Upon joining the AI-ACCESS program, each student is assigned an initial AI-ACCESS faculty advisor, who will meet with them to assess their background and advise them on track and course selection. AI-ACCESS trainees can choose one of three tracks (Computational, Environmental, or Social Sciences). We expect students to choose the track that is best aligned with their primary PhD program. For example, Computer Science and Engineering students are likely to join the Computational Science track, while Political Science students are likely to join the Social Science track.
The figure above provides an overview of the courses trainees will take based on their track; the year and semester is shown on the top-left of each course (e.g., 1F means Fall of Year 1). As the background and prior training of students in the different tracks may differ, we offer a number of remedial courses. Then, all students complete a common core cohort-building curriculum and also a domain depth requirement based on their track.
Remedial Courses
- Quantitative Methods I and II: PS 581 and 582 or Psych 5066 and 5067 (6 credits): A two-semester sequence covering essential probability and statistics, including hypothesis testing, inference, experimental methodology, using a modern statistical computing language like R, maximum-likelihood methods, Bayesian and nonparametric models, generalized linear models, and sampling techniques.
- Fundamentals of Computer Science: CSE 502 (3 credits): Students with very little computational training will take this remedial course, which is an existing fundamental course in algorithms and data structures, including significant implementation in an object-oriented programming language.
- Introduction to Machine Learning: CSE 417T or ESE 417 (3 credits): Students without exposure to an undergraduate ML course will take this existing course, which covers supervised learning, including generalization, overfitting, regularization, cross-validation, and model selection, and also the basics of core ML techniques and algorithms, including linear models like logistic regression, gradient descent, tree-based and ensemble methods, kernel methods, and artificial neural networks.
Core Cohort-Building Courses
- Introduction to Graduate Research: DCDS 501 (3 credits): To be taken in the first semester, the course is structured around topics that do not need detailed specific content background: Ethics, the nature of research, robustness and reproducibility of research, and presentations on the different AI-ACCESS research areas to give students an understanding of research opportunities in the program.
- Explorations in Computational and Data Sciences: DCDS 500 (3 credits): The course is designed to lay the foundation for conducting transdisciplinary research involving AI and, more broadly, computational science with environmental and social sciences. Opportunities exist to engage with the conceptual and technical challenges emerging from the increasingly ubiquitous availability of extensive datasets capturing many aspects of human life, social behavior, and the environment. Students work in diverse teams to apply methods to case studies.
- Data Wrangling: DCDS 510 (3 credits): This course will provide an introduction to conducting research using methods from data science. We examine the following sources of data: Static data sets, dynamic data accessed via an Application Programming Interface (API) such as tweets streamed from Twitter, web scraping as needed to harvest data where web pages contain desired data but no API is present, and surveys as needed to obtain data not already available. Students ingest data, perform analyses, and document their findings using an electronic notebook such as Jupyter, which allows the interspersing of data, code, analysis, and prose, serving to document and make reproducible the scientific process carried out by the students.
- Environmental Data Science: EECE 535 (3 credits): This course builds upon ML approaches taught in CSE 417T or ESE 417, and students will learn to build predictive ML models as well as effectively visualize and analyze environmental science datasets.
Domain Depth Courses
- Computational Science: Students must take an advanced algorithms course and an advanced AI/ML course in their area of interest.
- Environmental Science: Students must complete two advanced substantive courses in one subfield (geosciences, atmospheric sciences, or aquatic sciences) from a specified list for each subfield.
- Social Science: Students must take two advanced substantive courses from an approved list in their area of interest (political science, psychological and brain sciences, public health, or social work).