Georgia Institute of Technology

Featuring Andrew Wilson, New York University 

Abstract: Deep neural networks are often seen as different from other model classes by defying conventional notions of generalization. Popular examples of anomalous generalization behaviour include benign overfitting, double descent, and the success of overparametrization. We argue that these phenomena are not distinct to neural networks, or particularly mysterious. Moreover, this generalization behaviour can be intuitively understood, and rigorously characterized using long-standing generalization frameworks such as PAC-Bayes and finite hypothesis bounds. We present soft inductive biases as a key unifying principle in explaining these phenomena: rather than restricting the hypothesis space to avoid overfitting, embrace a flexible hypothesis space, with a soft preference for simpler solutions that are consistent with the data. This principle can be encoded in many model classes, and thus deep learning is not as mysterious or different from other model classes as it might seem. However, we also highlight how deep learning is relatively distinct in other ways, such as its ability for representation learning, phenomena such as mode connectivity, and its relative universality.

Bio: Andrew Gordon Wilson is a Professor at the Courant Institute of Mathematical Sciences and Center for Data Science at New York University. He is interested in developing a prescriptive foundation for 

building intelligent systems. His work includes the discovery of mode connectivity, the SWA optimization procedure, the popular GPyTorch library for scalable Gaussian processes, informative generalization bounds for billion parameter neural networks, Bayesian optimization techniques for protein engineering, the first LLM for time-series forecasting, and many contributions to Bayesian deep learning. His website is https://cims.nyu.edu/~andrewgw.