For older students, level 5 and above, our summer presearch projects will focus on various applications of neural networks. We will start with an introduction to neural networks and learn how to build and train basic networks with Keras, Pytorch and Tensor Flow. Then students will focus on studying specific kinds of neural networks for image recognition, classification, physics and time series analysis. Finally, students will complete individual projects in one of three areas:
- Using object detection for real-time data collection of ballistic projectiles. ( We’re going to shoot tennis balls and collect data about where they go. )
- Predicting ballistic trajectories with Physics Informed Neural Networks. ( We’re going to try to predict where the tennis balls will land )
- Life-course prediction from longitudinal social data. ( We’re going to use survey data from when people are young to predict what they do when they are older. )
The full projects are intended for advanced students who are very self-directed. We will have regular meetings, once or twice a week, and students will need to spend at least 10 hours per week independently to make progress.
The research will start with studying the book Neural Networks from Scratch to get a deep understanding of what neural and then we will continue Zero to Mastery Learn PyTorch for Deep Learning. After that, students will work on their own projects, or contribute to a project to predict ballistic trajectories with a neural network.
