I am an Assistant Professor within the Department of Analytics and Information Systems at Western Kentucky University's Gordon Ford College of Business. My research experience revolves around neuro-symbolic architectures called neuro-fuzzy networks (NFNs).
I am researching a new prototype NFN that can automatically reconfigure its knowledge base in response to its performance with respect to an objective function. These architectures natively handle missing data in the inputs and can easily transfer their knowledge to other models. The results of this novel work are reported in my dissertation.
- Created the first method to achieve fuzzy reinforcement learning in computer vision tasks (refer to my dissertation).
- First publication to create NFNs for tasks with high dimensions (to the best of my knowledge).
- Published the first work solely dedicated to offline model-free fuzzy reinforcement learning.
- First work to use fuzzy logic control to yield a pedagogical policy.
I may be interested in a research collaboration. Please feel free to reach out.
I implement a test-driven agile workflow that emphasizes modular and reusable code with high cohesion and low coupling. My code is clear, self-documenting, and minimal, yet I strongly advocate for comprehensive documentation, utilizing tools like Sphinx. I adhere to the relevant conventions of the programming language and maintain a consistent style, such as using Black for Python code formatting.
Email: john.hostetter@wku.edu
My "coding buddy" is an adopted four-year-old Siberian Husky named Zoey!
