OVERFITTING MITIGATION AND TRAINING OPTIMIZATION STRATEGIES IN DEEP LEARNING-BASED IMAGE CLASSIFICATION: A SYSTEMATIC REVIEW
DOI:
https://doi.org/10.33480/jitk.v12i1.7948Keywords:
Deep Learning Optimization, Image Classification, Overfitting Mitigation, Systematic Literature Review, Training StabilityAbstract
Overfitting remains a critical challenge in developing deep learning–based image classification models, particularly as modern architectures become increasingly complex and parameter-intensive. Although convolutional and transformer-based models have demonstrated strong predictive performance, their generalization capability depends strongly on the availability of large, well-annotated, and diverse training datasets. This condition is often difficult to achieve in real-world domains such as medicine, agriculture, and industrial inspection. Previous survey studies have examined various techniques related to image classification, data augmentation, and optimization strategies; however, these studies typically analyse individual approaches in isolation. As a result, opportunities remain to further synthesise the relationships among the underlying causes of overfitting, mitigation strategies, and training parameter configurations within a unified analytical perspective. This study employed the PRISMA framework to conduct a Systematic Literature Review (SLR). A total of 174 primary studies published between 2020 and 2025 were traced to address the gap. The review identifies three major sources of overfitting in image classification tasks: limited labeled data, model architecture complexity, and data and label quality issues. Based on these findings, the study synthesises the corresponding mitigation strategies reported in the literature. The main contribution of this study is not the proposal of a new theoretical taxonomy, but the systematic organisation and synthesis of methodological evidence into an integrated analytical framework. The resulting analytical framework relates the underlying causes of overfitting, mitigation strategies, training optimisation mechanisms, and parameter configuration practices within a unified perspective, thereby facilitating a more integrated interpretation of how these complementary aspects contribute to model generalisation in deep learning–based image classification.
Downloads
References
[1] F. Feng, M. Gao, R. Liu, S. Yao, and G. Yang, ‘A deep learning framework for crop mapping with reconstructed Sentinel-2 time series images’, Comput. Electron. Agric., vol. 213, 2023, doi: 10.1016/j.compag.2023.108227.
[2] Y. Asam and Z. Zhou, ‘A dual-path feature learning neural network for enhancing image classification in marine biology’, Multimedia Tools Appl, 2025, doi: 10.1007/s11042-025-21010-x.
[3] J. Qiu et al., ‘A Flattened-Tree Genetic Programming Approach with Multi-Scale Feature Extraction for Image Classification’, Memetic Comput., vol. 17, no. 3, 2025, doi: 10.1007/s12293-025-00472-4.
[4] H. Song, ‘A Leading but Simple Classification Method for Remote Sensing Images’, Ann. Emer. Tech. Comp., vol. 7, no. 3, pp. 1–20, 2023, doi: 10.33166/AETiC.2023.03.001.
[5] E. Keskin Bilgiç, İ. Gökbay, and Y. Kayar, ‘Innovative Approaches to Clinical Diagnosis: Transfer Learning in Facial Image Classification for Celiac Disease Identification’, Appl. Sci., vol. 14, no. 14, 2024, doi: 10.3390/app14146207.
[6] Y. Suo, Z. He, and Y. Liu, ‘Deep learning CS-ResNet-101 model for diabetic retinopathy classification’, Biomed. Signal Process. Control, vol. 97, 2024, doi: 10.1016/j.bspc.2024.106661.
[7] Z. He et al., ‘Deconv-transformer (DecT): A histopathological image classification model for breast cancer based on color deconvolution and transformer architecture’, Inf Sci, vol. 608, pp. 1093–1112, 2022, doi: 10.1016/j.ins.2022.06.091.
[8] S.-Y. Lu, Z. Zhu, Y. Tang, X. Zhang, and X. Liu, ‘CTBViT: A novel ViT for tuberculosis classification with efficient block and randomized classifier’, Biomed. Signal Process. Control, vol. 100, 2025, doi: 10.1016/j.bspc.2024.106981.
[9] X. Tong, Z. Liang, and F. Liu, ‘Succulent Plant Image Classification Based on Lightweight GoogLeNet with CBAM Attention Mechanism’, Appl. Sci., vol. 15, no. 7, 2025, doi: 10.3390/app15073730.
