Developing a Automated Feedback generator for students in a Human-Centered Way (Master's Thesis)

I built a system that generates automated feedback for students using Machine Learning. This helps Teachers, Professors and Teaching Assistants while also providing quick and easy guidance to students. We conducted a semester long user study to study the effects of this feedback on a student's narrative game development process.
MS Thesis PDF
MS Thesis Slides

NewHome: Buy and Sell App

A platform where WPI students can post items for sale and search for items. This application prompts sustainability and reduces wastage on campus. Students are expected to have a valid WPI ID, as the application makes use of Face Recognition to authenticate profiles. This application uses Object Recognition to classify the items being sold. The application was designed in by putting the WPI community needs first. We conducted iterative surveys and research with WPI students.

Consequences: AI story expansion

This is an AI system that can expand a given paragraph into a story! It uses Generative Adversarial Networks to do this. The neural networks were trained on a large dataset of short stories available from the project Gutenberg. At the end of the project, the model was able to construct sentences with a rough approximation of what the next word could be.

Importance and Need for Multimodal Emotion Detection in HCI

Studied why multimodal methods have more potential to recognize emotions and user context from an HCI perspective. Explored and compared different ML models to identify emotions through images and audios.

Publications