Comparative Analysis of Machine Learning Libraries for Neural Networks: A Benchmarking Study
Eid, A. I. A.; Mjlae, S. A.; Rababa'h, S. Y.; Hammad, A.; Rababah, M. A. I.
Show abstract
This comprehensive benchmarking study explores the performance of three prominent machine learning libraries: PyTorch, Keras with TensorFlow backend, and Scikit-learn with the same criteria, software, and hardware. The evaluation encompasses two diverse datasets, "student performance" and "College Attending Plan Classification," supported by Kaggle platforms utilizing feedforward neural networks (FNNs) as the modeling technique. The findings reveal that PyTorch and Keras with TensorFlow backend excel on the "College Attending Plan Classification" dataset, with PyTorch achieving impeccable precision, Recall, and F1-score for both classes. While Scikit-learn demonstrates commendable performance, it trails behind these libraries in this context. On the "Student Performance" dataset, all three libraries deliver comparable results, with Scikit-learn exhibiting the lowest accuracy at 16%. Keras with TensorFlow backend and PyTorch attain accuracy rates of 23%, respectively. Moreover, this study offers valuable insights into each librarys unique strengths and weaknesses when confronted with diverse dataset types. PyTorch emerges as the go-to choice for demanding tasks requiring high performance, while Scikit-learn proves advantageous for simpler tasks with modest computational demands. Keras with TensorFlow backend strikes a balance between performance and user-friendliness. This benchmarking endeavor equips machine learning practitioners with valuable guidance for selecting the most suitable library or framework tailored to their project requirements. It underscores the pivotal role of library choice in achieving optimal results in machine learning endeavors.
Matching journals
The top 8 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Cardiac disease diagnosis based on GAN in case of missing data 96%
- Improving prediction of drug-target interactions based on fusing multiple features with data balancing and feature selection techniques 95%
- A cautionary tale about properly vetting datasets used in supervised learning predicting metabolic pathway involvement 94%
Similar papers in this journal
- A machine-learning Approach for Stress Detection Using Wearable Sensors in Free-living Environments 96%
- Identification of Myocardial Infarction (MI) Probability from Imbalanced Medical Survey Data: An Artificial Neural Network (ANN) with Explainable AI (XAI) Insights 95%
- Employing Machine Learning Techniques to Detect Protein-Protein Interaction: A Survey, Experimental, and Comparative Evaluations 95%
Similar papers in this journal
- Uncovering the effects of model initialization on deep model generalization: A study with adult and pediatric chest X-ray images 94%
- Identification of predictive patient characteristics for assessing the probability of COVID-19 in-hospital mortality 93%
- From theoretical models to practical deployment: A perspective and case study of opportunities and challenges in AI-driven healthcare research for low-income settings 92%
Similar papers in this journal
- A novel interpretable deep transfer learning combining diverse learnable parameters for improved T2D prediction based on single-cell gene regulatory networks 94%
- Machine learning for classifying chronic kidney disease and predicting creatinine levels using at-home measurements 94%
- A Convolution Based Computational Approach Towards DNA N6-methyladenine Site Identification and Motif Extraction in Rice Genome 94%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.