Date of Award

6-2022

Document Type

Thesis

Publisher

Santa Clara : Santa Clara University, 2022.

Department

Computer Science and Engineering

First Advisor

Zhiqiang Tao

Abstract

Generative adversarial networks are being used to make original, AI generated content in many contexts, whether they are generating new works of art from a text description, or a completely fake person. Currently this technology is not being implemented in a way such that a user can practically interact with it in a way where users can actually use what is generated to improve their lives. We introduce OliveAI to explore a possible implementation of generative AI technology that users can practically and usefully interact with. OliveAI is a web app where users can provide a desired type of food (e.g. Pasta, Pizza, Cookies, etc.) and a list of ingredients which, once submitted, are processed through two different machine learning models on the backend to produce an original, completely AI generated recipe complete with instructions, ingredients, a title, and a picture of what the finished product could look like. Due to the computational complexity of this implementation and the difficulty of navigating two separate latent spaces OliveAI can generate some strange results, but oftentimes it does successfully generate a picture and recipe combination that matches with what the user provided. We believe we have successfully explored a new form of generative machine learning projects, where users interact with these models to better their lives in a practical day-to-day use kind of way. More time spent on this project could lead to better results as we allow for more food types to be used, dietary constraints to be taken into account, or the time one has to cook a meal.

Download

Share

COinS