Decomposition Methods for Machine Learning with Small, Incomplete or Noisy Datasets
Open accessThe DOI matches its Crossref record; 66 of the 66 references it cites were found in a registry or an index.
Abstract
In many machine learning applications, measurements are sometimes incomplete or noisy resulting in missing features. In other cases, and for different reasons, the datasets are originally small, and therefore, more data samples are required to derive useful supervised or unsupervised classification methods. Correct handling of incomplete, noisy or small datasets in machine learning is a fundamental and classic challenge. In this article, we provide a unified review of recently proposed methods based on signal decomposition for missing features imputation (data completion), classification of noisy samples and artificial generation of new data samples (data augmentation). We illustrate the application of these signal decomposition methods in diverse selected practical machine learning examples including: brain computer interface, epileptic intracranial electroencephalogram signals classification, face recognition/verification and water networks data analysis. We show that a signal decomposition approach can provide valuable tools to improve machine learning performance with low quality datasets.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Donoho, D. L. (2006) · IEEE Transactions on Information Theory · doi:10.1109/tit.2006.871582 · Cited by 23,994
Huang, N. E., Shen, Z., Long, S. R., Wu, M. C., Shih, H. H., Zheng, Q., Yen, N.-C., Tung, C. C., & Liu, H. H. (1998) · Proceedings of the Royal Society of London. Series A: Mathematical, Physical and Engineering Sciences · doi:10.1098/rspa.1998.0193 · Cited by 23,564 · cited 1×
“in [45] as a new method to analyze nonlinear and non-stationary signals.”
MethodLIII. On lines and planes of closest fit to systems of points in space
Pearson, K. (1901) · The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science · doi:10.1080/14786440109462720 · Cited by 11,934
Tensor Decompositions and Applications
Kolda, T. G., & Bader, B. W. (2009) · SIAM Review · doi:10.1137/07070111x · Cited by 10,301 · cited 2×
“As a particular case, when core tensor is diagonal with R = R1 = R2 = R3 this model is reduced to the CANDECOMP/PARAFAC or CP(R) decomposition model (see Figure 1f), which has demonstrated to be very useful in a wide range of applications [47,48].”
BackgroundMethodAn Introduction To Compressive Sampling
Candes, E. J., & Wakin, M. B. (2008) · IEEE Signal Processing Magazine · doi:10.1109/msp.2007.914731 · Cited by 10,017 · cited 1×
“Sparse coding has to date a large record of successful applications in signal processing tasks like compressed sensing [41,42], blind source separation [43], denoising [39], inpainting [44] and others.”
BackgroundSignal Recovery From Random Measurements Via Orthogonal Matching Pursuit
Tropp, J. A., & Gilbert, A. C. (2007) · IEEE Transactions on Information Theory · doi:10.1109/tit.2007.909108 · Cited by 9,653 · cited 4×
“TK (Tucker) and CP (CANDECOMP/PARAFAC) indicate the decomposition model used by the algorithms.”
MethodInfluentialImage Denoising Via Sparse and Redundant Representations Over Learned Dictionaries
Elad, M., & Aharon, M. (2006) · IEEE Transactions on Image Processing · doi:10.1109/tip.2006.881969 · Cited by 5,321
Atomic Decomposition by Basis Pursuit
Chen, S. S., Donoho, D. L., & Saunders, M. A. (2001) · SIAM Review · doi:10.1137/s003614450037906x · Cited by 5,085
ADASYN: Adaptive synthetic sampling approach for imbalanced learning
He, H., Bai, Y., Garcia, E. A., & Li, S. (2008) · 2008 IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence) · doi:10.1109/ijcnn.2008.4633969 · Cited by 4,521 · cited 2×
“Akter et al. [56] successfully applied data augmentation based on an adaptive synthetic oversampling approach (ADASYN) [66] to balance an in-hospital dataset of epileptic interictal iEEG signals to identify the seizure onset zones automatically.”
MethodCoSaMP: Iterative signal recovery from incomplete and inaccurate samples
Needell, D., & Tropp, J. A. (2009) · Applied and Computational Harmonic Analysis · doi:10.1016/j.acha.2008.07.002 · Cited by 4,326
Research works printed but not linked: 4
- 1. Goodfellow, I.; Bengio, Y.; Courville, A. Deep Learning; MIT Press: Cambridge, MA, USA, 2016.
- 68. Tomasi, C.; Manduchi, R. Bilateral filtering for gray and color images. In Proceedings of the Sixth International Conference on Computer Vision, Washington, DC, USA, 4-7 January 1998; pp. 839-846.
- 69. Gonzalez, R.C.; Woods, R.E. Digital Image Processing; Prentice Hall: Upper Saddle River, NJ, USA, 2008.
- 70. Tukey, J.W. Exploratory Data Analysis; Addison-Wesley: Reading, MA, USA, 1977.
