A Survey of Handwritten Character Recognition with MNIST and EMNIST
Open accessThe DOI matches its Crossref record; 84 of the 84 references it cites were found in a registry or an index.
Abstract
This paper summarizes the top state-of-the-art contributions reported on the MNIST dataset for handwritten digit recognition. This dataset has been extensively used to validate novel techniques in computer vision, and in recent years, many authors have explored the performance of convolutional neural networks (CNNs) and other deep learning techniques over this dataset. To the best of our knowledge, this paper is the first exhaustive and updated review of this dataset; there are some online rankings, but they are outdated, and most published papers survey only closely related works, omitting most of the literature. This paper makes a distinction between those works using some kind of data augmentation and works using the original dataset out-of-the-box. Also, works using CNNs are reported separately; as they are becoming the state-of-the-art approach for solving this problem. Nowadays, a significant amount of works have attained a test error rate smaller than 1% on this dataset; which is becoming non-challenging. By mid-2017, a new dataset was introduced: EMNIST, which involves both digits and letters, with a larger amount of data acquired from a database different than MNIST’s. In this paper, EMNIST is explained and some results are surveyed.
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
Figures
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.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Although the images are not of a real-world environment, they can be an appropriate alternative considering that the main objective of this study was to explore the possibility of deep learning for the objective quantification of the visual perception of the substituted signal.”
From Deep Learning-Based Optimization of Visual–Auditory Sensory Substitution (2023) · cites this paper as [31]
Experimental — accuracy not yet measured
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
Gradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805 · cited 7×
“For instance, Belongie et al. [5] achieved 0.63% and Keysers et al. [6] achieved 0.54% and 0.52% using K-NN, Kégl and Busa-Fekete [7] achieved 0.87% using boosted stumps on Haar features, LeCun et al. [4] achieved 0.8% and Decoste and Schölkopf [8] attained results from 0.56 to 0.68% using SVM.”
BackgroundMethodInfluentialExtreme learning machine: Theory and applications
Huang, G.-B., Zhu, Q.-Y., & Siew, C.-K. (2006) · Neurocomputing · doi:10.1016/j.neucom.2005.12.126 · Cited by 13,285 · cited 1×
“In 2015, McDonnell et al. [21] proposed an approach using the ‘extreme learning machine’ (ELM) [22] algorithm to train shallow non-convolutional neural networks, and they agree on the fact that data augmentation with distortions can reduce error rates even further.”
MethodShape matching and object recognition using shape contexts
Belongie, S., Malik, J., & Puzicha, J. (2002) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/34.993558 · Cited by 6,306 · cited 2×
“For instance, Belongie et al. [5] achieved 0.63% and Keysers et al. [6] achieved 0.54% and 0.52% using K-NN, Kégl and Busa-Fekete [7] achieved 0.87% using boosted stumps on Haar features, LeCun et al. [4] achieved 0.8% and Decoste and Schölkopf [8] attained results from 0.56 to 0.68% using SVM.”
MethodInfluentialDynamic Routing Between Capsules
Sabour, S., Frosst, N., & Hinton, G. E. (2017) · arXiv · doi:10.48550/arxiv.1710.09829 · Cited by 4,167 · cited 1×
“An interesting work using capsule layers (a concept recently introduced by Sabour et al. [76]) is TextCaps [77], a CNN with 3 convolutional layers and two capsule layers, which obtain accuracies of 95.36% in EMNIST Letters and 99.79% in EMNIST Digits.”
MethodInfluentialMulti-column deep neural networks for image classification
Ciresan, D., Meier, U., & Schmidhuber, J. (2012) · 2012 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2012.6248110 · Cited by 3,720
Best practices for convolutional neural networks applied to visual document analysis
Simard, P. Y., Steinkraus, D., & Platt, J. C. (2005) · Seventh International Conference on Document Analysis and Recognition, 2003. Proceedings. · doi:10.1109/icdar.2003.1227801 · Cited by 2,741 · cited 2×
“87% NN 2-layer 800 hidden units, MSE loss [9] 0.”
BackgroundSalakhutdinov, R., & Hinton, G. E. (2009) · International Conference on Artificial Intelligence and Statistics · Cited by 2,605
Lee, H., Grosse, R., Ranganath, R., & Ng, A. Y. (2009) · Proceedings of the 26th Annual International Conference on Machine Learning · doi:10.1145/1553374.1553453 · Cited by 2,412 · cited 2×
“Lee et al. [54] described convolutional deep belief networks with probabilistic max-pooling achieving an error rate of 0.82%.”
BackgroundInfluentialWhat is the best multi-stage architecture for object recognition?
Jarrett, K., Kavukcuoglu, K., Ranzato, M. A., & LeCun, Y. (2009) · 2009 IEEE 12th International Conference on Computer Vision · doi:10.1109/iccv.2009.5459469 · Cited by 2,153 · cited 3×
“…[39] 0.47% CNN (3 conv, 1 dense), stochastic pooling [25] 0.49% CNN (2 conv, 1 dense, relu) with dropout [24] 0.52% CNN, unsup pretraining [17] 0.53% CNN (2 conv, 1 dense, relu) with DropConnect [24] 0.57% SparseNet + SVM [15] 0.59% CNN (2 conv, 1 dense), unsup pretraining [16] 0.60%…”
MethodInfluential(2013) · Cited by 2,056
Research works printed but not linked: 5
- 23. Kasun, L.L.C.; Zhou, H.; Huang, G.B.; Vong, C.M. Representational learning with extreme learning machine for big data. IEEE Intell. Syst. 2013 , 28 , 31-34.
