46th Annual Summer Institute of Applied Statistics 2024

The featured speaker will be Dr. Chris Wikle, Curators' Distinguished Professor and Chair for the Department of Statistics at the University of Missouri.

Abstract

The “AI” revolution is upon us. Deep neural learning is a type of machine learning that exploits a connected hierarchical set of models to predict, classify, or generate elements of complex data sets. Although this revolution is relatively recent, many of the basic concepts have roots in statistics and optimization. With some exposure to the jargon and basic concepts, statisticians can easily add these methods to their data analysis toolkit or develop novel hybrid statistical/neural models. This course will present an introduction to deep models from a statistician’s perspective.
 
Topics will include an introduction to stochastic gradient optimization and concepts in regularization and dimension reduction, followed by discussion of the basic suite of neural models: feed forward neural networks, convolutional neural networks, recurrent neural networks, generative models, attention, and transformers (i.e., large language models). We will also discuss uncertainty quantification and approaches for explaining the inputs that are important for prediction and classification in these black box implementations. We will discuss some strengths and weaknesses and the motivation of these approaches. Time permitting, we will discuss some recent hybrid statistical/neural implementations such as neural Bayes estimation and deep reservoir models. The course will focus on concepts and modeling intuition and will include hands-on implementation with labs based on torch for R, with examples from different areas of application.

To view past presenters, click here.

To register and view registration pricing, click here.

For questions regarding SIAS, contact Kimri Mansfield at (801) 422-4506 or kmansfield@stat.byu.edu.