arrow
返回

ImageDataset2Vec: An image dataset embedding for algorithm selection

delete2021-10-01
delete12
PRE
AI
P
Péricles Miranda *
A
André Nascimento
F
Filipe R. Cordeiro
R
Rafael Ferreira Mello
R
Ricardo B. C. Prudêncio
DOI:10.1016/j.eswa.2021.115053delete
delete原文链接
delete分享
delete收藏
摘要

摘要

En 中文
Convolutional Neural Networks (CNNs) have become the main solution for image classification tasks in different applications. Although several CNN architectures are available, there is no best architecture regardless the problem at hand. The selection of the most suitable CNN architecture is usually performed by trial and error, which may take much time and computational resources. Meta-learning (MtL) is a framework developed in machine learning to perform algorithm selection based on the meta-features of each task being solved. Such meta-features are usually descriptive characteristics extracted from the training dataset available in the task at hand. Despite the increasing attention of MtL for algorithm selection, its success strongly depends on defining relevant meta-features to represent the classification tasks of interest. This paper proposes the ImageDataset2Vec method for extracting meta-features to describe image classification datasets. ImageDataset2Vec adopts a pretrained deep neural network to extract features from images datasets, embedding them in a single feature vector. The derived meta-features are employed by MtL to select CNN architectures for image classification. The proposed approach was evaluated for selecting among six CNN algorithms in 45 two-classes image datasets. The results showed that MtL using ImageDataset2Vec overcame different baseline methods, selecting the best possible CNN algorithm in 84.45% of the datasets. Furthermore, the proposal was statistically equivalent to the ground truth when the best CNN is recommended, i.e., when MtL does not select the best CNN, it selects a competitive algorithm. These results show that the proposal was able to extract representative features from image datasets.
Keyword:
Dataset embedding
Algorithm selection
Meta-learning
Image dataset description
AI总结

AI总结

对已上传原文的论文进行重点信息的提取,主要内容包括:简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。

期刊

Expert Systems with Applications 封面图
Expert Systems with Applications
IF:
7.5
论文数:
3.1W
被引数:
13.6W

机构

U
universidade federal rural de pernambuco (ufrpe)
学者数:
3.7K
论文数: 2.1K
被引数: 1.5K
U
universidade federal de pernambuco
学者数:
434
论文数: 136
被引数: 2.4K
引用论文

引用论文

err分享
err收藏
Learning-based algorithm selection for image segmentation
err2005-06-01
err20
PREAI
errYong, X; Feng, D; Rongchun, Z; Petrou, M
err分享
err收藏
Solitary intraretinal macular hemorrhage
err1984-10-01
err0
PREAI
errElmar P. Messmer; Achim Wessing; Klaus Ruprecht; Gottfried O. H. Naumann
err分享
err收藏
A meta-learning approach for selecting image segmentation algorithm
err2019-12-01
err30
errOAAI
errAguiar, Gabriel Jonas; Mantovani, Rafael Gomes; Mastelini, Saulo M.; de Carvalho, Andre C. P. F. L.; Campos, Gabriel F. C.; Barbon Junior, Sylvio
err分享
err收藏
Data complexity meta-features for regression problems回归问题的数据复杂性元特征
err2017-12-21
err42
errOAAI
errLorena, Ana C.; Maciel, Aron I.; de Miranda, Pericles B. C.; Costa, Ivan G.; Prudencio, Ricardo B. C.
err分享
err收藏
err分享
err收藏
没有更多内容