Teaching

• 20222020

  STAT 8112 - Mathematical Statistics II

  Coverage: 1) different modes of convergence and their relationships, 2) when to swap limit and expectation, 3) martingale and its convergence, 4) central limit theorems under various conditions, 5) the Delta method, 6) U-statistics, 7) maximum likelihood estimation, M-estimators, and Z-estimators, 8) asymptotical analysis, 9) information criteria and general model selection, 10) basics of Bayesian, 11) Bayesian asymptotics in terms of consistency and Bernstein-von Mises theorem, and 12) various ways to hypothesis testing, power analysis.

  Audience: Ph.D. students in statistics.

• 20222022

  STAT 4052 - Introduction to Statistical Learning

  Coverage: 1) linear regression, 2) logistic regression, 3) tree methods, 4) elements of deep learning, including feedforward and convolutional neural networks, recurrent neural networks, applications to object classification/detection and language modeling, 5) bias-var tradeoffs and model selection, 6) model diagnostics, 7) support vector machine, 8) margin-based classification, 9) k-means and model-based clustering methods, 10) PCA, sufficient dimension reduction, and deep autoencoders, 11) optimization methods such as GD, SGD, 12) various in-lab case studies.

  Optional material: 13) knowledge distillation, 14) adversarial learning, 15) computational frameworks such as Pytorch.

  Audience: undergraduate-level students who major in statistics.

• 20212018

  STAT 5302 - Applied Regression Analysis

  Coverage: 1) multiple regression and interpretations, 2) complex regressors, 3) analysis of variance, 4) nonlinear transformations, 5) regression diagnostics, 6) and variable selection, 7) intro to statistical learning, and 8) deep learning techniques, including feedforward neural network, auto-encoder, convolutional neural network, recurrent neural network, stochastic gradient descent.

  Audience: mostly master-level graduate students and some Ph.D. students from departments other than statistics.

• 20202019

  STAT 5021 - Statistical Analysis

  Coverage: 1) data description and basic probability, 2) random variables, 3) hypothesis tests and analysis of variance, 4) multiple regression and interpretations, 5) complex regressors, and 6) regression diagnostics.

  Audience: mostly master-level graduate students and some Ph.D. students from departments other than statistics.

• 20242023

  PHY/AST/CSCI/STAT 8581 - Big Data in Astrophysics

  Coverage: This course will cover key concepts and techniques used in working with large datasets in the field of astrophysics. In the first four weeks, the focus will be on modern approaches to creating and manipulating large datasets, with an emphasis on time series analysis and Bayesian methods applied to astrophysics survey data. The remaining part of the course will delve into various machine learning techniques for data processing, including classification algorithms (supervised and unsupervised learning), clustering algorithms, regression problems, recommender systems, and graphical models. Each type of algorithm will be covered for around 2 weeks, with an initial introduction in 1-2 lectures and then team projects where students will apply the algorithms to astrophysical data sets in order to answer specific questions in astrophysics.

  Audience: mostly master-level graduate students and Ph.D. students in astrophysics, CS, and Stat

• 20242024

  STAT 8931 – Advanced Topics in Statistics

  Coverage: This course will equip students with broad view on the recent AI landscape. The course will delve into a variety of topics outlined below, ensuring that students grasp both the algorithmic underpinnings and practical implementations: 1) Foundations of Deep Learning and Generative AI, 2) Evolution of Language Modeling, 3) Unpacking Transformer Models, 4) Training LLMs from scratch, 5) Diffusion Models, 6) Efficiency in Foundation Models’ Training and Finetuning, 7) Deployment Strategies for Foundation Models, 8) Ethical Considerations in Generative AI, and 9) Adversarial Learning in Generative AI.

  Audience: Ph.D. students in Stat, EE, and CS