Combined Convolutional and Perceptron Neural Networks for Handwritten Digits Recognition

Kayumov, Zufar; Tumakov, Dmitrii; Mosin, Sergey

doi:10.1109/dspa48919.2020.9213301

Cited by 11 publications

(5 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Momentum SGD optimization is used by CNN. This study tries to reduce model error by employing a small momentum decay factor [ 14 ], the literature's initialization approach [ 15 ], and the same batch normalization [ 16 ]. Samples are shuffled before being used to create a training set, followed by validation and testing on the new samples in the set.…”

Section: Introductionmentioning

confidence: 99%

“…The classification performance of the test set has been reported. These parameters are then fine-tuned via back propagation training [ 14 ].

where i is the number of iterations, v is the momentum variable, ε is the learning rate, and ( dL / dw | w i ) D i is the partial derivative of the objective function concerning the connection weight ω on the D i batch, which shows the optimization direction of the current collection.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

[Retracted] Classification of Bioinformatics EEG Data Signals to Identify Depressed Brain State Using CNN Model

Thakare¹,

Bhende²,

Deb

et al. 2022

BioMed Research International

View full text Add to dashboard Cite

Patients suffering from severe depression may be precisely assessed using online EEG categorization and their progress tracked over time, minimizing the risk of danger and suicide. Online EEG categorization systems, on the other hand, suffer additional challenges in the absence of empirical oversight. A lack of effective decoupling between brain regions and neural networks occurs during brain disease attacks, resulting in EEG data with poor signal intensity, high noise, and nonstationary characteristics. CNN employs momentum SGD optimization. By using a tiny momentum decay factor, the literature’s starting strategy, and the same batch normalization, this work attempts to decrease model error. Before being utilized to form a training set, samples are shuffled, followed by validation and testing on the new samples in the set. An online EEG categorization system driven by a convolution neural network has been developed to do this. The approach is applied directly to the EEG input and is able to accurately and quickly identify depressed states without the need for preprocessing or feature extraction. The healthy control group and the depression control group had accuracy, sensitivity, and specificity of 99.08 percent, 98.77 percent, and 99.42 percent, respectively, in experiments on depression evaluation based on publicly accessible data. The machine learning technique based on feature extraction is often getting more and more complex, making it only suited for offline EEG categorization. While neural networks have become increasingly important in the study of artificial intelligence in recent years, they are still essentially black-box function approximations with limited interpretability. In addition, quantitative study of the neural network shows that depressed patients and healthy persons have remarkable dissimilarity between the right and left temporal lobe brain regions.

show abstract

Section: Introductionmentioning

confidence: 99%

“…The classification performance of the test set has been reported. These parameters are then fine-tuned via back propagation training [ 14 ].

Section: Introductionmentioning

confidence: 99%

[Retracted] Classification of Bioinformatics EEG Data Signals to Identify Depressed Brain State Using CNN Model

Thakare¹,

Bhende²,

Deb

et al. 2022

BioMed Research International

View full text Add to dashboard Cite

show abstract

“…Многие алгоритмы проверяются на этой БД. Например, методы k-ближайшего соседа на основе MNIST дают ошибку 5 %, многослойные персептроны в зависимости от методов обучения и количества слоев -около 2-5 %, сверточные нейронные сети -чуть менее 1 %, а иерархические нейронные сети даже улучшают точность [26,27].…”

Section: материалы и методыunclassified

Применение Технологии CUDA Для Обучения Нейронной Сети Кохонена

Латыпова¹,

Latypova²,

Тумаков³

et al. 2022

ППС

View full text Add to dashboard Cite

Проведена кластеризация данных, содержащихся в обучающих выборках -базах данных MNIST и Fashion MNIST. Для кластеризации использована нейронная сеть Кохонена с евклидовой метрикой при оценке расстояний. Для каждой рукописной цифры (MNIST) и каждого типа предметов обихода (Fashion MNIST) определено оптимальное количество кластеров (не более 50).Обучение нейронной сети распараллелено на графическом устройстве NVidia с использованием технологии CUDA. Для каждой цифры приведены результаты, иллюстрирующие сравнение времен работы процессора и графического процессора. Как для цифр, так и для типов предметов обихода сделан вывод о 17-кратном ускорении параллельных вычислений над последовательными на игровом ноутбуке начального уровня. Тестовые выборки из тех же баз данных использованы для проверки правильности построения кластеров. Как для последовательного, так и для параллельного обучения сделан вывод о том, что векторы из тестовой выборки принадлежат правильному кластеру с вероятностью более 90 % в случае рукописных цифр. Кроме того, для каждой цифры и каждого типа предметов обихода вычислены F-меры для оценки попадания объектов в свои кластеры.Показано, что последовательная и параллельная кластеризации дают близкие результаты. Наилучшие значения F-меры получены для цифр 0 и 1 (F-среднее значение равно 0,974), в то время как худшее значение (0,903) получено для цифры 9. Для данных Fashion MNIST лучшее значение F-меры (0,96) получено для типа «Брюки», а худшее (0,34) -для типа «Рубашка». Несмотря на большие разбросы для значений F-метрик двух рассмотренных баз данных, отличия результатов кластеризации минимальны. Так, для MNIST максимальное отличие F-меры составляет порядка 0,01, а для Fashion MNISTоколо 0,04.

show abstract

“…The problem of recognizing objects such as handwritten letters or digits is often encountered in evaluating the performance of various CNN models [4]. The highest accuracy is achieved when using hierarchical or combined neural networks [5,6]. However, from the point of view of image processing, it is of interest, first of all, to analyze the recognition efficiency in the presence of various defects in the image.…”

Section: Introductionmentioning

confidence: 99%

On the recognition of weakly blurred, highly contrasting objects by neural networks

Tuliabaeva,

Dmitrii,

Elshin

2024

Third International Conference on Optics, Computer Applications, and Materials Science (CMSD-III 2023)

View full text Add to dashboard Cite

The influence of the degree of blur on the recognition of high-contrast simple objects is studied using images of handwritten digits and letters as an example. High-contrast objects are objects that are significantly brighter than the surrounding background. White images of handwritten digits (MNIST) and Latin letters (EMNIST) against a black background measuring 28 by 28 pixels are selected as training and testing data sets. It is found that recognition accuracy decreases linearly as the blur level increases. The work also shows that both an increase of blur and a decrease of blur from the blur level of the training sample worsens recognition. It is concluded that recognition when blur is reduced is worse than when blur is increased. Histograms are presented that demonstrate the dependence of recognition accuracy on the degree of blur. It is shown that the initial weights affect the recognition accuracy.

show abstract

Combined Convolutional and Perceptron Neural Networks for Handwritten Digits Recognition

Cited by 11 publications

References 18 publications

[Retracted] Classification of Bioinformatics EEG Data Signals to Identify Depressed Brain State Using CNN Model

[Retracted] Classification of Bioinformatics EEG Data Signals to Identify Depressed Brain State Using CNN Model

Применение Технологии CUDA Для Обучения Нейронной Сети Кохонена

On the recognition of weakly blurred, highly contrasting objects by neural networks

Contact Info

Product

Resources

About