- 39. Lin, M.; Chen, Q.; Yan, S. Network In Network. In Proceedings of the 2nd International Conference on Learning Representations, Banff, AB, Canada, 14-16 April 2014.
- 42. Calderón, A.; Roa-Valle, S.; Victorino, J. Handwritten digit recognition using convolutional neural networks and Gabor filters. In Proceedings of the 2003 International Conference on Computational Intelligence, Cancun, Mexico, 19-21 May 2003.
- 73. Peng, Y.; Yin, H. Markov Random Field Based Convolutional Neuralx Networks for Image Classification. In IDEAL 2017: Intelligent Data Engineering and Automated Learning ; Lecture Notes in Computer Science; Yin, H., Gao, Y., Chen, S., Wen, Y., Cai, G., Gu, T., Du, J., Tallón-Ballesteros, A., Zhang, M., Eds.; Springer: Guilin, China, 2017; Volume 10585, pp. 387-396.
- 75. Mor, S.S.; Solanki, S.; Gupta, S.; Dhingra, S.; Jain, M.; Saxena, R. Handwritten text recognition: With deep learning and Android. Int. J. Eng. Adv. Technol. 2019 , 8 , 172-178.
Cited by
Items 101–200 of 307
Handwritten Digit Recognition using Deep Learning Networks
Yu, S., Liu, J., Shu, H., & Cheng, Z. (2022) · IEEE Conference on Telecommunications, Optics and Computer Science (TOCS) · doi:10.1109/tocs56154.2022.10016012 · Cited by 6
“This article [12] describes information about MNIST with its extended EMNIST dataset and CIFAR-100.”
BackgroundYılmaz, Y. (2023) · Neural Computing and Applications · doi:10.1007/s00521-023-08995-y · Cited by 6
“The important metrics used to evaluate the adopted hard-ware-based learning approaches using optimisation meth-ods are accuracy, test error rate, i.e. misclassification rate, along with precision, sensitivity, and specificity [41].”
BackgroundOn the recognition of weakly blurred, highly contrasting objects by neural networks
Tuliabaeva, D., Tumakov, D. N., & Elshin, L. A. (2024) · International Conference on Optics, Computer Applications, and Materials Science (CMSD) · doi:10.1117/12.3024891 · Cited by 6
“The problem of recognizing objects such as handwritten letters or digits is often encountered in evaluating the performance of various CNN models [4].”
BackgroundImage feature learning with a genetic programming autoencoder
Ruberto, S., Terragni, V., & Moore, J. H. (2020) · Genetic and Evolutionary Computation Conference Companion (GECCO Companion) · doi:10.1145/3377929.3389981 · Cited by 6
“MNIST is widely-used as a standard benchmark in the ML community [1].”
MethodGangopadhyay, B., Hazra, S., & Dasgupta, P. (2021) · Pattern Recognition Letters · doi:10.1016/j.patrec.2021.10.004 · Cited by 6
Handwritten Digit Recognition with Flood Simulation and Topological Feature Extraction
Brociek, R., Pleszczyński, M., Błaszczyk, J., Czaicki, M., & Napoli, C. (2025) · Entropy · doi:10.3390/e27121218 · Cited by 5
“Support vector machines (SVMs) have shown strong performance with test error rates ranging from 0.6% to 1.4% depending on the kernel and preprocessing techniques used [16,17].”
InfluentialGhifary, M., Nugroho, M. A. A., Barokah, B., Purwarianti, A., Birowo, C. T., Machdi, I., Gaol, F. L., Purnama, I. K. E., Riandi, O., Azhar, T. N., et al. (2023) · Penerbit BRIN eBooks · doi:10.55981/brin.668 · Cited by 7
Privacy-Preserving Serverless Edge Learning With Decentralized Small-Scale Mobile Data
Lin, S., Lin, C.-H., & Lee, M. (2023) · IEEE Network · doi:10.1109/mnet.135.2200611 · Cited by 5
“MNIST handwriting image classification task and MNIST-fashion clothing image classification task [13] are implemented in distributed deep learning scenarios to obtain the simulation results in this section.”
MethodAutomated Measurement of Air Bubbles Dispersion in Ice Cream Using Machine Learning Methods
Korolev, I. A. (2023) · Food Processing Techniques and Technology · doi:10.21603/2074-9414-2023-3-2448 · Cited by 5
Optimized convolutional neural network deep learning for Arabian handwritten text recognition
Ritonga, M., Bangare, M. L., Bangare, P. M., Bangare, S. L., Vanjire, S. S., Moholkar, K., Kasat, K., & Rozak, P. (2025) · Bulletin of Electrical Engineering and Informatics · doi:10.11591/eei.v14i2.7696 · Cited by 5
Number Recognition Through Color Distortion Using Convolutional Neural Networks
Henshaw, C., Dennis, J., Nadzam, J., & Michaels, A. J. (2025) · Computers · doi:10.3390/computers14020034 · Cited by 5
Fischer-Brandies, L., Müller, L., Rebholz, B., & Buettner, R. (2025) · IEEE Access · doi:10.1109/access.2025.3556582 · Cited by 5
“Thus, it is possible to apply standard machine learning approaches such as CNN [16], [17], support vector machines [18], [19], [20], Convolutional-recurrent neural network [21], [22], Restricted Boltzmann Machines [23], Hidden Markov Models [24] or single-shot detector [25] within OCR.”
