Special Topics Course

Machine learning on graphs has emerged as an important research topic that finds applications in many domains including social networks, infrastructure design and maintenance, drug discovery, brain networks, and material design. This course will discuss methods to encode graphs (static/dynamic, attributed, undirected/directed, single/ensemble) into low-dimensional spaces to facilitate machine learning. Specific topics include random walks, kernels, spectral analysis, generative models, node embedding, subgraph embedding, and graph neural networks.

In this PhD level course, you will learn the models and algorithms of reinforcement learning (RL) as well as techniques in analyzing the sample-complexity of reinforcement learning algorithms in various settings.  The focus of the course would be on the statistical foundation of RL and for the most part we will consider the Markov decision process (MDP) model with finite states and finite actions, though towards the end of the course we will talk about function approximations as well as the ideas behind some deep RL algorithms. 

The goal of this course is to give an introduction to program synthesis, a new field at the intersection of programming languages, formal methods, and AI. The course will explore a number of fundamental questions around the problem of how to automatically discover programs that do what the user expects. The workloads include 2~3 programming assignments, 4~5 paper reviews, and a final project. In particular, the class will explore the following questions:

Enterprise systems are large-scale software applications to support business operations; they typically include software systems for data management, business process/workflow management, information flows, reporting, and data analytics.  Focusing only the data management aspect, a typical enterprise has to struggle with many data integration difficulties, since its data are usually spread around many database systems, workflow systems, file systems, etc. and in a variety of form possibly with no coherent semantics.

One of the perennial goals of computer graphics is creating high quality images which are indistinguishable from photographs: a goal referred to as photorealism. Another important goal is interactivity for visualization, simulation, gaming and other real-time applications. These two goals have historically been at odds with each other. In this course, we will review the history and some of the recent ideas that seek to bridge the gap between realism and interactivity. We will focus on the use of complex lighting and shading within limited computation time.

Perhaps the most powerful of the many connections between graphs and matrices is the Laplacian matrix of a graph. The definition is easy -- start with an undirected graph, put the vertex degrees on the diagonal, and put -1 in each off-diagonal position where there's an edge.

The course will cover techniques to compute on encrypted data, particularly homomorphic encryption, both partial homomorphic encryption, and fully homomorphic encryption. The course will have a large theory component. However, it will also discuss how homomorphic encryption is applied to real systems such as databases, media streaming services, anonymous messaging services, and machine learning systems. The course will be structured around paper readings, class discussions, high-quality paper review writing, and perhaps an individual research project.

Graphs are a remarkably versatile combinatorial object, used to model social, neural, road and computer networks, execution flow of programs, relations in databases, and countless other things. In this class we will investigate computational problems whose inputs are graphs, and how algorithms for such problems can exploit the structure of the input graph to improve performance. The focus will be on provable worst-case running time performance of our algorithms. On the way the class will expose students to some of the modern structural graph theory.

In this course, we will examine upcoming user interface technologies that will impact how we interact with our devices and digital content in the future. These include: immersive technologies (augmented and virtual reality), physiological interfaces (e.g., brain and body interfaces), wearable computing (e.g., devices both for reading and writing data to the user's body), multisensory and multimodal interactions in mixed, augmented and virtual realities (e.g., spatial audio, body movement), haptics (e.g., force feedback, sensing weight, feeling textures), and others.

Mixed and Augmented Reality, an active research field since the 1990s, has recently gained significant popularity because of the possibility of being implemented on smartphones and because of its unique approach of offering context-based computing directly in a person's field of vision. Augmented Reality is the concept of overlaying computer-generated information on top of the physical world. Mixed Reality is a bit broader and subsumes the fields of Augmented Reality, Augmented Virtuality, and Virtual Reality.