Research

Machine Learning and Deep Learning

One way to view machine learning is to try to find mathematical functions that represent our world. A function is nothing more than a map that always takes some input to an output. We could construct a function which would take inputs such as round, sweet, and red to an apple and map oblong, green, and sour to a lime. How is this done? In practice, the weights that parameterize a function are adjusted after looking at many (perhaps hundreds of thousands) examples of input and output. Through all of these training examples, the mapping can be adjusted and we can learn a proper map. The real value of this process is that after training, we have a function which we can apply to novel data and make intelligent predictions.

With recent advances in machine learning and computing resources (such as CUDA and general computing with GPUs), we can now learn very complex models. These models can learn not only the labels (or mappings) but also patterns in data. By learning the patterns in the data first, the methods are able to create better maps. Being larger, these newer models are also able to capture more complex patterns and relationships and better model aspects of our complex world.

Big Data and Data Mining

You may have heard the phrase "drowning in data". The idea is that in today's digitalized world, we create massive amounts of data. This data could be surveillance video, purchases you made at the store, images uploaded to the web or what sites you have visited on the Internet. The goal of collecting this data is the hope of understanding it. The two main challenges are effectively handling the large amount of data and identifying meaningful patterns in it.

Computer Science Education and Educational Technology

We live a world that is ever more entangled with technology. This presents complex challenges and opportunities. How can we use this technology to support learning? How can we prepare the next generation of computer scientists to solve problems, implement solutions, consider the ethical implications of those solutions and support the common good?

Web Tools and Services

Resources