ECE 69500 - Crowd-Powered Systems
Course Details
Lecture Hours: 3 Credits: 3
Counts as:
Experimental Course Offered:
Spring 2015, Spring 2017
Catalog Description:
Crowdsourcing is the practice of delegating human work to people via the internet through an open call or paid microtasking markets. Human computation is strategy of building systems to solve problems that current strictly computational methods are not yet able to solve. It leverages Internet-connected human labor pools to perform the elements of an otherwise "computational" task that machines are not able to solve to adequate quality levels. This course will explore how to use crowdsourcing and human computation to build systems that solve problems in information retrieval, natural language processing, image understanding, and artificial intelligence. The focus will be balanced between implementation, theory, and ethical consequences. The course will involve development of a system that integrates human and machine processing.
Required Text(s):
None.
Recommended Text(s):
None.
Lecture Outline:
| Lectures | Major Topics |
|---|---|
| 1 | Foundations, terminology |
| 1 | Exemplary Crowd-powered system |
| 1 | Design of web forms and other work interfaces |
| .5 | Open problems (overview) |
| 1.5 | Automated task decomposition |
| 1 | Real-time crowd-powered systems |
| 1 | Security of web applications |
| 1 | Information retrieval models |
| 1 | Ethics |
| 2 | Quality control - algorithms, financial incentives, design implications |
| 1 | Representations and programming models |
| 1 | Human computation games |
| 1 | Active Learning (machine learning) |
| 1 | Project presentations |