Sample Student Projects in the AI Research Laboratory
Hang (Hanna) Tran
Many businesses enhance their revenue-driven planning through forecasting analysis and business intelligence. Hanna’s honors thesis research uses Recurrent Neural Networks (RNNs), in particular LSTMs (Long Short-Term Memory) networks, applied to data from Starbucks.
These algorithms were chosen for their ability to drive business efficiency from latent data. The goal is to use data analysis to develop marketing strategies for mobile apps, train neural networks to predict human behavior, and, finally, apply these technologies to recommend products.
Central to the project is assessing the popularity of food and drink products through customer views. Given the vast amount of data involved, the biggest challenge is compressing and aggregating the data in a way that could provide essential insights for the business. A key goal of this research is to minimize the amount of error and prediction in the future through an ensemble of data science techniques.
Mahiteesh Alla & Naveen Lingam
In India, the agriculture sector is an unorganized sector. Through Mahiteesh’s and Lingam’s interest in agriculture and coming from an agribusiness family involved in chili breeding and marketing, they saw the opportunity to launch an agri-tech startup with an Al-enabled platform to help organize agriculture.
Historically, retailers and traders have had the most influence over agricultural product prices, stocks, and flow. It is Mahiteesh’s and Lingam’s hope to become the biggest distributors of agricultural products in India with the help of artificial intelligence. Using AI, their goals are three-fold. To:
Facial recognition is a tool for tracking an individual’s movements, locating criminals, and as a means to clear security. While much research has been done to improve facial-recognition algorithms, little has been done to study disguises and other means to avoid detection. Using a large data set of faces in various states of disguise, Duc is training AI to learn the specific details necessary to enable detection while the images are altered to thwart certain detection.