A survey on deep learning tools dealing with data scarcity: definitions, challenges, solutions, tips, and applications

被引:363
作者
Alzubaidi, Laith [1 ,2 ,3 ]
Bai, Jinshuai [1 ,2 ]
Al-Sabaawi, Aiman [4 ]
Santamaria, Jose [5 ]
Albahri, A. S. [6 ]
Al-dabbagh, Bashar Sami Nayyef [7 ]
Fadhel, Mohammed. A. A. [8 ]
Manoufali, Mohamed [9 ]
Zhang, Jinglan [4 ]
Al-Timemy, Ali. H. H. [10 ]
Duan, Ye [11 ]
Abdullah, Amjed [11 ]
Farhan, Laith [12 ]
Lu, Yi [4 ]
Gupta, Ashish [2 ]
Albu, Felix [13 ]
Abbosh, Amin [9 ]
Gu, Yuantong [1 ,2 ]
机构
[1] Queensland Univ Technol, Sch Mech Med & Proc Engn, Brisbane, Qld 4000, Australia
[2] Queensland Univ Technol, QUASR ARC Ind Transformat Training Ctr Joint Biome, Brisbane, Qld 4000, Australia
[3] Queensland Univ Technol, Ctr Data Sci, Brisbane, Qld 4000, Australia
[4] Queensland Univ Technol, Sch Comp Sci, Brisbane, Qld 4000, Australia
[5] Univ Jaen, Dept Comp Sci, Jaen 23071, Spain
[6] Univ Pendidikan Sultan, Dept Comp, Tanjung Malim 35900, Malaysia
[7] Charles III Univ Madrid, Dept Comp Sci & Technol, Madrid 28911, Spain
[8] Univ Sumer, Coll Comp Sci & Informat Technol, Thi Qar 64005, Iraq
[9] Univ Queensland, Sch Informat Technol & Elect Engn, St Lucia, Qld 4072, Australia
[10] Univ Baghdad, Al Khwarizmi Coll Engn, Biomed Engn Dept, Baghdad, Iraq
[11] Univ Missouri, Dept Elect Engn & Comp Sci, Columbia, MO 65211 USA
[12] Manchester Metropolitan Univ, Sch Engn, Manchester M1 5GD, Lancashire, England
[13] Valahia Univ Targoviste, Dept Elect, Targoviste 130082, Romania
基金
澳大利亚研究理事会;
关键词
Deep learning; Data scarcity; Machine learning; Convolutional neural network (CNN); Deep neural network architectures; Lack of training data; Small datasets; Medical image analysis; Transfer learning; PINN; SMOTE; DeepSMOTE; Generative Adversarial Networks; Electromagnetic imaging; Civil structural health monitoring; Meteorology; Wireless communications; Fluid mechanics; Cybersecurity; Trustworthy data; CONVOLUTIONAL NEURAL-NETWORKS; X-RAY IMAGES; FAULT-DIAGNOSIS; UNCERTAINTY QUANTIFICATION; INDUSTRIAL INTERNET; VEHICLE DETECTION; DATA AUGMENTATION; CROSS-VALIDATION; CLASSIFICATION; GAN;
D O I
10.1186/s40537-023-00727-2
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Data scarcity is a major challenge when training deep learning (DL) models. DL demands a large amount of data to achieve exceptional performance. Unfortunately, many applications have small or inadequate data to train DL frameworks. Usually, manual labeling is needed to provide labeled data, which typically involves human annotators with a vast background of knowledge. This annotation process is costly, time-consuming, and error-prone. Usually, every DL framework is fed by a significant amount of labeled data to automatically learn representations. Ultimately, a larger amount of data would generate a better DL model and its performance is also application dependent. This issue is the main barrier for many applications dismissing the use of DL. Having sufficient data is the first step toward any successful and trustworthy DL application. This paper presents a holistic survey on state-of-the-art techniques to deal with training DL models to overcome three challenges including small, imbalanced datasets, and lack of generalization. This survey starts by listing the learning techniques. Next, the types of DL architectures are introduced. After that, state-of-the-art solutions to address the issue of lack of training data are listed, such as Transfer Learning (TL), Self-Supervised Learning (SSL), Generative Adversarial Networks (GANs), Model Architecture (MA), Physics-Informed Neural Network (PINN), and Deep Synthetic Minority Oversampling Technique (DeepSMOTE). Then, these solutions were followed by some related tips about data acquisition needed prior to training purposes, as well as recommendations for ensuring the trustworthiness of the training dataset. The survey ends with a list of applications that suffer from data scarcity, several alternatives are proposed in order to generate more data in each application including Electromagnetic Imaging (EMI), Civil Structural Health Monitoring, Medical imaging, Meteorology, Wireless Communications, Fluid Mechanics, Microelectromechanical system, and Cybersecurity. To the best of the authors' knowledge, this is the first review that offers a comprehensive overview on strategies to tackle data scarcity in DL.
引用
收藏
页数:82
相关论文
共 130 条
[11]  
[Anonymous], Common Weakness Enumeration
[12]  
Avci O., 2022, Dynamics of Civil Structures, V2, P93, DOI [10.1007/978-3-030-77143-0_10, DOI 10.1007/978-3-030-77143-0_10]
[13]   A systematic review on Deep Learning approaches for IoT security [J].
Aversano, Lerina ;
Bernardi, Mario Luca ;
Cimitile, Marta ;
Pecori, Riccardo .
COMPUTER SCIENCE REVIEW, 2021, 40
[14]   Robust Classification and Detection of Big Medical Data Using Advanced Parallel K-Means Clustering, YOLOv4, and Logistic Regression [J].
Awad, Fouad H. ;
Hamad, Murtadha M. ;
Alzubaidi, Laith .
LIFE-BASEL, 2023, 13 (03)
[15]   Structural health monitoring using extremely compressed data through deep learning [J].
Azimi, Mohsen ;
Pekcan, Gokhan .
COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING, 2020, 35 (06) :597-614
[16]   Computer vision and deep learning-based data anomaly detection method for structural health monitoring [J].
Bao, Yuequan ;
Tang, Zhiyi ;
Li, Hui ;
Zhang, Yufeng .
STRUCTURAL HEALTH MONITORING-AN INTERNATIONAL JOURNAL, 2019, 18 (02) :401-421
[17]   A systematic review on overfitting control in shallow and deep neural networks [J].
Bejani, Mohammad Mahdi ;
Ghatee, Mehdi .
ARTIFICIAL INTELLIGENCE REVIEW, 2021, 54 (08) :6391-6438
[18]  
Black PE, 2018, J RES NATL INST STAN, V123, DOI [10.6028/123.005, 10.6028/jres.123.005]
[19]   Juliet 1.1 C/C++ and Java']Java Test Suite [J].
Boland, Tim ;
Black, Paul E. .
COMPUTER, 2012, 45 (10) :88-90
[20]  
Brownlee J., 2022, Data preparation for machine learning