Efficient and Robust Indian Number Plate Recognition Through Modified and Tuned LPRNet
Dangi, R. S., Kuvelkar, A., Maity, S. K., & Wandhekar, S. (2021) · Lecture notes in electrical engineering · doi:10.1007/978-981-16-3690-5_12 · Cited by 5
KurdSet: A Kurdish Handwritten Characters Recognition Dataset Using Convolutional Neural Network
Ali, S. H., & Abdulrazzaq, M. B. (2024) · Computers, materials & continua/Computers, materials & continua (Print) · doi:10.32604/cmc.2024.048356 · Cited by 5
A new SVM solver applied to Skin Lesion Classification
SILVA, J. D. C., Alves, A., Santos, P. S. M., & Matioli, L. C. (2024) · Statistics Optimization & Information Computing · doi:10.19139/soic-2310-5070-2005 · Cited by 6
Thamaraiselvan, S., Venugopal, V., & Vekkot, S. (2024) · International Conference on System Modeling & Advancement in Research Trends (SMART) · doi:10.1109/smart63812.2024.10882523 · Cited by 5
A methodology for character recognition and revision of the linear equations solving procedure
Neri, M. C. G., Villegas, O. O. V., Sánchez, V. G. C., Domínguez, H. D. J. O., Nandayapa, M., & Sossa, H. (2022) · Information Processing & Management · doi:10.1016/j.ipm.2022.103088 · Cited by 7
Low-Cost Malware Detection with Artificial Intelligence on Single Board Computers
Steadman, P., Jenkins, P., Rathore, R. S., & Hewage, C. T. E. R. (2026) · Future Internet · doi:10.3390/fi18010046 · Cited by 5
A neural network with competitive layers for character recognition
Goltsev, A., & Gritsenko, V. A. (2022) · ELCVIA Electronic Letters on Computer Vision and Image Analysis · doi:10.5565/rev/elcvia.1392 · Cited by 4
“There is a more recent overview of handwritten character recognition of the MNIST database in [13].”
BackgroundText extraction and recognition method for license plates
Moussaoui, H., Akkad, N. E., & Benslimane, M. (2023) · E3S Web of Conferences · doi:10.1051/e3sconf/202346900069 · Cited by 4
“In order to detect and recognize a text from images, a variety of devices support optical character recognition ([1], [2], [3]), or OCR which refers to Optical Character Recognition ([4], [5]).”
BackgroundAn efficient parallel DCNN algorithm in big data environment
Mao, Y., Nanehkaran, Y. A., Chandrasekaran, N., Huo, Y., Wen, Z. Q., Gong, K., & Miao, D. (2025) · PeerJ Computer Science · doi:10.7717/peerj-cs.2871 · Cited by 4
“To evaluate the effectiveness of the PDCNN-MI algorithm, four datasets from different domains were used in the experiments: EMNIST-Balanced ( Baldominos, Saez & Isasi, 2019 ), CompCars ( Xu et al., 2022 ), ImageNet 1K ( Li et al., 2022 ), and COCO5 ( Lin et al., 2014 ).”
Object classification using a single-pixel camera and neural networks
Reutov, A. A., Валерьевич, Б. Д., & Васильевич, С. Д. (2025) · Computer Optics · doi:10.18287/2412-6179-co-1538 · Cited by 4
Becker, D., & Klonowski, J. (2023) · Drones · doi:10.3390/drones7020094 · Cited by 4
Short Words for Writer Identification Using Neural Networks
Κούκιου, Γ. (2023) · Applied Sciences · doi:10.3390/app13116841 · Cited by 4
Vivekrabinson, K., Jebaraj, B. S., Ragavan, K., Kumaresan, J., Dinesh, S. E. V., & Mareeswari, G. (2023) · International Conference on Research Methodologies in Knowledge Management, Artificial Intelligence and Telecommunication Engineering (RMKMATE) · doi:10.1109/rmkmate59243.2023.10369098 · Cited by 4
“The EMNIST (Extended MNIST) dataset is an extension of the popular MNIST dataset for handwritten character recognition [18].”
MethodStrategic data navigation: information value-based sample selection
Balogh, C. L., Pelenczei, B., Kővári, B., & Bécsi, T. (2024) · Artificial Intelligence Review · doi:10.1007/s10462-024-10813-3 · Cited by 4
“Deep Learning proved to be an outstanding tool for computer vision applications, starting with the MNIST (LeCun et al, 1998) dataset, that contains 60 , 000 images of 28 × 28 grayscale images (Baldominos et al, 2019).”
