Filters








19 Hits in 6.0 sec

Evolving Deep Architecture Generation with Residual Connections for Image Classification Using Particle Swarm Optimization

Tom Lawrence, Li Zhang, Kay Rogage, Chee Peng Lim
2021 Sensors  
A PSO variant is proposed which incorporates a new encoding scheme and a new search mechanism guided by non-uniformly randomly selected neighboring and global promising solutions for the search of optimal  ...  Owing to the guidance of diverse non-uniformly selected neighboring promising solutions in combination with the swarm leader at fine-grained and global levels, the proposed model produces a rich assortment  ...  Conflicts of Interest: The authors declare no conflict of interest. Sensors 2021, 21, 7936  ... 
doi:10.3390/s21237936 pmid:34883940 fatcat:a3pl6q4iq5gkvdbrae7uaqg4yy

Dynamic Hierarchical Mimicking Towards Consistent Optimization Objectives [article]

Duo Li, Qifeng Chen
2021 arXiv   pre-print
While the depth of modern Convolutional Neural Networks (CNNs) surpasses that of the pioneering networks with a significant margin, the traditional way of appending supervision only over the final classifier  ...  However, it is still vulnerable to issues including interference to the hierarchical representation generation process and inconsistent optimization objectives, as illustrated theoretically and empirically  ...  R-1 denotes Rank-1 accuracy. w/ pretrain and w/o pretrain means with and without ImageNet pretrained weights loaded respectively.  ... 
arXiv:2003.10739v2 fatcat:mn5e5yc45vbspnpl4axhczcuwe

Cyclic Differentiable Architecture Search [article]

Hongyuan Yu, Houwen Peng, Yan Huang, Jianlong Fu, Hao Du, Liang Wang, Haibin Ling
2022 arXiv   pre-print
First, the search network generates an initial architecture for evaluation, and the weights of the evaluation network are optimized.  ...  Repeating the above cycle results in joint optimization of the search and evaluation networks and thus enables the evolution of the architecture to fit the final evaluation network.  ...  The weights of the new evaluation network are initialized with the parameters inheriting from previous training.  ... 
arXiv:2006.10724v4 fatcat:vbq6g2arfndxlmyceqoqfg22qi

RelativeNAS: Relative Neural Architecture Search via Slow-Fast Learning [article]

Hao Tan, Ran Cheng, Shihua Huang, Cheng He, Changxiao Qiu, Fan Yang, Ping Luo
2021 arXiv   pre-print
The proposed RelativeNAS brings several unique advantages: (1) it achieves state-of-the-art performance on ImageNet with top-1 error rate of 24.88%, i.e. outperforming DARTS and AmoebaNet-B by 1.82% and  ...  1.12% respectively; (2) it spends only nine hours with a single 1080Ti GPU to obtain the discovered cells, i.e. 3.75x and 7875x faster than DARTS and AmoebaNet respectively; (3) it provides that the discovered  ...  A large network of 20 cells (i.e. s is set to 6) is built with the selected normal and reduction cells while the initial number of channels is set to 36.  ... 
arXiv:2009.06193v3 fatcat:6ymzh27cpzgv5drgsemoybmdde

Efficient Visual Recognition with Deep Neural Networks: A Survey on Recent Advances and New Directions [article]

Yang Wu, Dingheng Wang, Xiaotong Lu, Fan Yang, Guoqi Li, Weisheng Dong, Jianbo Shi
2021 arXiv   pre-print
Deep neural networks (DNNs) have largely boosted their performances on many concrete tasks, with the help of large amounts of training data and new powerful computation resources.  ...  In this paper, we present the review of the recent advances with our suggestions on the new possible directions towards improving the efficiency of DNN-related visual recognition approaches.  ...  topology, e.g., SqueezeNet [75] uses amounts of 1 × 1 convolutions to replace 3 × 3 convolutions and reduce the counts of channels in the rest 3 × 3 convolutions.  ... 
arXiv:2108.13055v2 fatcat:nf3lymdbvzgl7otl7gjkk5qitq

