Salt Lake City, Utah
June 23, 2018
June 23, 2018
July 27, 2018
Computing and Information Technology
Laboratory work is essential for students in the Science, Technology, Engineering, and Mathematics (STEM) fields. The laboratory work provides the students with practical experience of the theory and has the potential to increase enthusiasm by motivating the students to learn via experimentation. This is an important process for students as the active learning achieves positive educational results and prepares the students for real-world problems in the STEM fields.
This paper emphasizes the development of the Artificial Intelligence (AI) laboratory work for a new course in Embedded Artificial Intelligence (EAI). The course and the laboratory work are designed as an upper-level undergraduate elective to develop an AI system to run on an embedded device. It is open to all majors in the STEM fields who meet the prerequisite of a basic programming course and a linear algebra course. After an introduction to embedded programming and sensor interface, students will be introduced to machine learning and AI. During the corresponding lab sessions, the students are given a dataset to apply the theory in a sequential fashion. The laboratories start by employing traditional statistical classification algorithms, such as logistic regression, transitioning towards a deep neural network, such as a convolutional neural network (CNN). The accuracy of each model will be noted for each laboratory, starting with a lower accuracy but smaller model size for the statistical models and culminating with a relatively high accuracy with the CNN.
This paper outlines the design of the laboratories for the AI section of the EAI course, as well the feedback received during their development. The research to create and assess the AI exercises was conducted by a senior computer engineering student without any prior experience in AI. Two different forms of learning were taken into consideration: top-down approach where the student begins with a fully functional model and works backwards to understand each step of the processes and a bottom-up approach where the student begins from scratch and implements each component, working towards a fully functionally model. A comparative study of both approaches is presented from the point of view of the student. The assessment also asked the student to rate the assignment topics, to list how many hours were spent per each lab, and to propose suggestions for improvement.
Ergezer, M., & Kucharski, B., & Carpenter, A. (2018, June), Work in Progress: Designing Laboratory Work for a Novel Embedded AI Course Paper presented at 2018 ASEE Annual Conference & Exposition , Salt Lake City, Utah. 10.18260/1-2--31280
ASEE holds the copyright on this document. It may be read by the public free of charge. Authors may archive their work on personal websites or in institutional repositories with the following citation: © 2018 American Society for Engineering Education. Other scholars may excerpt or quote from these materials with the same citation. When excerpting or quoting from Conference Proceedings, authors should, in addition to noting the ASEE copyright, list all the original authors and their institutions and name the host city of the conference. - Last updated April 1, 2015