MethodKannada-Mnist Classification Using Skip CNN
Selasi, G. E., Nimo, B. D., & ASIAMAH, E. K. (2019) · International Computer Conference on Wavelet Active Media Technology and Information Processing (ICCWAMTIP) · doi:10.1109/iccwamtip47768.2019.9067521 · Cited by 4
“[6] outlined the various contributions reported on the MNIST dataset for handwritten digit recognition.”
BackgroundHandwritten Character Recognition using Neural Networks forBanking Applications
Mhalgi, S., Ganu, K., Marne, P., Phadke, R., & Shekapure, S. (2020) · International Journal of Computer Applications · doi:10.5120/ijca2020920296 · Cited by 5
“[19] Although these intensive tasks are handled much better with the advancements in the hardware like GPUs, the errors in handwritten text recognition differ to an extent and as a result, the outputs are averaged.”
BackgroundCharacter Recognition using Tesseract enabling Multilingualism
Niharika, G. L., Bano, S., Kumar, P. S., Deepika, T. J., & Thumati, H. (2020) · International Conference on Electronics, Communication and Aerospace Technology (ICECA) · doi:10.1109/iceca49313.2020.9297609 · Cited by 4
“The analysis to recognize the handwritten character recognition [1] using the Modern National Institute of Standards and Technology (MNIST) database [16].”
MethodMid-Layer Visualization in Convolutional Neural Network for Microstructural Images of Cast Irons
Lee, H.-J., Hwang, I.-K., Jeong, S.-J., Cho, I.-S., & Kim, H. (2021) · 대한금속재료학회지 · doi:10.3365/kjmm.2021.59.6.430 · Cited by 4
Design of Handwriting-based Text Interface for Support of Mobile Platform Education Contents
Cho, Y., Cho, S.-H., & Kim, J. (2021) · Journal of the Korea Computer Graphics Society · doi:10.15701/kcgs.2021.27.5.81 · Cited by 4
“이는 사용자가이해하기쉽고,상대적으로적은용량의데이터셋이기 때문에 학습, 분류 및 컴퓨터비전 관점에서 표준 모델로 활용된 다 [29, 30].”
MethodZhou, X., Jia, Q., & Xie, R. (2022) · ACM Annual International Conference on Mobile Computing and Networking (MobiCom) · doi:10.1145/3495243.3558248 · Cited by 4
“To demonstrate the superiority of NestFL across applications, we experiment with the MNIST [5] and CIFAR10 [6] image classification tasks and the HAR [7] human activity recognition tasks.”
MethodЧичкарьов, Є., Zinchenko, O. V., Балалаєва, О., Сергієнко, А. В., & Ковальов, О. (2023) · ГРААЛЬ НАУКИ · doi:10.36074/grail-of-science.23.12.2022.36 · Cited by 4
A mobility-aware federated learning coordination algorithm
Macedo, D., Santos, D. F. S., Perkusich, Â., & Valadares, D. C. G. (2023) · The Journal of Supercomputing · doi:10.1007/s11227-023-05372-3 · Cited by 4
“We considered an MNIST dataset for handwritten digit recognition [22].”
MethodEikonal depth: An optimal control approach to statistical depths
Molina-Fructuoso, M., & Murray, R. (2024) · Foundations of Data Science · doi:10.3934/fods.2024016 · Cited by 4
SyncNet: Harmonizing Nodes for Efficient Learning
Roy, A., Gyanchandani, B. S., Sahu, J., & Rao, K. M. (2024) · International Conference on Software, Telecommunications and Computer Networks (SoftCOM) · doi:10.23919/softcom62040.2024.10721994 · Cited by 3
“[9] The ImageNet [10] dataset is a large collection of images with metadata that is often employed to test and train computer vision algorithms.”
BackgroundMethodEnhancing Handwritten Character Recognition Using a Hybrid SVM-CNN Model
Fethi, M. R., Farhaoui, O., Alaoui, A. O., Zeroual, I., & Allaoui, A. E. (2025) · Lecture notes in networks and systems · doi:10.1007/978-3-031-90921-4_94 · Cited by 3
Feijoo‐Garcia, M. A., Zhang, Y., Gu, Y., Magana, A. J., Beneš, B., & Popescu, V. S. (2025) · Computer Applications in Engineering Education · doi:10.1002/cae.70084 · Cited by 3
“The modified National Institute of Standards and Technology (MNIST) data set is popularly used for machine learning or deep learning tasks [49, 50].”
Hopkins, M., Fil, J., Jones, E. G., & Furber, S. B. (2023) · Frontiers in Neuroinformatics · doi:10.3389/fninf.2023.1125844 · Cited by 3
“, 2017) with our basic mechanism, though more recent work has moved the achievable bound upwards by a few percent over ours (Baldominos et al., 2019) as can be seen from the CNN results here.”
ResultAli, A., Senan, N., & Murli, N. (2024) · Lecture notes in networks and systems · doi:10.1007/978-3-031-66965-1_33 · Cited by 3
Salman, O. S., & Salman, A. S. (2022) · Lecture notes in networks and systems · doi:10.1007/978-3-030-98015-3_17 · Cited by 3
Incremental Learning of Handwritten Characters in EMNIST dataset
Dana, S. K. (2023) · International Conference for Advancement in Technology (ICONAT) · doi:10.1109/iconat57137.2023.10080724 · Cited by 3
“The EMNIST dataset is an important dataset from application point of view [40] as it contains English uppercase and lowercase characters [a-z, A-Z] as well as the digits [0-9].”