Cited by
Machine Learning for Perovskite Solar Cells and Component Materials: Key Technologies and Prospects
Liu, Y., Tan, X., Liang, J., Han, H., Xiang, P., & Yan, W. (2023) · Advanced Functional Materials · doi:10.1002/adfm.202214271 · Cited by 174
A Comprehensive Survey of Machine Learning Methodologies with Emphasis in Water Resources Management
Drogkoula, M., Kokkinos, K., & Samaras, N. (2023) · Applied Sciences · doi:10.3390/app132212147 · Cited by 209
“Our findings are depicted in the PRISMA diagram (Figure 3).”
MethodA Comparison of Machine Learning Methods to Forecast Tropospheric Ozone Levels in Delhi
Juarez, E. K., & Petersen, M. (2021) · Atmosphere · doi:10.3390/atmos13010046 · Cited by 64
“Additional preprocessing could also compensate for missing data with decomposition methods [43].”
MethodTower crane safety technologies: A synthesis of academic research and industry insights
Ali, A. H., Zayed, T. M., Wang, R. D., & Kit, M. Y. S. (2024) · Automation in Construction · doi:10.1016/j.autcon.2024.105429 · Cited by 58
Machine Learning Methods with Noisy, Incomplete or Small Datasets
Caiafa, C. F., Sun, Z., Tanaka, T., Martí-Puig, P., & Solé‐Casals, J. (2021) · Applied Sciences · doi:10.3390/app11094132 · Cited by 39
“In [1], the EMD method was applied to remove noise in face images, thus improving the classification accuracy of a machine learning classifier.”
BackgroundMethodSelf-citationWang, S., Liu, C., Hao, W., Zhuang, Y., Zhu, X., Wang, L., Niu, X., Liu, S., Chen, B., & Zhao, Q. (2025) · ACS Nano · doi:10.1021/acsnano.5c07494 · Cited by 39
Low-Quality Video Target Detection Based on EEG Signal Using Eye Movement Alignment
Shi, J., Bi, L., Xu, X., Feleke, A. G., & Fei, W. (2024) · Cyborg and Bionic Systems · doi:10.34133/cbsystems.0121 · Cited by 32
Machine learning as a characterization method for analysis and design of perovskite solar cells
Liu, Y., Tan, X., Xiang, P., Tu, Y., Shao, T., Zang, Y., Li, X., & Yan, W. (2024) · Materials Today Physics · doi:10.1016/j.mtphys.2024.101359 · Cited by 29
Machine learning-assisted design of high-performance perovskite photodetectors: a review
Li, X. F., Mai, Y., Lan, C., Yang, F., Zhang, P., & Li, S. (2024) · Advanced Composites and Hybrid Materials · doi:10.1007/s42114-024-01113-z · Cited by 33
“Data collection is a key step in the ML workflow, which determines the quality of the data underlying the model and the effectiveness of the subsequent work [35, 36].”
Wang, J., Jia, J.-N., Sun, S., & Zhang, T. (2021) · Engineering Fracture Mechanics · doi:10.1016/j.engfracmech.2021.108160 · Cited by 17
Soize, C., & Ghanem, R. (2023) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2023.116498 · Cited by 36
Evaluating the effectiveness of decomposed Halstead Metrics in software fault prediction
Khan, B. A., & Nadeem, A. (2023) · PeerJ Computer Science · doi:10.7717/peerj-cs.1647 · Cited by 18
“The decomposition enables ML algorithms to comprehend features and thus improves model performance by uncovering potential information ( Caiafa et al., 2020 ).”
BackgroundHe, Z., Naphade, S., & Huang, T.-H. (2025) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706598.3714319 · Cited by 26
Rigorous validation of machine learning in laboratory medicine: guidance toward quality improvement
Miller, H. A., & Valdes, R. (2025) · Critical Reviews in Clinical Laboratory Sciences · doi:10.1080/10408363.2025.2488842 · Cited by 15
Exploring Effects of Modified Machine Learning Pipelines of Astrochemical Inventories
Shay, H. T., Scolati, H. N., Wenzel, G., Lee, K. L. K., Marimuthu, A. N., & McGuire, B. A. (2025) · The Astrophysical Journal · doi:10.3847/1538-4357/adc80b · Cited by 9
“This con fl uence of both a small and diverse input data set makes for a model that is extremely susceptible to even small changes or additions ( C. F. Caiafa et al. 2020 ) .”
Drogkoula, M., Samaras, N., Iatrellis, O., Nathanail, E., & Kokkinos, K. (2025) · Discover Sustainability · doi:10.1007/s43621-025-01832-3 · Cited by 9
Solar Generated Hydrogen: 2D Heterostructures for Photocatalysts
Shoaib, S., Arshad, F., Sun, Z., Wu, Z., Li, Z., Lu, X., Guo, W., & Chen, Z. (2026) · ACS Applied Nano Materials · doi:10.1021/acsanm.5c05371 · Cited by 5
Qiao, Y., Eisfeld, C., von Eisenhart‐Rothe, R., & Hinterwimmer, F. (2026) · BMC Medical Informatics and Decision Making · doi:10.1186/s12911-026-03401-8 · Cited by 5
“Currently, in addition to multicenter collaborations to expand patient samples, the development of ML algorithms for small samples is an approach that could be useful to apply to diseases such as bone tumors, which have a low incidence and prevalence [98, 99].”
