IntuiShop: Intuitive shopping for styles.

This project was completed as part of a consulting project, in which my client is building a platform that allows users to visually search for e-commerce clothing items. In order to help my client make important decisions about their proprietary AI algorithm, I transformed e-commerce product images and text descriptions into high-dimensional, high-level (semantic) vectors using deep learning. Then, I used a combination of supervised and unsupervised learning to evaluate the hidden patterns in their data in order to make recommendations to improve the performance of their algorithm. As an extension of this work, I also build a pipeline that integrates image and text information, allowing users to use intuitive search terms along with images (for example: "I want this dress, but more flowy"). [Read more about this project here!]


Collaborative memory foraging:

How do the dynamics of memory search change when we're remembering in a social, or collaborative, context? Does collaborative memory foraging follow the same Lévy-type processes observed in and hypothesized to be beneficial for individual recall (see Rhodes & Turvey, 2007). 1, 2, 3

Collaborative crosswords:

Although much of the literature predicts that collaboration disrupts an individual's recall, we have an intuitive sense that "two heads are better than one". In what ways can collaboration effect the retrieval of existing knowledge or information, and does the process of collaborating effect future recall of the information? For this project, subjects get to play a trivia game. 1

Category landscapes:

If memory search can be conceptualized as foraging across a mental landscape (as in the above project), how does the search process change when the landscape changes? We look at how search patterns may reflect different distributions of resources in the environment. This is a computational modeling project, in which we can vary parameters of the environmental and social contexts to investigate their effects on search behavior. 1

Rhythmic sequence coordination:

To better understand how two (or more) systems can become coordinated, synchronized, or aligned, we can study rhythmic sequence production. When a participant tries to synchronize with a rhythm, which channels of information are used? For this project, I am using the Vicon Motion Capture System in Ramesh Balasubramaniam's lab. 4, 5