Yao-Yi Chiang email@example.com
04:00 PM‑05:15 PM MW Amundson Hall 120
The location of things in space and how they change over time is the key to understanding complex environmental phenomena and human-environmental interactions. A significant amount of data now contains location and time information, either explicitly, e.g., traffic sensors, air quality sensors, satellite imagery, or implicitly, e.g., images and text documents.
This course aims to explore the foundation and the state-of-the-art on 1) spatial data management and 2) machine learning & data mining technologies that can exploit the unique spatial data properties (e.g., autocorrelations) to solve real-world problems. This is a seminar course consisting of lectures and paper presentations. Specifically, this course has two main themes. The first theme explores current ways to store and manage spatial data, including topics in spatial databases, spatial Big Data platforms, and/or knowledge graphs & ontology. The second theme looks into how machine learning & data mining technologies solve real-world problems utilizing the unique spatial data properties, including topics in computer vision (e.g., object detection from overhead imagery), location time-series data prediction & forecasting (e.g., air quality prediction and traffic forecasting), and optionally natural language processing (e.g., toponym detection from documents). The course will include several programming assignments and a final project.
The students should have excellent knowledge in applied machine learning (e.g., can select an appropriate machine learning model for solving a problem at hand) and databases (e.g., can write SQL queries with the help of the internet) and solid programming skills. Some background in handling spatial data is a plus but not required.
Students will be able to identify the role of spatial data and challenges in using them to solve a real-world problem. They will define the problem scope by identifying appropriate machine learning and data mining technologies and then leveraging the unique spatial data properties to solve the problem. For example, students will learn how to find and integrate spatial data from heterogeneous sources in the assignments. Then they will learn how to build machine learning or data mining methods to handle these spatial data for descriptive and predictive analysis.
The course will include several programming assignments and a final project. The programming assignments evaluate the students' capability in completing individual tasks towards the final project. The final project evaluates the students' overall capability in combining all knowledge learned from this course. Specifically, the final project should show working artificial intelligence technologies to tackle a real-world problem involving spatial data.
MS/Senior Undergrad Students
Write a comparison of selected state-of-the-art methods for solving a spatial AI problem (e.g., object detection from satellite imagery).
Develop a complete research work, which could be related to your research direction.