurrent Neural Network), GRU (Gated Recurrent Unit), LSTM (Long Short-Term Memory), and Bi-LSTM (Bi-directional Long Short-Term Memory) to classify the presence or absence of 16 disease conditions from patients' discharge summary notes. We analyzed this question as a composition of 16 binary separate classification problems. The model performance of the seven models on each of the 16 datasets with various levels of imbalance between classes were compared in terms of AUC-ROC (Area Under the Curve of the Receiver Operating Characteristic), AUC-PR (Area Under the Curve of Precision and Recall), F1 Score, and Balanced Accuracy as well as the training time. The model performances were also compared in combination with different word embedding approaches (GloVe, BioWordVec, and no pre-trained word embeddings). Results The analyses of these 16 binary classification problems showed that the transformer encoder model performs the best in nearly all scenarios. In addition, when the disease prevalence is close to or greater than 50%, the Convolutional Neural Network model achieved a comparable performance to the transformer encoder, and its training time was 17.6% shorter than the second fastest m

关键词： Medical notes Text classification BERT CNN Deep learning Embedding transformer encoder

来源：评论

学校读者我要写书评

暂无评论

High-Resolution Range Profile Recognition of Midcourse Ballistic Targets Based on AMTrans 2

High-Resolution Range Profile Recognition of Midcourse Balli...

引用

2nd International Conference on Image Processing, Computer Vision and Machine Learning, ICICML 2023

作者： Liu, Yushan Wang, Jian Cui, Dekun Peng, Shuyang Air Force Engineering University College of Air Missile Defense Xi'an China

ISBN: (纸本)9798350331417

To address the problem of high-precision time-series recognition of high-resolution range profiles (HRRP) of ballistic midcourse targets, we propose a HRRP recognition method based on the multivariate time-series learning approach of transformer encoder to further improve the recognition efficiency and accuracy. The AMTrans model is based on a multi-headed attention mechanism which can take into account multiple correlations among the input elements and adaptively adjust the attention weight in different distance units. The experimental results show that this proposed method has significant performance advantages over other methods and has better robustness. © 2023 IEEE.

关键词： high resolution range profile label smoothing recognition of midcourse ballistic missile target transformer encoder

来源：评论

学校读者我要写书评

暂无评论

Computer Vision and Voice Assisted Image Captioning Framework for Visually Impaired Individuals using Deep Learning Approach 4

Computer Vision and Voice Assisted Image Captioning Framewor...

引用

4th IEEE Global Conference for Advancement in Technology, GCAT 2023

作者： Safiya, K.M. Pandian, R. Chennai India

ISBN: (纸本)9798350305258

Blind or visually challenged have considerable obstacles when obtaining visual material, restricting their capacity to derive meaningful information from pictures. This research proposing a novel framework combining computer vision with voice-based image captioning using deep learning techniques. The proposed system employs VGGNet-16 model for photo processing and long short-term memory networks (LSTMs) for natural language processing. The models used in this research were trained on Flickr8k, Flickr30k, and a bespoke dataset. Subsequently, these models were deployed on a Raspberry Pi 4B single-board computer that was equipped with a graphics processing unit. A two-fold approach is used, whereby the first phase entails exposing the input image to preprocessing using a pre-established VGGNet-16 model. This methodology enables the retrieval of relevant visual attributes, therefore capturing the intrinsic semantic content of the picture. The aforementioned attributes are fed into a language model based on LSTM to generate explanatory captions. To enhance communication with visually impaired people, the generated captions are converted into audible speech using text-to-speech synthesis technology. The effectiveness of the proposed architecture is evaluated via extensive experiments using benchmark datasets and real-time images obtained with a NoIR camera unit. The generated captions are evaluated using quantitative assessment criteria, namely BLEU and ROUGE scores. The developed VGGNet16 model has exceptional performance in terms of accuracy (95.620%), precision (96.928%), recall (87.879%), and F1 score (92.182%). The results suggest that using computer vision in conjunction with voice-based image captioning offers a promising solution for addressing the challenges faced by those with visual impairments in perceiving and comprehending visual content. © 2023 IEEE.

关键词： computer vision image captioning LSTM self-attention transformer encoder VGGNet-16 visually impaired

来源：评论

学校读者我要写书评

暂无评论

A Multi-Task Joint Learning Model Based on transformer and Customized Gate Control for Predicting Remaining Useful Life and Health Status of Tools

引用

SENSORS 2024年第13期24卷 4117页

作者： Hou, Chunming Zheng, Liaomo Chinese Acad Sci Shenyang Inst Comp Technol Shenyang 110168 Peoples R China Univ Chinese Acad Sci Beijing 100049 Peoples R China Shenyang CASNC Technol Co Ltd Shenyang 110168 Peoples R China

Previous studies have primarily focused on predicting the remaining useful life (RUL) of tools as an independent process. However, the RUL of a tool is closely related to its wear stage. In light of this, a multi-task joint learning model based on a transformer encoder and customized gate control (TECGC) is proposed for simultaneous prediction of tool RUL and tool wear stages. Specifically, the transformer encoder is employed as the backbone of the TECGC model for extracting shared features from the original data. The customized gate control (CGC) is utilized to extract task-specific features relevant to tool RUL prediction and tool wear stage and shared features. Finally, by integrating these components, the tool RUL and the tool wear stage can be predicted simultaneously by the TECGC model. In addition, a dynamic adaptive multi-task learning loss function is proposed for the model's training to enhance its calculation efficiency. This approach avoids unsatisfactory prediction performance of the model caused by unreasonable selection of trade-off parameters of the loss function. The effectiveness of the TECGC model is evaluated using the PHM2010 dataset. The results demonstrate its capability to accurately predict tool RUL and tool wear stages.

