ºÚÁϳԹÏ

This class will cover the principles and techniques underlying programming language implementation. They include language grammar: both regular expressions and context free grammar, functional and imperative languages, variable binding and scoping, stack versus heap allocations, lazy versus eager evaluations, functional closures, etc. The goal of this class is not to design and implement a new programming language; it is to enhance the student’s appreciation and understanding of modern programming languages and the inherent tradeoffs language designers make.

An introduction to the fundamental concepts and ideas in natural language processing (NLP). The goals of the field range from document retrieval to text translation to enabling humans to converse with robots. We will study language processing starting from the word level, then moving to syntactic structure and finally to the semantic meaning of text. Approaches include symbolic as well as statistical methods, including an introduction to cutting-edge transformer-based deep learning models. Students will build and modify NLP systems and will use large existing corpora for validating their systems. We will also explore connections between NLP and the social sciences and highlight data ethics throughout the course.

The fundamental problems involved in computer networking remain the same even as our network hardware and software get better and faster. Modern integrated networking hardware and operating systems encode, hide, and enforce particular design decisions that are primarily side-effects of how commercial/commodity networks evolved. This is a team-based project-based course in which teams will be expected to independently research/rebuild computer networking from the link-layer through the application-layer and discuss/present their architecture/design decisions in context.

Description: In order for computer code to solve real-world problems, we need to represent those problems in a way the computer can understand: mathematically. This course will discuss how this is done, and give an overview of the strategies we can use for finding optimal solutions. Some specific topics covered will include: convex optimization, non-convex and combinatorial optimization, gradient descent, genetic algorithms, and simulated annealing. In particular, optimization is the heart of what makes machine learning and AI function, so we will pay special attention to these applications. Finally, we will critically analyze these tools, comparing their efficiency and discussing issues of implicit bias.

One could argue that the most important part of the software development process is requirements analysis--deciding what a software system should do. During this process, the primary emphasis is on 'business value' - the economic value provided by the system to the company developing and selling it. We also incorporate technical values such as performance and security. But what about other types of values? Can we build a software system by also incorporating human values such as diversity, compassion, and social responsibility? This class will look at requirements analysis and design and explore how human values could be incorporated into the process. This class will build on concepts taught in the humanities and social sciences. Ideally, students taking this class should have had at least one of the classes on the Computer Science Race, Gender, and Ethics course list.