[10] Y. Hu, J. Tang, Y. Xu, R. Xu, and B. Huang, ‘EL-DenseNet: a novel method for identifying the flame state of converter steelmaking based on dense convolutional neural networks’, Signal Image Video Process., vol. 18, no. 4, pp. 3445–3457, 2024, doi: 10.1007/s11760-024-03011-9.
[11] M. Piao, Y. Sheng, J. Yan, and C. H. Jin, ‘Image Hash Layer Triggered CNN Framework for Wafer Map Failure Pattern Retrieval and Classification’, ACM Trans. Knowl. Discov. Data, vol. 18, no. 4, 2024, doi: 10.1145/3638053.
[12] Y. Kong, X. Ma, and C. Wen, ‘A New Method of Deep Convolutional Neural Network Image Classification Based on Knowledge Transfer in Small Label Sample Environment’, Sensors, vol. 22, no. 3, 2022, doi: 10.3390/s22030898.
[13] F. Xu, P. Wang, and H. Xu, ‘Deep pyramidal residual networks with inception sub-structure in image classification’, J. Intelligent Fuzzy Syst., vol. 45, no. 4, pp. 5885–5906, 2023, doi: 10.3233/JIFS-230569.
[14] Q. Huang, H. Zhang, M. Xue, J. Song, and M. Song, ‘A Survey of Deep Learning for Low-shot Object Detection’, ACM Comput Surv, vol. 56, no. 5, 2024, doi: 10.1145/3626312.
[15] D. Zhai, R. Shi, J. Jiang, and X. Liu, ‘Rectified Meta-learning from Noisy Labels for Robust Image-based Plant Disease Classification’, ACM Trans. Multimedia Comput. Commun. Appl., vol. 18, no. 1s, 2022, doi: 10.1145/3472809.
[16] S. Li, H. Hu, S. Huo, and H. Liang, ‘Clean, performance-robust, and performance-sensitive historical information based adversarial self-distillation’, IET Comput. Vision, vol. 18, no. 5, pp. 591–612, 2024, doi: 10.1049/cvi2.12265.
[17] M. Kang et al., ‘Efficient one-shot federated learning on medical data using knowledge distillation with image synthesis and client model adaptation’, Med. Image Anal., vol. 105, 2025, doi: 10.1016/j.media.2025.103714.
[18] M. Liu, Y. Yu, Z. Ji, J. Han, and Z. Zhang, ‘Tolerant Self-Distillation for image classification’, Neural Netw., vol. 174, 2024, doi: 10.1016/j.neunet.2024.106215.
[19] Y. Zhou, Z. Wang, and J. Li, ‘Knowledge Distillation Based on Narrow-Deep Networks’, Neural Process Letters, vol. 56, no. 3, 2024, doi: 10.1007/s11063-024-11646-5.
[20] J. Zhang, Z. Chen, and L. Dai, ‘Unleashing the Power of Each Distilled Image’, IEEE Trans Image Process, 2025, doi: 10.1109/TIP.2025.3624626.
[21] J. Mi, C. Ma, L. Zheng, M. Zhang, M. Li, and M. Wang, ‘WGAN-CL: A Wasserstein GAN with confidence loss for small-sample augmentation’, Expert Sys Appl, vol. 233, 2023, doi: 10.1016/j.eswa.2023.120943.
[22] K. Alomar, H. I. Aysel, and X. Cai, ‘Data Augmentation in Classification and Segmentation: A Survey and New Strategies’, J. Imaging, vol. 9, no. 2, 2023, doi: 10.3390/jimaging9020046.
[23] Z. Wang, Y. Guo, Q. Li, G. Yang, and W. Zuo, ‘DualAug: Exploiting additional heavy augmentation with OOD data rejection’, Neurocomputing, vol. 654, 2025, doi: 10.1016/j.neucom.2025.131209.
[24] Y. Chen, X. Fang, and H. Han, ‘An improved transfer learning algorithm using integration method and its application in image segmentation and recognition’, Multimedia Tools Appl, vol. 84, no. 24, pp. 28085–28114, 2025, doi: 10.1007/s11042-024-20290-z.
[25] J. Su, X. Yu, X. Wang, Z. Wang, and G. Chao, ‘Enhanced transfer learning with data augmentation’, Eng Appl Artif Intell, vol. 129, 2024, doi: 10.1016/j.engappai.2023.107602.