MethodLi, F., Wang, S., & Zhu, Y. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10096448 · Cited by 3
“To empirically evaluate the proposed method, we conducted experiments on MNIST [19], CIFAR-10 [20], and Cal-tech101 [21] with the residual network [22] as the backbone DNN .”
MethodDeep Learning for Handwritten Character Recognition
Bhatt, C., Akanksha, A., Chauhan, R. S., Kumar, D., Singh, T., & Sharma, S. (2023) · International Conference on Sustainable Emerging Innovations in Engineering and Technology (ICSEIET) · doi:10.1109/icseiet58677.2023.10303610 · Cited by 3
Carboni, L., Nwaigwe, D., Mainsant, M., Bayle, R., Reyboz, M., Mermillod, M., Dojat, M., & Achard, S. (2025) · Neural Networks · doi:10.1016/j.neunet.2025.107125 · Cited by 5
“Different ANN architectures and algorithms of varying complexity have been proposed to tackle the classification of the MNIST dataset (Baldominos et al. [2019], Ciregan et al. [2012], Jarrett et al. [2009].”
Fully generated mammogram patch dataset using CycleGAN with de-identification texture analysis
Richmond, L., Trivedi, H., & Deshpande, P. (2024) · International Journal of Data Science and Analytics · doi:10.1007/s41060-024-00638-2 · Cited by 3
“This was inferior to validation accuracy results when training on a real dataset, in which precision exceeds 90% [17].”
Global and local combined contrastive learning for multi-view clustering
Gu, W., & Zhu, C. (2024) · Multimedia Systems · doi:10.1007/s00530-024-01512-8 · Cited by 3
“• NoisyMNIST The NoisyMNIST [39, 40] dataset is a variant based on the classic MNIST handwritten digits dataset extended and modified.”
Machine Learning for Sensor-Based Handwritten Character Recognition: A Brief Survey
Singh, S. K., & Chaturvedi, A. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-81404-4_21 · Cited by 3
Handwritten digit recognition based on SVM
Tang, W., & Li, F. (2024) · International Conference on Frontiers of Intelligent Manufacturing and Automation · doi:10.1145/3704558.3705521 · Cited by 3
A novel framework for generating handwritten datasets
Anwar, S., Mehrban, B., Ali, M., Hussain, F., & Halim, Z. (2020) · Multimedia Tools and Applications · doi:10.1007/s11042-020-09545-7 · Cited by 3
Pruning Distorted Images in MNIST Handwritten DigitsPreprint
Amarnath, R. N., & Kumar, V. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2307.14343 · Cited by 3
“Their contributions have been discussed in the literature [21]”
BackgroundThe dataset for extending EMNIST evaluation
Szymański, J., Skarżyński, K., Szutenberg, B., Ratkowska, K., & Drywa, S. (2026) · Scientific Data · doi:10.1038/s41597-025-06291-z · Cited by 3
“EMNIST is intended to represent a more challenging classification task compared to the widely used MNIST dataset, which has been considered a solved problem for some time now 8 .”
Determining the blur factor of handwritten characters using a convolutional neural network
Tuliabaeva, D., Tumakov, D. N., & Elshin, L. A. (2025) · Procedia Computer Science · doi:10.1016/j.procs.2024.12.030 · Cited by 2
Three quantum neural network architectures for multiclass image classification
Qu, G., Yu, X., Wang, Z., & Gu, Y. (2025) · International Conference on Algorithms, Network, and Communication Technology (ICANCT) · doi:10.1117/12.3060077 · Cited by 2
“We employed the MNIST dataset 21 to evaluate the performance of the three proposed QNN architectures on the task of eight-class image classification.”
Zhou, X., Hu, Y., Jia, Q., & Xie, R. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2025.3562633 · Cited by 2
“To demonstrate the superiority of NestFL across applications, we experiment with the MNIST [32] and CIFAR10 [33] image classification tasks and the HAR [34] human activity recognition tasks.”
Ai-guided vectorization for efficient storage and semantic retrieval of visual data
Harby, A. A., Zulkernine, F. H., & Abdulsalam, H. M. (2025) · Discover Artificial Intelligence · doi:10.1007/s44163-025-00713-y · Cited by 3
“MNIST [43] is a widely-used collection of handwritten digits, consisting of 60,000 training images and 10,000 testing images, each sized 28x28 pixels.”
InfluentialVicchietti, M. L., Ramos, F. M., & Campanharo, A. S. L. O. (2025) · Scientific Reports · doi:10.1038/s41598-025-27825-w · Cited by 2
“The MNIST dataset has been employed in numerous image processing studies as a standard reference point for machine learning and deep learning methodologies 81–84 .”