Deep Learning for Generic Object Detection: A Survey [article]

Li Liu, Wanli Ouyang, Xiaogang Wang, Paul Fieguth, Jie Chen, Xinwang Liu, Matti Pietikäinen
2019 arXiv   pre-print
Object detection, one of the most fundamental and challenging problems in computer vision, seeks to locate object instances from a large number of predefined categories in natural images.  ...  Given this period of rapid evolution, the goal of this paper is to provide a comprehensive survey of the recent achievements in this field brought about by deep learning techniques.  ...  This work has been supported by the Center for Machine Vision and Signal Analysis at the University of Oulu (Finland) and the National Natural Science Foundation of China under Grant 61872379.  ... 
arXiv:1809.02165v4 fatcat:b7ozzcy46bek5jx7l3qomj6e3q

Shape Prediction of Nasal Bones by Digital 2D-Photogrammetry of the Nose Based on Convolution and Back-Propagation Neural Network

Ho Nguyen Anh Tuan, Nguyen Dao Xuan Hai, Nguyen Truong Thinh, Luminita Moraru
2022 Computational and Mathematical Methods in Medicine  
To boost performance and efficacy, it is deliberately constructed with many layers and different filter sizes, with less filters and optimizing parameters.  ...  In conclusion, the proposed model performed the potential hybrid of CNN and BPNN with its application to give expected accuracy in terms of keypoint localization and nasal morphology regression.  ...  Acknowledgments The authors would like to specially thank the support of Pham Ngoc Thach University of Medicine and Ho Chi Minh City University of Technology and Education in experimenting and collecting  ... 
doi:10.1155/2022/5938493 pmid:35069786 pmcid:PMC8767378 fatcat:7mjukzr3qfhhritqxohsyup5gu

Squeeze-and-Excitation Networks [article]

Jie Hu, Li Shen, Samuel Albanie, Gang Sun, Enhua Wu
2019 arXiv   pre-print
In this work, we focus instead on the channel relationship and propose a novel architectural unit, which we term the "Squeeze-and-Excitation" (SE) block, that adaptively recalibrates channel-wise feature  ...  The central building block of convolutional neural networks (CNNs) is the convolution operator, which enables networks to construct informative features by fusing both spatial and channel-wise information  ...  The work is supported in part by NSFC Grants (61632003, 61620106003, 61672502, 61571439) , National Key R&D Program of China (2017YFB1002701), and Macao FDCT Grant (068/2015/A2).  ... 
arXiv:1709.01507v4 fatcat:ofry6usryze7dlcrbzztvozqhm

HASA: Hybrid Architecture Search with Aggregation Strategy for Echinococcosis Classification and Ovary Segmentation in Ultrasound Images [article]

Jikuan Qian, Rui Li, Xin Yang, Yuhao Huang, Mingyuan Luo, Zehui Lin, Wenhui Hong, Ruobing Huang, Haining Fan, Dong Ni, Jun Cheng
2022 arXiv   pre-print
However, manual design and selection of suitable network architectures are time-consuming and require substantial effort of human experts.  ...  The hybrid framework consists of a pre-trained backbone and several searched cells (i.e., network building blocks), which takes advantage of the strengths of both NAS and the expert knowledge from existing  ...  Acknowledgment This study was supported by National Natural Science Foundation of China (Nos. 62171290, 61901275, and 62101343)  ... 
arXiv:2204.06697v2 fatcat:jsgs47nhlrcwrasebu2yfkmgy4

Deep Learning in Mobile and Wireless Networking: A Survey [article]

Chaoyun Zhang, Paul Patras, Hamed Haddadi
2019 arXiv   pre-print
We first briefly introduce essential background and state-of-the-art in deep learning techniques with potential applications to networking.  ...  The rapid uptake of mobile devices and the rising popularity of mobile applications and services pose unprecedented demands on mobile and wireless networking infrastructure.  ...  [323] Indoor localization CSI RBM Explores features of wireless channel data and obtains optimal weights as fingerprints Wang et al.  ... 
arXiv:1803.04311v3 fatcat:awuvyviarvbr5kd5ilqndpfsde