[26] M. Sharma, J. Heard, E. Saber, and P. Markopoulos, ‘Convolutional Neural Network Compression via Dynamic Parameter Rank Pruning’, IEEE Access, vol. 13, pp. 18441–18456, 2025, doi: 10.1109/ACCESS.2025.3533419.
[27] W. Wei, X. Fu, S. Ma, Y. Zhu, and N. Lu, ‘Reducing overfitting in vehicle recognition by decorrelated sparse representation regularisation’, IET Comput. Vision, vol. 18, no. 8, pp. 1351–1361, 2024, doi: 10.1049/cvi2.12320.
[28] Z. Ren, Y.-D. Zhang, and S. Wang, ‘A Hybrid Framework for Lung Cancer Classification’, Electronics (Switzerland), vol. 11, no. 10, 2022, doi: 10.3390/electronics11101614.
[29] H. Sun et al., ‘An Improved Adam’s Algorithm for Stomach Image Classification’, Algorithms, vol. 17, no. 7, 2024, doi: 10.3390/a17070272.
[30] Y. Zheng et al., ‘Cyclic Learning Rate-Based Co-Training for Image Classification With Noisy Labels’, IEEE Access, vol. 13, pp. 6292–6305, 2025, doi: 10.1109/ACCESS.2025.3526332.
[31] L. Hao, K. Hao, B. Wei, and X.-S. Tang, ‘Boosting the transferability of adversarial examples via stochastic serial attack’, Neural Netw., vol. 150, pp. 58–67, 2022, doi: 10.1016/j.neunet.2022.02.025.
[32] J. Hu et al., ‘APDL: an adaptive step size method for white-box adversarial attacks’, Complex Intell. Syst., vol. 11, no. 1, 2025, doi: 10.1007/s40747-024-01748-x.
[33] S. Q. Gilani and O. Marques, ‘Skin lesion analysis using generative adversarial networks: a review’, Multimedia Tools Appl, vol. 82, no. 19, pp. 30065–30106, 2023, doi: 10.1007/s11042-022-14267-z.
[34] B. Shi, W. Li, J. Huo, P. Zhu, L. Wang, and Y. Gao, ‘Global- and local-aware feature augmentation with semantic orthogonality for few-shot image classification’, Pattern Recogn., vol. 142, 2023, doi: 10.1016/j.patcog.2023.109702.
[35] M. Momeny et al., ‘Greedy Autoaugment for classification of mycobacterium tuberculosis image via generalized deep CNN using mixed pooling based on minimum square rough entropy’, Comput. Biol. Med., vol. 141, 2022, doi: 10.1016/j.compbiomed.2021.105175.
[36] X. Zhang, C. Yuan, W. Sun, and S. K. Jha, ‘Image Emotion Classification Network Based on Multilayer Attentional Interaction, Adaptive Feature Aggregation’, Comput. Mater. Continua, vol. 75, no. 2, pp. 4273–4291, 2023, doi: 10.32604/cmc.2023.036975.
[37] C. P. Lau, J. Liu, H. Souri, W.-A. Lin, S. Feizi, and R. Chellappa, ‘Interpolated Joint Space Adversarial Training for Robust and Generalizable Defenses’, IEEE Trans Pattern Anal Mach Intell, vol. 45, no. 11, pp. 13054–13067, 2023, doi: 10.1109/TPAMI.2023.3286772.
[38] Z.-Z. Wu et al., ‘Domain Adaptation via Feature Disentanglement for cross-domain image classification’, Appl. Soft Comput., vol. 172, 2025, doi: 10.1016/j.asoc.2025.112868.
[39] C. Li et al., ‘Domain generalization on medical imaging classification using episodic training with task augmentation’, Comput. Biol. Med., vol. 141, 2022, doi: 10.1016/j.compbiomed.2021.105144.
[40] Q. Han et al., ‘DM-CNN: Dynamic Multi-scale Convolutional Neural Network with uncertainty quantification for medical image classification’, Comput. Biol. Med., vol. 168, 2024, doi: 10.1016/j.compbiomed.2023.107758.
