Teaching
Special topics courses
EECS 4980/5980: Social and Information Networks
This course introduces students to social and information networks, with a focus on analyzing social and information network data. The course will cover aspects including
The local and global structure of social and information networks, including the small-world effect, power-law degree distributions, and community structure.
Algorithms to reveal such structures on a variety of real social and information network data.
Generative models for the formation and growth of social and information networks.
Several applications involving social and information networks, including information diffusion and the spread of epidemics over social networks.
Required textbook: M. E. J. Newman, Networks: An Introduction, Oxford University Press, 2010. ISBN-13: 978-0199206650
Tentative list of topics:
Mathematics of networks
Centrality measures
Local and global structures of social and information networks
Data structures and algorithms for networks
Matrix algorithms
Community detection
Epidemics on networks
Random graph models
Dynamic and multi-layer networks
Link prediction
EECS 6980/8980: Probabilistic Methods in Data Science
The interdisciplinary field of data science draws upon concepts from electrical engineering, computer science, mathematics, statistics, and many other fields with the goal of extracting knowledge or patterns from data. Today we live in a data deluge which calls for automated methods of data analysis, such as machine learning methods that automatically detect these patterns and use them to make decisions or make predictions on future data. Some example applications of data science include
Automatically detecting and filtering out spam emails or posts on social media.
Identifying different types of objects in digital images.
Predicting patient lifetime following an operation.
Discovering groups of similarly behaving people within a population.
Identifying factors that affect achievement of college students.
There are many different components and approaches to data science. This course covers probabilistic methods for analyzing a variety of data types, both continuous and discrete, with and without dependencies between variables, including time-dependent data.
Required textbook: K. P. Murphy, Machine Learning: A Probabilistic Perspective, MIT Press, 2012. ISBN-13: 978-0262018029
Tentative list of topics:
Models for discrete data
Gaussian models for continuous data
Bayesian and frequentist decision theory
Linear and logistic regression
Bayesian networks
Mixture models
Hidden Markov models and state-space models
Variational inference
Latent variable models for relational and network data