Variable Length Digit Recognition for Gujarati Language
Malvi, S., Patel, N., & Prajapati, P. (2022) · International Conference on Artificial Intelligence Trends and Pattern Recognition (ICAITPR) · doi:10.1109/icaitpr51569.2022.9844182 · Cited by 2
“There are multiple publicly available data set for the English numerals and alphabet (e.g. MNIST and Extended MNIST), and many research studies have proposed various methods to get over 95% accuracy [1].”
MethodKumbhar, R. R., Radhika, P., & Mane, D. (2023) · International Conference on Recent Advances in Electrical, Electronics, Ubiquitous Communication, and Computational Intelligence (RAEEUCCI) · doi:10.1109/raeeucci57140.2023.10134408 · Cited by 2
“The MIST dataset is one of the most diverse datasets for handwritten images and is considered a benchmark for other datasets[12].”
BackgroundYang, S.-Y., Wey, I., Hsu, H.-K., & Teo, T. H. (2023) · IEEE International Symposium on Embedded Multicore/Many-core Systems-on-Chip (MCSoC) · doi:10.1109/mcsoc60832.2023.00031 · Cited by 2
Handwritten Text Recognition Using Machine Learning
Nandan, T. P. K., Ankitha, G., Narutham, K., Bharadhwaz, M., Anusha, M., & Chinnaiah, M. C. (2024) · International Conference on Intelligent Technologies (CONIT) · doi:10.1109/conit61985.2024.10626948 · Cited by 2
“The paper [5] acknowledges the contribution of MNIST dataset.”
BackgroundMNSIT Handwritten Digit Recognition using CNN
Ahmareen, S., Alabdouli, A. K. K., & Polturi, S. (2024) · International Conference on Image Processing and Capsule Networks (ICIPCN) · doi:10.1109/icipcn63822.2024.00018 · Cited by 2
“[11] According to the research, the MNIST dataset is becoming less challenging as state-of-the-art methods have reached test error rates below 1%.”
MethodMNIST Image Classification with A Novel 2D-CIMA Method and Pareto Front Multi-Objective Grid Search
Putrada, A. G., Abdurohman, M., Perdana, D., & Nuha, H. H. (2024) · doi:10.1109/icoict61617.2024.10698573 · Cited by 2
“MNIST is a well-known image dataset researchers use to validate novel image classification techniques [6].”
BackgroundManimaran, N. H., Sutton, C. L. M., Streamer, J. W., Merkel, C. E., & Xu, K. W. (2024) · Journal of Physics Materials · doi:10.1088/2515-7639/ad9ee1 · Cited by 4
Handwriting Recognition of English Digits: A Deep Learning Perspective
Mahmud, T., Rahman, T., Aziz, M. T., Hasan, I., Barua, K., Barua, A., Basnin, N., Hossain, M. S., & Andersson, K. (2024) · Lecture notes in networks and systems · doi:10.1007/978-3-031-73324-6_10 · Cited by 2
CMOS compatible SiN-based memristive synapses for edge computing applications
Park, H., Noh, M., Cho, S., & Kim, S. (2025) · Journal of Alloys and Compounds · doi:10.1016/j.jallcom.2025.181398 · Cited by 2
Transcribed Digits and Apparel Classification using Convolutional Neural Network
Chutani, P., Bhatia, K., Gulati, S., & Arora, N. (2021) · International Conference on Communication Information and Computing Technology (ICCICT) · doi:10.1109/iccict50803.2021.9510145 · Cited by 2
Handwritten Character Recognition using Deep Neural Networks
Kotriwal, S., Pradhan, N., & Dhaka, V. S. (2021) · International Conference on Data Science, Machine Learning and Artificial Intelligence (DSMLAI) · doi:10.1145/3484824.3484901 · Cited by 2
“Baldominos et al. [7], present a review work on handwritten character recognition.”
BackgroundStrategies for Improving the Error Robustness of Convolutional Neural Networks
Morais, A., Barbosa, R., Lourenço, N., Cerveira, F., Lombardi, M., & Madeira, H. (2022) · IEEE International Conference on Software Quality, Reliability and Security (QRS) · doi:10.1109/qrs57517.2022.00092 · Cited by 4
“MNIST [28] is a collection of handwritten digits, divided into 60000 training and 10000 testing samples.”
BackgroundQuantum-Inspired Neural Network on Handwriting Datasets
Shah, M. R., Sarraf, J., Pattnaik, P. K., & Bandyopadhyay, A. (2023) · Lecture notes in networks and systems · doi:10.1007/978-981-19-9483-8_25 · Cited by 2
Triple‐loss driven generative adversarial network for pansharpening
Huang, B., Li, X., & Zhang, X. (2023) · IET Image Processing · doi:10.1049/ipr2.12943 · Cited by 2
Libby-Novick Beta-Liouville Distribution for Enhanced Anomaly Detection in Proportional Data
Sghaier, O., Amayri, M., & Bouguila, N. (2024) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3675405 · Cited by 2
An improved nearest neighbour classifier
Setterqvist, E., Kruglyak, N., & Forchheimer, R. (2025) · Pattern Analysis and Applications · doi:10.1007/s10044-024-01399-1 · Cited by 2
Consistent Lock-free Parallel Stochastic Gradient Descent for Fast and Stable ConvergencePreprint
Bäckström, K., Walulya, I., Papatriantafilou, M., & Tsigas, P. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2102.09032 · Cited by 2
“Among the most prominently used are multi-layer perceptrons (MLPs) and convolutional neural networks (CNNs) [5], where MLPs consist of layers densely connected through a weight matrix, and CNNs of sparsely connected performing filter convolutions, used in conjunction with MaxPool downsampling layers.”
