Search
3 articles for “MNIST”
-
Assessing the Performance of DL Methods in Handwritten Digit Recognition
Abstract: Handwritten digit recognition is a computer vision task that involves the automatic identification and classification of hand-written digits. The objective is to develop models capable of accurately recognizing and distinguishing digits handwritten by humans. With the development of machine learning and deep learning techniques, this field has advanced remarkably. The convolutional neural network (CNN) is the most often used technique for this purpose. By utilizing CNN, the model can learn …
Published in International Journal of Data Structure Studies · Vol. 1, Issue 1, 2023 · pp. 25–32 Read article
-
Accelerating Unsupervised Feature Learning: Parallelized Training of Denoising Autoencoders
Abstract: Unsupervised representation learning has become a cornerstone of contemporary machine learning, enabling algorithms to extract informative features from un-labelled, high-dimensional data. This work investigates the efficacy of stacked denoising autoencoders (SDAEs) trained via parallelized stochastic gradient descent (SGD) as a scalable approach to feature extraction. By strategically leveraging multi-threaded computation, our study systematically examines the trade-offs between increased parallelism, training efficiency, and the preservation of model accuracy. Experiments on the …
Published in Journal of Image Processing & Pattern Recognition Progress · Vol. 12, Issue 3, 2025 · pp. 56–64 Read article
-
Parallel Privacy-Preserving Adaptive Federated Learning on GPU-Enabled Multi-Core Architectures
Abstract: The increasing deployment of parallel and distributed intelligent systems has intensified the need for privacy-preserving learning frameworks that can exploit multi-core and GPU-based architectures without centralizing sensitive data. This work proposes a parallel Adaptive Federated Learning (AFL) framework that integrates Differential Privacy and Secure Aggregation over heterogeneous multi-core and GPU platforms to enhance both data confidentiality and convergence efficiency. The framework dynamically adjusts client participation, learning rates, and aggregation weights …
Published in Recent Trends in Parallel Computing · Vol. 13, Issue 1, 2026 Read article