Master of Computing and Information Systems, Youngstown State University, 2023, Department of Computer Science and Information Systems

The performance of machine learning (ML) methods, including deep learning, for classification and regression tasks applied to tabular datasets is sensitive to hyperparameters values. Therefore, finding the optimal values of these hyperparameters is integral to improving the prediction accuracy of a machine learning algorithm and the model selection. However, manually searching for the best configuration is a tedious task, and many AutoML (automated machine learning) frameworks have been proposed recently to help practitioners solve this problem. Hyperparameters are the values or configurations used to control the algorithm's behavior while building the model. Hyperparameter optimization is the guided process of finding the best combination of hyperparameters that delivers the best performance on the data and task at hand in a reasonable amount of time. In this work, the performance of two frequently used AutoML hyperparameter optimization frameworks, Optuna and HyperOpt, are compared on popular OpenML tabular datasets to identify the best framework for tabular data. The results of the experiments show that the performance score of Optuna is better than that of HyperOpt, while HyperOpt is the fastest for hyperparameter optimization.

Committee: Alina Lazar PhD (Advisor); Feng Yu PhD (Committee Member); John R. Sullins PhD (Committee Member) Subjects: Artificial Intelligence; Comparative; Computer Science; Information Systems

Master of Science in Computer Engineering, University of Dayton, 2022, Electrical and Computer Engineering

Image compression algorithms are the basis of media transmission and compression in the field of image processing. Decades after their inception, algorithms such as the JPEG image codec continue to be the industry standard. A notable research topic gathering momentum in the field of compression is deep learning (DL). This paper explores the opti- mization of DL models for ideal image compression and object detection (OD) applications. The DL model to be optimized is based upon an existing compression framework known as the CONNECT model. This framework wraps the traditional JPEG image codec within two convolutional neural networks (CNNs). The first network, ComCNN, focuses on com- pressing an input image into a compact representation to be fed into the image codec. The second network, RecCNN, focuses on reconstructing the output image from the codec as similarly as possible to the original image. To enhance the performance of the CONNECT model, an optimization software called Optuna wraps the framework. Hyperparameters are selected from each CNN to be evaluated and optimized by Optuna. Once the CONNECT model produces ideal results, the output images are applied to the YOLOv5 OD network. This paper explores the impact of DL hyperparameters on image quality and compres- sion metrics. In addition, a detection network will provide context to the effect of image compression on computer vision applications.

Committee: Bradley Ratliff (Committee Chair); Eric Balster (Committee Member); Barath Narayanan (Committee Member) Subjects: Computer Engineering