Filters








3,250 Hits in 7.9 sec

Prototypical Priors: From Improving Classification to Zero-Shot Learning [article]

Saumya Jetley, Bernardino Romera-Paredes, Sadeep Jayasumana, Philip Torr
2018 arXiv   pre-print
In zero-shot learning scenarios, the same system can be directly deployed to draw inference on unseen classes by simply adding the prototypical information for these new classes at test time.  ...  Using prototypes as prior information, the deepnet pipeline learns the input image projections into the prototypical embedding space subject to minimization of the final classification loss.  ...  The first stage of our network consists of a CNN to enable learning of image features starting from original RGB patches of 48 × 48 (size suitable for both traffic-sign and logo samples in experimental  ... 
arXiv:1512.01192v2 fatcat:6r7lo54tibhwropgayzm3iqe5e

Prototypical Priors: From Improving Classification to Zero-Shot Learning

Saumya Jetley, Bernardino Romera-Paredes, Sadeep Jayasumana, Philip Torr
2015 Procedings of the British Machine Vision Conference 2015  
In zero-shot learning scenarios, the same system can be directly deployed to draw inference on unseen classes by simply adding the prototypical information for these new classes at test time.  ...  Using prototypes as prior information, the deepnet pipeline learns the input image projections into the prototypical embedding space subject to minimization of the final classification loss.  ...  The first stage of our network consists of a CNN to enable learning of image features starting from original RGB patches of 48 × 48 (size suitable for both traffic-sign and logo samples in experimental  ... 
doi:10.5244/c.29.120 dblp:conf/bmvc/JetleyRJT15 fatcat:aph7jcjonnhv3br7n7remq6cli

Deep Metric Learning to Rank

Fatih Cakir, Kun He, Xide Xia, Brian Kulis, Stan Sclaroff
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
On three few-shot image retrieval datasets, FastAP consistently outperforms competing methods, which often involve complex optimization heuristics or costly model ensembles. −  ...  We propose a novel deep metric learning method by revisiting the learning to rank approach.  ...  , feature matching [14] , and fewshot learning [38] .  ... 
doi:10.1109/cvpr.2019.00196 dblp:conf/cvpr/Cakir0XKS19 fatcat:dp7zjz36ovhtlgqhe4zgjchvhm

Data-driven geophysics: from dictionary learning to deep learning [article]

Siwei Yu, Jianwei Ma
2020 arXiv   pre-print
We present a coding tutorial and a summary of tips for beginners and interested geophysical readers to rapidly explore deep learning.  ...  Some promising directions are provided for future research involving deep learning in geophysics, such as unsupervised learning, transfer learning, multimodal deep learning, federated learning, uncertainty  ...  Acknowledgments The work was supported in part by the National Key Research and Development Program Data Availability Statement Data were not used, nor created for this research.  ... 
arXiv:2007.06183v2 fatcat:ow45ejo7izbkpmssedwd74rbym

Learning to Group: A Bottom-Up Framework for 3D Part Discovery in Unseen Categories [article]

Tiange Luo, Kaichun Mo, Zhiao Huang, Jiarui Xu, Siyu Hu, Liwei Wang, Hao Su
2021 arXiv   pre-print
At the core of our approach is to restrict the local context for extracting part-level features, which encourages the generalizability to unseen categories.  ...  Being able to learn the geometry prior of parts and transfer this prior to unseen categories pose fundamental challenges on data-driven shape segmentation approaches.  ...  Here also exists a cascaded structure where the verification network will focus on the pairs selected by the policy network and make a double verification.  ... 
arXiv:2002.06478v4 fatcat:knms24hbdbeilmk24xdy27lxlu

A Comprehensive Survey of Machine Learning Applied to Radar Signal Processing [article]

Ping Lang, Xiongjun Fu, Marco Martorella, Jian Dong, Rui Qin, Xianpeng Meng, Min Xie
2020 arXiv   pre-print
With the rapid development of machine learning (ML), especially deep learning, radar researchers have started integrating these new methods when solving RSP-related problems.  ...  This paper aims at helping researchers and practitioners to better understand the application of ML techniques to RSP-related problems by providing a comprehensive, structured and reasoned literature overview  ...  Few-shot learning can rapidly generalize to new tasks of limited supervised experience by turning to prior knowledge, which mimics human's ability to acquire knowledge from few examples through generalization  ... 
arXiv:2009.13702v1 fatcat:m6am73324zdwba736sn3vmph3i

Machine Learning Approach to Detect Drowsiness on Behavioral Parameters

Aannd R, BMS Institute of Technology and Management, Yelahanka, Bangalore 560064, India., Anil G N, Rishika Sankaran, Anushruti Adhikari, Kruthika Ravishankar, BMS Institute of Technology and Management, Yelahanka, Bangalore 560064, India., BMS Institute of Technology and Management, Yelahanka, Bangalore 560064, India., BMS Institute of Technology and Management, Yelahanka, Bangalore 560064, India., BMS Institute of Technology and Management, Yelahanka, Bangalore 560064, India.
2022 Ymer  
Our project is focused on building a single-access platform for various object detection tasks.  ...  Thanks to the availability of large amounts of data, faster GPUs, and improved algorithms, we can now quickly train computers to detect and classify many elements inside a picture with high accuracy.  ...  The method focuses on multiple subregions of images (grayscale images) and works on recognition of faces by looking for certain attributes in each subregion.  ... 
doi:10.37896/ymer21.01/01 fatcat:rdmpkrttdngf7gijqqhwcdz7uu

