| CARVIEW |
Biography
Susan Wei is an Associate Professor in the Department of Econometrics and Business Statistics at Monash University. She previously held a Discovery Early Career Researcher Award (DECRA) and was a visiting faculty researcher at Google DeepMind in Sydney. Her research focuses on statistical machine learning, particularly in Bayesian approaches to deep learning, alongside variational inference and singular learning theory. She is part of the Melbourne Deep Learning Group and a founding organiser of GDG AI for Science - Australia.
Interests
- Statistics
- Deep Learning
- Singular learning theory
Education
PhD in Statistics, 2014
University of North Carolina, Chapel Hill
BA in Mathematics, 2009
University of California, Berkeley
Featured Work
Temperature Optimization for Bayesian Deep Learning
The Local Learning Coefficient: A Singularity-Aware Complexity Measure
A prior version of this work was released on arXiv under a different title and focused more heavily on the mathematical underpinnings.
Pathwise Gradient Variance Reduction with Control Variates in Variational Inference
Interventional Fairness on Partially Known Causal Graphs -- A Constrained Optimization Approach
The Developmental Landscape of In-Context Learning
Received a best papers award at ICML 2024 Workshop on High-dimensional Learning Dynamics (HiLD): The Emergence of Structure and Reasoning
Deep Learning is Singular, and That's Good
Counterfactual Fairness with Partially Known Causal Graph
A Shooting Formulation of Deep Learning
Selected for oral presentation
Direction-Projection-Permutation for High-Dimensional Hypothesis Tests
Experience
Associate Professor
Department of Econometrics and Business Statistics, Monash University
Visiting Faculty Researcher
Google Deepmind
Lecturer (Assistant Professor)
School of Mathematics and Statistics, University of Melbourne
Assistant Professor
Division of Biostatistics, University of Minnesota
Postdoc
Institute of Mathematics, Ecole Polytechnique Federale de Lausanne
Deep Learning Down Under
In summer 2024, I co-organized a deep learning workshop in Lorne, Australia with Peter Bartlett. Here is the website of the workshop, which was generously supported by Google Research.
Teaching
In Winter 2021, I gave a week-long lecture series on Neural Networks and Related Models as part of the Australian Mathematical Sciences Institute (AMSI) Winter School program, an annual event open to graduate students, early career researchers, and industry members across Australia. The course was an introduction to deep learning as well as some probabilistic models involving neural networks (flow-based models and deep generative models).
You can find my lecture slides here for Part 1 of the course where I covered the following topics:
- An Introduction to Neural Networks: key components of DL pipeline, multilayer perception, forward/backward propagation, computational graphs
- Stochastic Optimization and Extensions
- The Art of Model Training and Regularization: Model selection, weight decay, dropout, initialization
- Convolutional Neural Networks and Recurrent Neural Networks
The second part of the module on deep generative modeling was given by Robert Salomone. You can find his excellent teaching materials here.