关键词： remaining useful life wear stage multi-task joint learning dynamic adaptive transformer encoder

来源：评论

学校读者我要写书评

暂无评论

UnifiedCut: A Simple and Efficient Neural Model for Thai, Burmese and Khmer Word Segmentation

引用

APPLIED SCIENCES-BASEL 2024年第23期14卷 11435页

作者： Wen, Yonghua Xian, Yantuan Wang, Yuehan Yu, Zhengtao Kunming Univ Sci & Technol Fac Informat Engn & Automat 727 South Jingming Rd Kunming 650500 Peoples R China Yunnan Key Lab Artificial Intelligence 727 South Jingming Rd Kunming 650500 Yunnan Peoples R China

Word segmentation is a critical task in natural language processing for southeast Asian Abugida languages, including Thai, Burmese, and Khmer. Existing approaches demonstrate that models using fixed-length windowed context inputs can achieve high segmentation accuracy;however, they often rely on low-level character features or language-specific preprocessing. Character-based methods can limit feature learning, while language-specific features add complexity due to specialized preprocessing requirements. This paper introduces UnifiedCut, which is a neural model that leverages multiple n-grams within a windowed multi-head attention mechanism. This design captures segmentation features from local contexts and multi-perspective n-gram inputs, enhancing generalization and recall, particularly for out-of-vocabulary words. Compared to CNN- and RNN-based approaches, UnifiedCut's multi-head attention enables finer-grained feature extraction and greater parallelism, resulting in a faster, more scalable solution. Comprehensive experiments on public datasets for Thai, Burmese, and Khmer show that UnifiedCutachieves state-of-the-art performance in word segmentation.

关键词： word segmentation transformer encoder multiple n-grams Thai Burmese Khmer

来源：评论

学校读者我要写书评

暂无评论

Fault diagnosis of induction motor in the cooling water supply system using a multi-channel data fusion transformer with limited sample conditions

引用

MEASUREMENT SCIENCE AND TECHNOLOGY 2024年第10期35卷 106103-106103页

作者： Niu, Tianyu Wang, Jianyu Miao, Qiang Sichuan Univ Coll Elect Engn Chengdu 610065 Peoples R China

Induction motors play a vital role in the cooling water supply system of hydropower facilities. However, it is not feasible to collect sufficient fault samples in a hydropower station. The scarcity of labeled samples poses a challenge in developing powerful diagnostic models with high classification accuracy. To address this challenge, this paper proposes a multi-channel data fusion strategy based on a transformer for feature enhancement. Initially, the original signals are transferred into non-overlapping single-channel data patches to preserve correlation features across different channels. Next, temporal and spatial attention modules are applied to process the data patches, which can learn and fuse temporal and spatial information, respectively. Subsequently, the data patches are embedded to retain position information and represent fault-related features through class embedding, which are further processed by a transformer encoder with self-attention mechanisms. Finally, the classification task is achieved by using a multilayer perceptron layer connected to the class embedding. While dealing with limited training samples, the proposed method can learn robust features that are beneficial to improve the fault diagnosis ability of induction motors. The comparison of the proposed method with three basic models and two improved methods demonstrates the superiority of the proposed method in accuracy and feature clustering performance under limited sample conditions. In addition, ablation experiments demonstrate the necessity of each module in the proposed method.

关键词： fault diagnosis induction motor limited samples data fusion transformer encoder

来源：评论

学校读者我要写书评

暂无评论

Development and Analysis of Convolutional Neural Network based Accurate Speech Emotion Recognition Models 19

Development and Analysis of Convolutional Neural Network bas...

引用

19th IEEE-India-Council International Conference (INDICON)

作者： Vijayan, Divya M. Arun, A., V Ganeshnath, R. Nath, Ajay S. A. Roy, Rajesh Cherian Model Engn Coll Kochi Dept Elect Ernakulam India TKM Coll Engn Dept Elect Kollam India Muthoot Inst Technol & Sci Dept Comp Sci Ernakulam India

ISBN: (纸本)9781665473507

Automatic speech recognition is a major topic in artificial intelligence and machine learning, with the intent of developing machines that can communicate with humans through speech. Recently, with the emergence of the deep-learning paradigm, end to-end models that extract features and train directly from the raw speech signal have been developed. With the goal of more precisely classifying emotions from speech, this paper presents a comparative analysis on two deep-learning architectures that improves on the models available in the literature in terms of accuracy. Using a combined CNN-LSTM architecture and a CNN-transformer encoder architecture, this work analyses the complete deep learning strategy for extracting distinct spatial and temporal features and classifying the emotions from speech. Experiments are carried out on the RAVDESS dataset. The CNN-transformer encoder network achieves high accuracy 82% in these networks, while the CNN-LSTM network achieves 74%.

关键词： Speech Emotion Recognition CNN LSTM transformer encoder Accuracy RAVDESS dataset

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：