Deep Learning in Mobile and Wireless Networking: A Survey

Chaoyun Zhang, Paul Patras, Hamed Haddadi
2019 IEEE Communications Surveys and Tutorials  
The rapid uptake of mobile devices and the rising popularity of mobile applications and services pose unprecedented demands on mobile and wireless networking infrastructure.  ...  We first briefly introduce essential background and state-of-theart in deep learning techniques with potential applications to networking.  ...  After transfer through an additive white Gaussian noise channel, a receiver employs another MLP to decode messages and select the one with the highest probability of occurrence.  ... 
doi:10.1109/comst.2019.2904897 fatcat:xmmrndjbsfdetpa5ef5e3v4xda

A Survey on Graph-Based Deep Learning for Computational Histopathology [article]

David Ahmedt-Aristizabal, Mohammad Ali Armin, Simon Denman, Clinton Fookes, Lars Petersson
2021 arXiv   pre-print
With the remarkable success of representation learning for prediction problems, we have witnessed a rapid expansion of the use of machine learning and deep learning for the analysis of digital pathology  ...  The phenotypical and topological distribution of constituent histological entities play a critical role in tissue diagnosis.  ...  that optimally represent the data [2] .  ... 
arXiv:2107.00272v2 fatcat:3eskkeref5ccniqsjgo3hqv2sa

Deep Learning for Generic Object Detection: A Survey

Li Liu, Wanli Ouyang, Xiaogang Wang, Paul Fieguth, Jie Chen, Xinwang Liu, Matti Pietikäinen
2019 International Journal of Computer Vision  
Object detection, one of the most fundamental and challenging problems in computer vision, seeks to locate object instances from a large number of predefined categories in natural images.  ...  Given this period of rapid evolution, the goal of this paper is to provide a comprehensive survey of the recent achievements in this field brought about by deep learning techniques.  ...  This work has been supported by the Center for Machine Vision and Signal Analysis at the University of Oulu (Finland) and the National Natural Science Foundation of China under Grant 61872379.  ... 
doi:10.1007/s11263-019-01247-4 fatcat:isdmz4febvbthgowo33c6ifhm4

Exploring the Design Space of Deep Convolutional Neural Networks at Large Scale [article]

Forrest Iandola
2016 arXiv   pre-print
Judiciously choosing benchmarks and metrics. 2. Rapidly training CNN models. 3. Defining and describing the CNN design space. 4. Exploring the design space of CNN architectures.  ...  To our knowledge, there is no single CNN/DNN architecture that solves all problems optimally. Instead, the "right" CNN/DNN architecture varies depending on the application at hand.  ...  [128] use a combination of pruning, quantization, and Huffman encoding to compress the weights of pretrained models by 35x with no reduction in accuracy.  ... 
arXiv:1612.06519v1 fatcat:jwo2gyfjvfh3lbkfdntctx24o4

Comparative Analysis of Different Machine Learning Classifiers for the Prediction of Chronic Diseases [chapter]

Rajesh Singh, Anita Gehlot, Dharam Buddhi
2022 Comparative Analysis of Different Machine Learning Classifiers for the Prediction of Chronic Diseases  
This paper forms the basis of understanding the difficulty of the domain and the amount of efficiency achieved by the various methods recently.  ...  Chronic Diseases are the most dangerous diseases for humans and have significant effects on human life. Chronic Diseases like heart disease & Diabetes are the main causes of death.  ...  The performance of existing and proposed algorithms is analysed with regard to several metrics.  ... 
doi:10.13052/rp-9788770227667 fatcat:da47mjbbyzfwnbpde7rgbrlppe
« Previous Showing results 1 — 15 out of 19 results