MethodTraining of Offline Handwritten Text Recognisers Using Computer-Generated TextPreprint
Ehiorobo, E. E., Koleoso, R. A., & Uwadia, C. O. (2022) · SSRN Electronic Journal · doi:10.2139/ssrn.4055077 · Cited by 2
Mitigating the Impact of Labeling Errors on Training via Rockafellian Relaxation
Chen, L. L., Chern, B., Eckstrand, E., Mahapatra, A., & Røyset, J. Ø. (2024) · International Conference on Learning Representations · doi:10.48550/arxiv.2405.20531 · Cited by 2
“MNIST-10 : The MNIST dataset has been studied extensively and harnessed to investigate novel machine-learning methods, including CNNs (Baldominos et al., 2019).”
Standardness Clouds Meaning: A Position Regarding the Informed Usage of Standard DatasetsPreprint
Cech, T., Wegen, O., Atzberger, D., Richter, R., Scheibel, W., & Döllner, J. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2406.13552 · Cited by 2
“With Chhavi and Bottou [84] and Baldominos et al. [5], two papers discuss specifics of the dataset.”
InfluentialCarboni, L., Nwaigwe, D., Mainsant, M., Bayle, R., Reyboz, M., Mermillod, M., Dojat, M., & Achard, S. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4904745 · Cited by 2
Recognizing Handwritten Arabic Characters Using Deep Learning Techniques in Educational Platforms
Abood, A., AbdElrazek, E., & Saad, A. M. (2024) · المجلة العلمية لكلية التربية النوعية جامعة دمياط · doi:10.21608/sjeud.2024.284278.1032 · Cited by 1
Handwritten Text Recognition Algorithm
Zabala, M. I. S., Ferreiro, M. D., Maldonado, A. C., Dávila, L. C. D., & Beltramone, D. A. (2024) · World Congress on Medical Physics and Biomedical Engineering, September 7 - 12, 2009, Munich, Germany · doi:10.1007/978-3-031-61960-1_7 · Cited by 1
Image Recognition Imitating Saccadic Eye Movement in Grid Cell Net
Matsuda, Y., Niwa, K., Aoki, T., Takadama, K., & Sato, H. (2024) · Joint International Conference on Soft Computing and Intelligent Systems and International Symposium on Advanced Intelligent Systems (SCIS&ISIS) · doi:10.1109/scisisis61014.2024.10759985 · Cited by 1
“The previous study addressed recognizing numbers from zero to nine from images in the MNIST handwritten digit dataset [3].”
Kaushik, P., Dewal, I. S., Rathore, S. S., Singh, V. K., Kumar, S., & Bhavani, P. M. S. (2024) · International Conference on Advances in Computing, Communication and Networking (ICAC2N) · doi:10.1109/icac2n63387.2024.10894857 · Cited by 1
Dincic, M. R., Perić, Z., Denić, D. B., Denić, B., & Perić, S. (2025) · Transactions of the Institute of Measurement and Control · doi:10.1177/01423312251319575 · Cited by 1
Building a photonic neural network based on multi-operand multimode interference ring resonators
Do, T. T., Pham, H. Y., & Thanh, T. (2025) · International Journal of Reconfigurable and Embedded Systems (IJRES) · doi:10.11591/ijres.v14.i2.pp311-319 · Cited by 1
Backdoor Detection Method for Vision Transformer Model Based on Attention Map Analysis
Xu, W., Jiang, Y., Zhang, X., Liang, Z., Yang, Y., Bi, L., Xu, H., Yang, Q., & Huang, Y. (2025) · IEEE/CIC International Conference on Communications in China (ICCC) · doi:10.1109/iccc65529.2025.11148748 · Cited by 1
“We will conduct experiments on two datasets, MNIST [19] and CIFAR-10 [20], to fully evaluate the accuracy and robustness of the method in detecting backdoor samples.”
Ko, M., Byun, Y., & Kim, S. (2025) · Advanced Materials Technologies · doi:10.1002/admt.202501570 · Cited by 1
“Unlike the MNIST dataset, which is limited to numerical digits, EMNIST includes alphabetic characters, offering a more diverse environment for character pattern learning [64, 65].”
Solution-processed metal oxide synaptic transistor with bilayer Li-ion-conducting gate dielectric
Chakraborty, R., Singh, R. K., Pramanik, S., Chetri, P., Yadav, A. K., Aich, P. K., Barman, P., & Pal, B. N. (2025) · Journal of Materials Chemistry C · doi:10.1039/d5tc02255j · Cited by 1
Energy-efficient serverless federated learning with blockchain-enhanced optimized raft consensus
Lu, J., Qi, P., Yu, S., Liu, J., Cao, S., & Jin, Y. (2026) · Future Generation Computer Systems · doi:10.1016/j.future.2026.108473 · Cited by 1
Rahman, J. U., & Almarzooqi, S. (2024) · International Journal of Innovations in Science and Technology · doi:10.33411/ijist/890 · Cited by 1
A Study on the Effect of DropConnect to Control Overfitting in Designing Neural Networks
Lim, H.-I. (2020) · Frontiers in artificial intelligence and applications · doi:10.3233/faia200780 · Cited by 1
“The MNIST database is popularly used in designing and evaluating machine learning models for image recognition [10-12].”
