TY - JOUR TI - Automated machine learning for supervised and unsupervised models with artificial neural networks DO - https://doi.org/doi:10.7282/t3-8k6c-b717 PY - 2021 AB - Artificial Neural Networks (ANNs) are powerful machine learning tools to find and apply patterns for intelligent decision making. These tools can be combined with automation to select few results among many trials. Since ANNs are used for both supervised and unsupervised learning, automation can lead to more trusted learning methods across many fields and lead to exploring possibilities that are considered impossible with current technology. In this thesis, at first, I introduce a new form of ANN architecture which is used exclusively for automated robot navigation. By doing so, I provide a high-level overview of both computational neuroscience and the potential of automation. Next, I introduce Greedy AutoAugment to automate the learning of state-of-the-art neural networks for both big and small datasets. I also create an efficient model to evaluate clustering in unsupervised learning. The model is further expanded to introduce unsupervised learning for deep subspace clustering. In the end, I provide discussion and the future research plan for automating ANNs in machine learning applications. KW - ANN KW - Neural networks (Computer science) KW - Computer Science LA - English ER -