Overcoming challenges in artificial intelligence training: data limitations, computational costs and model robustness

Keywords: artificial intelligence, AI training, computational costs, environmental impact, model robustness, interpretability, energy efficiency, AI ethics, sustainable AI

Abstract

This paper explores challenges in AI training, focusing on data limitations, computational costs, and the need for robust models. It discusses innovative solutions like synthetic data generation, efficient neural architectures, and robustness techniques, highlighting the importance of AI model interpretability.

References

Conditional Synthetic Data Generation for Robust Machine Learning Applications with Limited Pandemic Data / H. P. Das et al. ResearchGate.

Bex T. 5 Powerful Cross-Validation Methods to Skyrocket Robustness of Your ML Models. Medium.

Carneiro G., Santana E., R. Cordeiro F. A Study on the Impact of Data Augmentation for Training Convolutional Neural Networks in the Presence of Noisy Labels.

Neural Network Pruning by Cooperative Coevolution. Thirty-First International Joint Conference on Artificial Intelligence (IJCAI-22).

Mirzasoleiman B., Yu Liu T. Data-Efficient Augmentation for Training Neural Networks.


Abstract views: 0
PDF Downloads: 0
Published
2023-12-16
How to Cite
Bortnyk, K., Yaroshchuk, B., Bahnіuk N., & Pekh, P. (2023). Overcoming challenges in artificial intelligence training: data limitations, computational costs and model robustness. COMPUTER-INTEGRATED TECHNOLOGIES: EDUCATION, SCIENCE, PRODUCTION, (53), 37-43. https://doi.org/10.36910/6775-2524-0560-2023-53-06
Section
Computer science and computer engineering