[41] S. Wang, H. Ben, Y. Hao, X. He, and M. Wang, ‘Boosting Hyperspectral Image Classification with Dual Hierarchical Learning’, ACM Trans. Multimedia Comput. Commun. Appl., vol. 19, no. 1, 2023, doi: 10.1145/3522713.
[42] C. Ningthoujam, T. C. Chinghtham, B. Brahma, and A. K. Bhoi, ‘Hybrid CNN-KNN Model for Image Annotation: Combining Deep Learning and Instance-Based Learning’, J. Inst. Eng. Ser. B, 2025, doi: 10.1007/s40031-025-01239-8.
[43] L. Yan, Y. Ye, C. Wang, and Y. Sun, ‘LocMix: local saliency-based data augmentation for image classification’, Signal Image Video Process., vol. 18, no. 2, pp. 1383–1392, 2024, doi: 10.1007/s11760-023-02852-0.
[44] X. Xiong et al., ‘VariMix: A variety-guided data mixing framework for explainable medical image classifications’, Computer Methods and Programs in Biomedicine, vol. 271, p. 109016, Nov. 2025, doi: 10.1016/j.cmpb.2025.109016.
[45] T. Li et al., ‘Salient Features Guided Augmentation for Enhanced Deep Learning Classification in Hematoxylin and Eosin Images’, Comput. Mater. Continua, vol. 84, no. 1, pp. 1711–1730, 2025, doi: 10.32604/cmc.2025.062489.
[46] K. Sun, Y. Yin, F. Dong, and X. Sun, ‘Hyperspectral classification method based on M-ResHSDC’, Multimedia Tools Appl, vol. 83, no. 16, pp. 49767–49785, 2024, doi: 10.1007/s11042-023-17515-y.
[47] B. Wang, X. Hu, C. Zhang, P. Li, and P. S. Yu, ‘Hierarchical GAN-Tree and Bi-Directional Capsules for multi-label image classification’, Knowl Based Syst, vol. 238, 2022, doi: 10.1016/j.knosys.2021.107882.
[48] B. Gholami, Q. Liu, M. El-Khamy, and J. Lee, ‘Multiexpert Adversarial Regularization for Robust and Data-Efficient Deep Supervised Learning’, IEEE Access, vol. 10, pp. 85080–85094, 2022, doi: 10.1109/ACCESS.2022.3196780.
[49] D. Song, Y. Tang, B. Wang, J. Zhang, and C. Yang, ‘Two-Branch Generative Adversarial Network With Multiscale Connections for Hyperspectral Image Classification’, IEEE Access, vol. 11, pp. 7336–7347, 2023, doi: 10.1109/ACCESS.2022.3232152.
[50] D. Mukherkjee, P. Saha, D. Kaplun, A. Sinitca, and R. Sarkar, ‘Brain tumor image generation using an aggregation of GAN models with style transfer’, Sci Rep, vol. 12, no. 1, p. 9141, Jun. 2022, doi: 10.1038/s41598-022-12646-y.
[51] Z. Liao, S. Hu, Y. Zhang, and Y. Xia, ‘Unleashing the potential of open-set noisy samples against label noise for medical image classification’, Med. Image Anal., vol. 105, 2025, doi: 10.1016/j.media.2025.103702.
[52] C. Zhang, X. Zhang, and D. Tu, ‘A Set of Comprehensive Evaluation System for Different Data Augmentation Methods’, Mob. Inf. Sys., vol. 2022, 2022, doi: 10.1155/2022/8572852.
[53] S. Cheng, P. Li, K. Han, Y. Zheng, H. Xu, and Y. Yao, ‘DMFP: Dynamic multiscale feature perturbations for transferable adversarial attacks’, Knowl Based Syst, vol. 330, 2025, doi: 10.1016/j.knosys.2025.114469.
[54] P. Huang, Z. Yang, W. Wang, and F. Zhang, ‘Denoising Low-Rank Discrimination based Least Squares Regression for image classification’, Inf Sci, vol. 587, pp. 247–264, 2022, doi: 10.1016/j.ins.2021.12.031.
[55] Z. Yang, D. Wang, P. Huang, M. Wan, and G. Yang, ‘Regularisation constrained denoising discriminant least squares regression for image classification’, Expert Sys Appl, vol. 252, 2024, doi: 10.1016/j.eswa.2024.124253.