MethodGuo, X.-F., & Wang, S. (2020) · European Journal of Horticultural Science · doi:10.17660/ejhs.2020/85.6.2 · Cited by 1
NGHIÊN CỨU TRIỂN KHAI MẠNG HỌC SÂU LENET5 TRÊN VI ĐIỀU KHIỂN STM32 ỨNG DỤNG TRONG NHẬN DẠNG HÌNH ẢNH
Thắng, H. V. (2021) · TNU Journal of Science and Technology · doi:10.34238/tnu-jst.4497 · Cited by 1
Kirupakaran, A. M., Yadav, K. S., & Laskar, R. H. (2022) · National Conference on Communications · doi:10.1109/ncc55593.2022.9806782 · Cited by 1
“This is equivalent to that reported for deep networks solely trained [22-23] or developed [24] for EMNIST.”
MethodResultA convolutional neural network model of multi-scale feature fusion: MFF-Net
Yi, Y., Wang, J., Ding, X., & Li, C. (2022) · Journal of Computational Methods in Sciences and Engineering · doi:10.3233/jcm-226356 · Cited by 1
“The test set has the same handwritten digit data source as the training set and it has a total of 60,000 samples, including 50,000 training set and 10,000 test set samples [7–9].”
Clustering of handwritten digits by Kohonen neural network
Латыпова, Д. С., & Tumakov, D. N. (2022) · Program systems theory and applications · doi:10.25209/2079-3316-2022-13-3-225-239 · Cited by 1
A Dataset to Investigate First-Person Shooter Players
Halbhuber, D., Höpfinger, J., Schwind, V., & Henze, N. (2022) · Annual Symposium on Computer-Human Interaction in Play · doi:10.1145/3505270.3558331 · Cited by 1
“Multiple teams attempted, and still attempt, to find the best computer-aided method for optical character recognition (OCR) [4, 9] in the MNIST challenge.”
BackgroundAn Automated System to Calculate Marks from Answer Scripts
Rizvee, R. A., Arefin, M. F., Khan, M. R., Islam, M. N., & Rabbi, K. F. (2022) · IEEE Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON) · doi:10.1109/iemcon56893.2022.9946626 · Cited by 1
“In [5], it was stated that in the handwritten digit prediction problem, simple 2 convolutional layers and 1 dense layer combination provided the least error or best accuracy among various tested architectures.”
BackgroundMethodLine Segmentation and Reading Order Detection Algorithm for Digitization of Korean Historical Texts
Lee, A., Min, G., Kim, K. S., Kim, J. E., & Kang, H. S. (2022) · Journal of Digital Contents Society · doi:10.9728/dcs.2022.23.11.2239 · Cited by 1
Agrafiotis, G., Makri, E., Kalamaras, I., Lalas, A. X., Votis, K., & Tzovaras, D. K. (2023) · Proceedings of the Northern Lights Deep Learning Workshop · doi:10.7557/18.6825 · Cited by 1
“One popular dataset used for this purpose is the MNIST collection of handwritten numbers [5].”
BackgroundDeteksi Objek dan Pengukuran Panjang serta Berat Ikan Menggunakan YOLOv3-ResNet18
Liawatimena, S., Abdurahman, E., Trisetyarso, A., & Wibowo, A. (2023) · Penerbit BRIN eBooks · doi:10.55981/brin.668.c554 · Cited by 1
Metadata
- Authors
- Baldominos, A., Saez, Y., & Isasi, P.
- Affiliations
- Universidad Carlos III de Madrid (ES)
- Computer Science Department, Universidad Carlos III of Madrid, 28911 Leganés, Madrid, Spain
- Corresponding author
- Baldominos, Alejandro
- Venue
- Applied Sciences
- Publisher
- MDPI AG
- Published
- 2019-08-04
- Volume
- 9
- Issue
- 15
- Pages
- 3169
- ISSN
- 2076-3417 (ISSN-L)
- OpenAlex
- W2966275995
- Cited by
- 279
- References
- 76
- Citations by year
- 2026: 20 · 2025: 39 · 2024: 64 · 2023: 54 · 2022: 41 · 2021: 36 · 2020: 24 · 2019: 1
- FWCI
- 11.41
- Type
- journal article
- Language
- en
- Open access
- gold
- dx.doi.org
- Licence
- Locations
- Applied Sciences (publishedVersion, cc-by)
- e-Archivo (Carlos III University of Madrid) (submittedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
- Funders
- Ministerio de Educación, Cultura y Deporte (FPU13/03917)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Handwritten Text Recognition Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Vehicle License Plate Recognition
- Keywords
- handwritten character recognition, handwritten digit recognition, MNIST, EMNIST, convolutional neural networks, data augmentation, letter recognition
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.


