Filters








2,286 Hits in 5.7 sec

Deep Imbalanced Attribute Classification using Visual Attention Aggregation [article]

Nikolaos Sarafianos and Xiang Xu and Ioannis A. Kakadiaris
2018 arXiv   pre-print
With that in mind, we propose an effective method that extracts and aggregates visual attention masks at different scales.  ...  For many computer vision applications, such as image description and human identification, recognizing the visual attributes of humans is an essential yet challenging problem.  ...  We argue that a solution to the deep imbalanced attribute classification problem should: (i) extract discriminative information, (ii) leverage visual information that is specific for each attribute, and  ... 
arXiv:1807.03903v2 fatcat:76gup5jrazfjlofuh2jjtv3dam

Deep Imbalanced Attribute Classification Using Visual Attention Aggregation [chapter]

Nikolaos Sarafianos, Xiang Xu, Ioannis A. Kakadiaris
2018 Lecture Notes in Computer Science  
With that in mind, we propose an effective method that extracts and aggregates visual attention masks at different scales.  ...  For many computer vision applications, such as image description and human identification, recognizing the visual attributes of humans is an essential yet challenging problem.  ...  We argue that a solution to the deep imbalanced attribute classification problem should: (i) extract discriminative information, (ii) leverage visual information that is specific for each attribute, and  ... 
doi:10.1007/978-3-030-01252-6_42 fatcat:jft7ls2jxvhuflyczq7c6dhzgu

Data Augmentation Imbalance For Imbalanced Attribute Classification [article]

Yang Hu, Xiaying Bai, Pan Zhou, Fanhua Shang, Shengmei Shen
2020 arXiv   pre-print
Pedestrian attribute recognition is an important multi-label classification problem.  ...  Extensive empirical evidence shows that our DAI algorithm achieves state-of-the-art results, based on pedestrian attribute datasets, i.e. standard PA-100K and PETA datasets.  ...  Commonly-used practice such as attention model [9] extract and aggregate visual attention masks at different scales.  ... 
arXiv:2004.13628v3 fatcat:y74gv3odlfcbhd4s4epvs2kud4

SANet:Superpixel Attention Network for Skin Lesion Attributes Detection [article]

Xinzi He, Baiying Lei, Tianfu Wang
2019 arXiv   pre-print
To solve these problems, we propose a deep learning framework named superpixel attention network (SANet).  ...  However, unlike lesion segmentation and melenoma classification, there are few deep learning methods and literatures focusing on this task.  ...  The proposed model can also reformulate super-pixel classification problems as the superpixel segmentations and superpixel attention mechanism is leveraged to enhance the discriminative of visual features  ... 
arXiv:1910.08995v1 fatcat:g7wsfakianai3k6ji5kbpnpive

SiamAPN++: Siamese Attentional Aggregation Network for Real-Time UAV Tracking [article]

Ziang Cao, Changhong Fu, Junjie Ye, Bowen Li, Yiming Li
2021 arXiv   pre-print
By virtue of the attention mechanism, we conduct a special attentional aggregation network (AAN) consisting of self-AAN and cross-AAN for raising the representation ability of features eventually.  ...  To this concern, in this paper, a novel attentional Siamese tracker (SiamAPN++) is proposed for real-time UAV tracking.  ...  In visual tracking, RASNet [16] adopts three attention branches to adapt the model without updating the model online in visual tracking.  ... 
arXiv:2106.08816v2 fatcat:tdkwspfkm5fibpa3urirt2oegy

Graphsite: Ligand-binding site classification using Deep Graph Neural Network [article]

Wentao Shi, Manali Singha, Limeng Pu, J. Ramanujam, Michal Brylinski
2021 bioRxiv   pre-print
Deep learning is a modern artificial intelligence technology. It utilizes deep neural networks to handle complex tasks such as image classification and language translation.  ...  Graph neural networks (GNNs) are deep learning models that operate on graphs.  ...  Classification performance Two GNN-based methods are evaluated: GraphSite and GIN. GIN uses a sum aggregator, so the edge attributes are ignored.  ... 
doi:10.1101/2021.12.06.471420 fatcat:ctadiun7bfewnipdlm7oe55evq

Intrusion Detection of Imbalanced Network Traffic Based on Machine Learning and Deep Learning

Lan Liu, Pengcheng Wang, Jun Lin, Langzhou Liu
2020 IEEE Access  
INDEX TERMS IDS, imbalanced network traffic, machine learning, deep learning, CSE-CIC-IDS2018.  ...  This paper researches machine learning and deep learning for intrusion detection in imbalanced network traffic.  ...  We use t-SNE to visualize the NSL-KDD and CSE-CIC-IDS2018 by dimensionality reduction [40] .  ... 
doi:10.1109/access.2020.3048198 fatcat:tln3mv5u4nadzkz4alexcw65uu

Reinforced Pedestrian Attribute Recognition with Group Optimization Reward [article]

Zhong Ji, Zhenfei Hu, Yaodong Wang, Shengjia Li
2022 arXiv   pre-print
Then we employ an agent to recognize each group of attributes, which is trained with Deep Q-learning algorithm.  ...  Two key challenges in PAR include complex alignment relations between images and attributes, and imbalanced data distribution. Existing approaches usually formulate PAR as a recognition task.  ...  Although the visual attention mechanism has been successfully applied to PAR, the complexity of pedestrian images makes attention masks fail to obtain the position of a specific attribute.  ... 
arXiv:2205.14042v1 fatcat:4pzdeqexe5b77fb5fhka2f7oi4

