Terrick Boyd

Graphical User Interface for Machine Learning Visualization Techniques

Visualization is a key component in machine learning tasks. Viewing high dimensional feature based results is challenging. Several dimensionality reduction techniques have been proposed. Different techniques provide different results based on their underlying algorithm.  My research project focuses on designing and implementing a Graphical User Interface (GUI) for various visualization methods including Principal Component Analysis (PCA), T-Distributed Stochastic Neighbor Embedding (t-SNE), and Isometric Mapping (ISOMAP). My project allows a one-stop place to upload machine learning results with specified labels and compare the outcomes of different visualization techniques on it. The parameters can be readily modified and the user can decide which model to save and utilize for further research. This cluster visualization application is available via simple web download of the executable file without the need for the user to utilize a python interpreter. The intuitive design of this interface enables anyone to use it, regardless if they have a background in statistics or computer science. My goal is to post the completed interface on GitHub as soon as possible in order to allow other investigators to take advantage of this useful tool to augment their research.