The Use of Additive Decomposition and Deep Neural Network for Photovoltaic Power Forecasting
Dehghan, F., Moghaddam, M. P., & Imani, M. (2023) · doi:10.1109/icee59167.2023.10334838 · Cited by 3
Lammers, R., Jansma, S. R., Veldkamp, B. P., Machens, A., de Visser, M., & Gosselt, J. F. (2023) · Sustainability · doi:10.3390/su152216106 · Cited by 3
Small, imbalanced, and noisy datasets in healthcare research: A systematic review
Ramos, J., Pais-Vieira, M., & Brás, S. (2026) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2026.115472 · Cited by 4
A robust denoising diffusion framework for completing missing regions of multiple biomedical signals
Wang, S., Tang, H., Himeno, R., Solé‐Casals, J., Caiafa, C. F., Aoki, S., & Sun, Z. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2025.108788 · Cited by 1
Self-citationMachine Learning Methods with Noisy, Incomplete or Small Datasets
Solé-Casals, J., Sun, Z., Caiafa, C. F., Martí-Puig, P., & Tanaka, T. (2021) · MDPI eBooks · doi:10.3390/books978-3-0365-1288-4 · Cited by 1
Self-citationGraph Empirical Mode Decomposition-Based Data Augmentation Applied to Gifted Children MRI Analysis
Chen, X., Li, B., Jia, H., Feng, F., Duan, F., Sun, Z., Caiafa, C. F., & Solé‐Casals, J. (2022) · Frontiers in Neuroscience · doi:10.3389/fnins.2022.866735 · Cited by 1
“The training model needs large amount of data to avoid overfitting (Caiafa et al., 2020).”
BackgroundSelf-citationSchneider, J., Eisenhardt, D., Utama, C., & Meske, C. (2025) · Lecture notes in information systems and organisation · doi:10.1007/978-3-031-80122-8_5 · Cited by 1
Nguyen, N. A. T., Tao, Q.-B., Yang, H.-J., Huynh, H. T., Vinh, N. T. Q., & Ninh, D. K. (2024) · IEEE Access · doi:10.1109/access.2024.3393413
Ramiotis, G., & Mania, K. (2025) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc58623.2025.11251707
“However, motor imagery suffers from long training sessions that cause user fatigue, thus, producing unidentifiable EEG patterns [5], [6] degrading classification performance.”
Multi-stage Large Segment Missing Value Filling Algorithm Based on Distribution Measure
Zhang, Y., Li, Y. N., & Pei, Y. (2025) · International Symposium on AI and Cybersecurity (ISAICS) · doi:10.1109/isaics66888.2025.11349880
“With the advancement of data mining techniques, machine learning has become crucial for time series analysis [7], though most algorithms require complete datasets [8].”
Ichsan, D. K., Hidayat, M. I. P., Pramata, A. D., & Hantoro, R. (2026) · Results in Engineering · doi:10.1016/j.rineng.2026.112383
Metadata
- Authors
- Caiafa, C. F., Solé-Casals, J., Marti-Puig, P., Zhe, S., & Tanaka, T.
- Affiliations
- Consejo Nacional de Investigaciones Científicas y Técnicas (AR)
- Data and Signal Processing Research Group, University of Vic-Central University of Catalonia, 08500 Vic, Catalonia, Spain
- Computational Engineering Applications Unit, Head Office for Information Systems and Cybersecurity, RIKEN, Wako-Shi 351-0198, Japan
- Tokyo University of Agriculture and Technology (JP)
- Corresponding author
- Caiafa, Cesar Federico
- Venue
- Applied Sciences
- Publisher
- MDPI AG
- Published
- 2020-11-27
- Volume
- 10
- Issue
- 23
- Pages
- 8481
- ISSN
- 2076-3417 (ISSN-L)
- OpenAlex
- W3109443741
- Cited by
- 33
- References
- 66
- Citations by year
- 2026: 4 · 2025: 9 · 2024: 5 · 2023: 10 · 2022: 1 · 2021: 4
- FWCI
- 1.47
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Applied Sciences (publishedVersion, cc-by)
- LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas) (submittedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas) (submittedVersion, cc-by-nc-sa)
- CONICET Digital (CONICET) (submittedVersion, cc-by-nc-sa)
- Funders
- Japan Science and Technology Agency (JPMJCR1784)
- University of Vic (R0947)
- European Cooperation in Science and Technology (CA18106)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
- Physical Sciences › Computer Science › Artificial Intelligence › Neural Networks and Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Imbalanced Data Classification Techniques
- Keywords
- signal decomposition, missing data imputation, noisy data classification, data augmentation, small datasets, missing data, brain-computer interface, epileptic EEG, face recognition, face verification, water networks
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.