Network Embedding with Completely-imbalanced Labels

Zheng Wang, Xiaojun Ye, Chaokun Wang, Jian Cui, Philip Yu
2020 IEEE Transactions on Knowledge and Data Engineering  
However, existing semi-supervised methods would get unappealing results in the completely-imbalanced label setting where some classes have no labeled nodes at all.  ...  Specifically, to benefit from the completely-imbalanced labels, RSDNE guarantees both intra-class similarity and inter-class dissimilarity in an approximate way.  ...  For example, we can define some attributes like "wing", "climb" or "tail" for animals. Then we can train attribute recognizers using images and attribute information from seen classes.  ... 
doi:10.1109/tkde.2020.2971490 fatcat:fbgwesqkw5cq7gand5z6wva2nq

Deep Multi-Scale Resemblance Network for the Sub-class Differentiation of Adrenal Masses on Computed Tomography Images [article]

Lei Bi, Jinman Kim, Tingwei Su, Michael Fulham, David Dagan Feng, Guang Ning
2020 arXiv   pre-print
We also augmented the training data with randomly sampled paired adrenal masses to reduce the influence of imbalanced training data. We used 229 CT scans of patients with adrenal masses.  ...  Methods: We developed a deep multi-scale resemblance network (DMRN) to overcome these limitations and leveraged paired CNNs to evaluate the intra-class similarities.  ...  AMs Related Work Current automated medical imaging classification methods are: (i) traditional, using handcrafted features, with conventional classifiers; and (ii) deep learning using deep convolutional  ... 
arXiv:2007.14625v1 fatcat:3zce7zccnrfm7ize756imydjq4

Cross-Modality Interaction Network for Equine Activity Recognition Using Imbalanced Multi-Modal Data

Axiu Mao, Endai Huang, Haiming Gan, Rebecca S. V. Parkes, Weitao Xu, Kai Liu
2021 Sensors  
In conclusion, CMI-Net and CB focal loss effectively enhanced the equine activity classification performance using imbalanced multi-modal sensor data.  ...  With the recent advances in deep learning, wearable sensors have increasingly been used in automated animal activity recognition.  ...  recall confusion matrix aggregating the classification results under 6-fold cross-validation when using CMI-Net with CB focal loss (γ = 0.5).  ... 
doi:10.3390/s21175818 pmid:34502709 fatcat:cru27grwzzdcpiqv4ul74zeym4

Data Augmentation for Skin Lesion using Self-Attention based Progressive Generative Adversarial Network [article]

Ibrahim Saad Ali, Mamdouh Farouk Mohamed, Yousef Bassyouni Mahdy
2019 arXiv   pre-print
One significant problem with adopting DNNs for skin cancer classification is that the class frequencies in the existing datasets are imbalanced.  ...  Self-attention mechanism is used to directly model the long-range dependencies in the feature maps.  ...  Despite the using of self-attention mechanism, the generated samples suffer from some artifacts due to unstable training behavior. The imbalanced learning rate (TTUR) is used to tackle this issue.  ... 
arXiv:1910.11960v1 fatcat:3ozy75wygbgy5k3qiejavlt2g4

A survey on generative adversarial networks for imbalance problems in computer vision tasks

Vignesh Sampath, Iñaki Maurtua, Juan José Aguilar Martín, Aitor Gutierrez
2021 Journal of Big Data  
It is particularly important that GANs can not only be used to generate synthetic images, but also its fascinating adversarial learning idea showed good potential in restoring balance in imbalanced datasets  ...  Image level imbalances in classification, 2. object level imbalances in object detection and 3. pixel level imbalances in segmentation tasks.  ...  Deep Recurrent Attentive Writer (DRAW) [107] networks combine spatial attention mechanism with a sequential variational autoencoder.  ... 
doi:10.1186/s40537-021-00414-0 pmid:33552840 pmcid:PMC7845583 fatcat:g3p6hbjuj5c5vbe23ms4g6ed6q

Graph Neural Network with Curriculum Learning for Imbalanced Node Classification [article]

Xiaohe Li, Lijie Wen, Yawen Deng, Fuli Feng, Xuming Hu, Lei Wang, Zide Fan
2022 arXiv   pre-print
Traditional solutions for imbalanced classification (e.g. resampling) are ineffective in node classification without considering the graph structure.  ...  The proposed framework is evaluated via several widely used graph datasets, showing that our proposed model consistently outperforms the existing state-of-the-art methods.  ...  From this visualization, we can see that proposed GNN-CL model achieves the best classification results as well as a stable training process.  ... 
arXiv:2202.02529v1 fatcat:dsfblf6l4ve4dprmebiurr2xmy

Exploring to establish an appropriate model for image aesthetic assessment via CNN-based RSRL: An empirical study [article]

Ying Dai
2021 arXiv   pre-print
This method employs the recurrent attention network to focus on some key regions to extract visual features.  ...  In [7] , the authors propose a novel multimodal recurrent attention CNN, which incorporates the visual information with the text information.  ... 
arXiv:2106.03316v2 fatcat:dbxm4u4c6bh6lhpcld3m75fcq4
« Previous Showing results 1 — 15 out of 2,286 results