The Murphy Lab

Hello and welcome!

We are a research group that designs and studies information processing systems using tools from information theory. We study representation learning in deep neural networks, and build algorithms that distill high-dimensional data into reduced descriptions for the express purpose of interpretability.

If you’re curious about how to design and understand AI systems, consider joining! Please reach out.

Visualization can be a powerful route to building intuition around how complex systems work. Below is a visualization of a randomly initialized neural network that warps two-dimensional space. The input starts as a square and then what you’re seeing is the square after passing through the network. Try varying the number of layers (64 units each) and the activation function!