A Survey on Text Classification: From Shallow to Deep Learning [article]

Qian Li, Hao Peng, Jianxin Li, Congying Xia, Renyu Yang, Lichao Sun, Philip S. Yu, Lifang He
2021 arXiv   pre-print
We create a taxonomy for text classification according to the text involved and the models used for feature extraction and classification.  ...  This paper fills the gap by reviewing the state-of-the-art approaches from 1961 to 2021, focusing on models from traditional models to deep learning.  ...  Thanks for computing infrastructure provided by Huawei MindSpore platform.  ... 
arXiv:2008.00364v6 fatcat:a6zp52rtf5awlh253yp62wqt3a

Learning to Recognize Faces by Successive Meetings

M. Castrillón-Santana, O. Déniz-Suárez, J. Lorenzo-Navarro, M. Hernández-Tejera
2006 Journal of Multimedia  
In this paper we focus on the face recognition problem.  ...  However, instead of following the usual approach of manually gathering and registering face images to build a training set to compute a classifier off-line, the system will start with an empty training  ...  A simple way to implement that using recognition and/or verification techniques is to apply both approaches in a cascade configuration.  ... 
doi:10.4304/jmm.1.7.1-8 fatcat:ui4xsq7t5vanhp4ac3a5i22bli

A Deep Learning based Light-weight Face Mask Detector with Residual Context Attention and Gaussian Heatmap to Fight Against COVID-19

Xinqi Fan, Mingjie Jiang, Hong Yan
2021 IEEE Access  
To cope with this problem, we propose two novel modules -RCAM, to focus on learning important information, in section III-B, and SGHR, to learn more discriminating features for faces with and without masks  ...  To focus on the important face mask related features, we cascade a convolutional block attention module (CBAM) [42] after the CEB, and add a skip connection.  ... 
doi:10.1109/access.2021.3095191 fatcat:yma42liatzgdnj6wlbnl3ulnri

Personalized Fashion Recommendation from Personal Social Media Data: An Item-to-Set Metric Learning Approach [article]

Haitian Zheng, Kefei Wu, Jong-Hwi Park, Wei Zhu, Jiebo Luo
2020 arXiv   pre-print
Meanwhile, social networks provide an open and new data source for personalized fashion analysis.  ...  To this end, we present an item-to-set metric learning framework that learns to compute the similarity between a set of historical fashion items of a user to a new fashion item.  ...  The feature averaging operation is also proposed in [33] for few-shot learning.  ... 
arXiv:2005.12439v1 fatcat:xy3gsq4hrbg2jlt74b2nrqdbba

Glance and Focus: a Dynamic Approach to Reducing Spatial Redundancy in Image Classification [article]

Yulin Wang, Kangchen Lv, Rui Huang, Shiji Song, Le Yang, Gao Huang
2020 arXiv   pre-print
are strategically selected from the original image with reinforcement learning.  ...  The accuracy of deep convolutional neural networks (CNNs) generally improves when fueled with high resolution images. However, this often comes at a high computational cost and high memory footprint.  ...  Very deep convolutional networks for large-scale image recognition.  ... 
arXiv:2010.05300v1 fatcat:7gwawa2usbcttbqlyex6xybbke

Learning to Reconstruct Shapes from Unseen Classes [article]

Xiuming Zhang, Zhoutong Zhang, Chengkai Zhang, Joshua B. Tenenbaum, William T. Freeman, Jiajun Wu
2018 arXiv   pre-print
From a single image, humans are able to perceive the full 3D shape of an object by exploiting learned shape priors from everyday life.  ...  surfaces, and 3D voxel-based representations, in a principled manner that exploits the causal structure of how 3D shapes give rise to 2D images.  ...  Acknowledgements We thank the anonymous reviewers for their constructive comments.  ... 
arXiv:1812.11166v1 fatcat:jd6l3xva3nbglmh7l73miqgxoi

Deep learning in radiology: An overview of the concepts and a survey of the state of the art with focus on MRI

Maciej A. Mazurowski, Mateusz Buda, Ashirbani Saha, Mustafa R. Bashir
2018 Journal of Magnetic Resonance Imaging  
Deep-learning algorithms have shown groundbreaking performance in a variety of sophisticated tasks, especially those related to images. They have often matched or exceeded human performance.  ...  Deep learning is a branch of artificial intelligence where networks of simple interconnected units are used to extract patterns from data in order to solve complex problems.  ...  Acknowledgments: The authors would like to thank Gemini Janas for reviewing and editing this article.  ... 
doi:10.1002/jmri.26534 pmid:30575178 pmcid:PMC6483404 fatcat:7jg5sr7z6bbehd6xabsjw6bcde

Review of end-to-end speech synthesis technology based on deep learning [article]

Zhaoxi Mu, Xinyu Yang, Yizhuo Dong
2021 arXiv   pre-print
Due to the limitations of high complexity and low efficiency of traditional speech synthesis technology, the current research focus is the deep learning-based end-to-end speech synthesis technology, which  ...  Moreover, this paper also summarizes the open-source speech corpus of English, Chinese and other languages that can be used for speech synthesis tasks, and introduces some commonly used subjective and  ...  For example, in the SPSS model based on deep neural network (DNN), DNN can learn the mapping function from linguistic features (input) to acoustic features (output).  ... 
arXiv:2104.09995v1 fatcat:q5lx74ycx5hobjox4ktl3amfta
« Previous Showing results 1 — 15 out of 3,250 results