[56] J. Guo, H. Wang, X. Xue, M. Li, and Z. Ma, ‘Real-time classification on oral ulcer images with residual network and image enhancement’, IET Image Proc., vol. 16, no. 3, pp. 641–646, 2022, doi: 10.1049/ipr2.12144.
[57] M. Aamir, Z. Rahman, W. Ahmed Abro, U. Aslam Bhatti, Z. Ahmed Dayo, and M. Muhammad, ‘Brain tumor classification utilizing deep features derived from high-quality regions in MRI images’, Biomed. Signal Process. Control, vol. 85, 2023, doi: 10.1016/j.bspc.2023.104988.
[58] S. Iqbal, A. N. Qureshi, K. Aurangzeb, M. Alhussein, S. I. Haider, and I. Rida, ‘AMIAC: adaptive medical image analyzes and classification, a robust self-learning framework’, Neural Comput. Appl., vol. 37, no. 25, pp. 20451–20479, 2025, doi: 10.1007/s00521-023-09209-1.
[59] N. Abdel Samee, E. H. Houssein, E. Saber, G. Hu, and M. Wang, ‘Integrated deep learning-based IRACE and convolutional neural networks for chest X-ray image classification’, Knowl Based Syst, vol. 329, 2025, doi: 10.1016/j.knosys.2025.114293.
[60] D. Xue, J. Huang, R. Zhou, Y. Tai, and J. Zhang, ‘Secured COVID-19 CT image classification based on human-centric IoT and vision transformer’, J. Ambient Intell. Humanized Comput., 2024, doi: 10.1007/s12652-024-04797-9.
[61] X. Hu, S. Wen, and H. R. Karimi, ‘An improved algorithm for deep convolutional neural network structures based on randomness’, Inf Sci, vol. 713, 2025, doi: 10.1016/j.ins.2025.122162.
[62] Y. Tai, Y. Tan, E. Zou, B. Lei, Q. Fan, and Y. He, ‘Where to model the epistemic uncertainty of Bayesian convolutional neural networks for classification’, Neurocomputing, vol. 583, 2024, doi: 10.1016/j.neucom.2024.127568.
[63] M. Shi, X. Zeng, J. Ren, and Y. Shi, ‘A multi-scale residual capsule network for hyperspectral image classification with small training samples’, Multimedia Tools Appl, vol. 82, no. 26, pp. 40473–40501, 2023, doi: 10.1007/s11042-023-15017-5.
[64] X. Li, C. Zhao, X. Deng, and W. Jiang, ‘VTFR-AT: Adversarial Training With Visual Transformation and Feature Robustness’, IEEE Trans. Emerging Topics Comp. Intell., vol. 8, no. 4, pp. 3129–3140, 2024, doi: 10.1109/TETCI.2024.3370004.
[65] S. Yuan, Y. Chen, C. Ye, M. W. Bhatt, M. Sardeshmukh, and M. S. Hossain, ‘Cross-modal multi-label image classification modeling and recognition based on nonlinear’, Nonlenier Eng., vol. 12, no. 1, 2023, doi: 10.1515/nleng-2022-0194.
[66] T. K. Dutta, D. R. Nayak, and Y.-D. Zhang, ‘ARM-Net: Attention-guided residual multiscale CNN for multiclass brain tumor classification using MR images’, Biomed. Signal Process. Control, vol. 87, 2024, doi: 10.1016/j.bspc.2023.105421.
[67] L. Zhu et al., ‘Bayes-CAL: Robust Cross-Modal Alignment by Bayesian Approach for Few-Shot OoD Generalization’, Int J Comput Vision, vol. 133, no. 10, pp. 7076–7109, 2025, doi: 10.1007/s11263-025-02527-y.
[68] H. Wang et al., ‘Optimized lightweight CA-transformer: Using transformer for fine-grained visual categorization’, Ecol. Informatics, vol. 71, 2022, doi: 10.1016/j.ecoinf.2022.101827.
[69] Z. Jin and Y. Wei, ‘UMPA: Unified multi-modal prompt with adapter for vision-language models’, Multimedia Syst, vol. 31, no. 2, 2025, doi: 10.1007/s00530-025-01707-7.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Budiman Budiman, Arief Setyanto, Andi Sunyoto, Dhani Ariatmanto
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
-a.jpg)
-b.jpg)
