A “style transfer” or image students created using convolutional neural networks during camp. Style transfer is the process of modifying the style of an image while preserving its content.
The department of statistics at the University of Georgia (UGA) partnered with Wells Fargo to hold the 2021 Data Science and Artificial Intelligence Summer Camp for high-school students July 12–23. Due to COVID-19 concerns, the camp was held virtually.
The co-organizers of the camp, Abhyuday Mandal and T. N. Sriram, selected a diverse set of 23 high-school students to participate.
Content for each session, developed by Wells Fargo, was made available to instructors through GitHub. UGA faculty members Ray Bai, Yuan Ke, Abhyuday Mandal, Qian Xiao, and Hani Safadi teamed up with seven data scientists from Wells Fargo to deliver the content to the students. In addition, six teaching assistants—Kaiwen Han, Jeevan Jankar, Ilsuk Kang, Beibei Xu, Mengyun Yu, and Tianyi Zhang—provided assistance to the instructors.
Students were introduced to Python programming and a wide variety of data science and artificial intelligence (AI) topics, including neural networks, natural language processing, machine learning, deep learning, reinforcement learning, and generative adversarial networks. After each presentation, students were divided into breakout rooms, where they worked in groups on a variety of hands-on activities.
On the last day of camp, each team gave two presentations—one technical and one nontechnical. The technical presentation focused on developing neural network models for object recognition in low-resolution images with discussion of model outputs, algorithm, and accuracy on testing data. The nontechnical presentation was an idea for an AI-based start-up company, where the teams proposed applying one or more of the AI techniques discussed during camp to a novel application.
Leave a Reply