Automated Machine Learning: Understanding the relation between Data and Neurons

Sivasamy Kalamani, Mugil (2024) Automated Machine Learning: Understanding the relation between Data and Neurons. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration manual) Download (749kB) | Preview

Abstract

Automated Machine Learning is one of the fields that can improve productivity in every possible field as it automates most of the ML tasks, allowing everyone to develop AI systems. Neural networks are the base of all AI systems in the current world, and they are very complex, making it difficult to understand the deep workings of the dataset. Without that knowledge, automating neural network model selection and tuning is impossible. This research study mainly focuses on selecting the no of neurons in each layer and the number of layers in the neural network. As a bonus, the analysis gave a deep understanding of the contribution of the Activation function to fitting the curve. The basic straight-line equation y = mx + c is the basis of the neural network, and a neuron holds every straight line or a single-valued function. Thus, from all the experiments, it is found that the minimum no of neurons required in the first layers is (e + 1), where e is the total no of extrema in the data column. For n-dimensional Dataset, the sum of all extremas in all the columns + no of columns. The first layer of the neural network is the most important, and the activation function is chosen Based on how the data relation is; if it is a smooth curve, smooth functions like sigmoid and tanh can be used, and for sharp-edged curves, ReLU and other functions should be used.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Syed, Muslim Jameel UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Artificial Intelligence
Depositing User:	Tamara Malone
Date Deposited:	07 Apr 2025 10:38
Last Modified:	07 Apr 2025 10:38
URI:	https://norma.ncirl.ie/id/eprint/7375

Actions (login required)

View Item