Addressing Overfitting Problem in Deep Learning‐Based Solutions for Next Generation Data‐Driven Networks
Open Access
- 1 January 2021
- journal article
- research article
- Published by Wiley in Wireless Communications and Mobile Computing
- Vol. 2021 (1)
Abstract
Next‐generation networks are data‐driven by design but face uncertainty due to various changing user group patterns and the hybrid nature of infrastructures running these systems. Meanwhile, the amount of data gathered in the computer system is increasing. How to classify and process the massive data to reduce the amount of data transmission in the network is a very worthy problem. Recent research uses deep learning to propose solutions for these and related issues. However, deep learning faces problems like overfitting that may undermine the effectiveness of its applications in solving different network problems. This paper considers the overfitting problem of convolutional neural network (CNN) models in practical applications. An algorithm for maximum pooling dropout and weight attenuation is proposed to avoid overfitting. First, design the maximum value pooling dropout in the pooling layer of the model to sparse the neurons and then introduce the regularization based on weight attenuation to reduce the complexity of the model when the gradient of the loss function is calculated by backpropagation. Theoretical analysis and experiments show that the proposed method can effectively avoid overfitting and can reduce the error rate of data set classification by more than 10% on average than other methods. The proposed method can improve the quality of different deep learning‐based solutions designed for data management and processing in next‐generation networks.Keywords
Funding Information
- Natural Science Foundation of Hunan Province (2020JJ6088, 2020JJ7007, 2021JJ50050)
- Education Department of Hunan Province (20A162)
This publication has 20 references indexed in Scilit:
- Overfitting effect of artificial neural network based nonlinear equalizer: from mathematical origin to transmission evolutionScience China Information Sciences, 2020
- Learning URL Embedding for Malicious Website DetectionIEEE Transactions on Industrial Informatics, 2020
- Blockchain-Enabled Accountability Mechanism Against Information Leakage in Vertical Industry ServicesIEEE Transactions on Network Science and Engineering, 2020
- Trustworthy Network Anomaly Detection Based on an Adaptive Learning Rate and Momentum in IIoTIEEE Transactions on Industrial Informatics, 2020
- An efficient privacy‐enhanced attribute‐based access control mechanismConcurrency and Computation: Practice and Experience, 2019
- Android HIV: A Study of Repackaging Malware for Evading Machine-Learning DetectionIEEE Transactions on Information Forensics and Security, 2019
- Reducing overfitting of a deep learning breast mass detection algorithm in mammography using synthetic images2019
- Reply to Mummadi et al. : Overfitting and Use of Mismatched Cohorts in Deep Learning Models: Preventable Design LimitationsAmerican Journal of Respiratory and Critical Care Medicine, 2018
- Reduction of Overfitting in Diabetes Prediction Using Deep Learning Neural Network2017
- GPU-based parallel optimization of immune convolutional neural network and embedded systemEngineering Applications of Artificial Intelligence, 2017