Dr. Anthony Deese, Chair of the Electrical and Computer Engineering Department and Associate Professor, recently gave a talk to School of Engineering students about his most recent research endeavors on machine learning via serverless cloud computing in AWS Lambda.
Serverless computing is becoming more and more important as its usage grows due to mobile devices. Dr. Deese worked to apply training data to a machine learning algorithm in a serverless environment in the hopes that our low powered mobile devices will be able to perform complex tasks such as machine learning. His work focuses on finding a novel approach to machine learning for big data processing through the usage of modular lambda functions and does the following:
- Employs a lean design philosophy such that serverless cloud computing resources are used efficiently.
- Interfaces lambda functions directly with cloud storage to efficiently access large amounts of data.
- Emphasizes code modularity and reliance on micro-services to maximize scalability and re-usability.
- Interacts with mobile device applications seamlessly as gateways for user interaction.
Dr. Deese wanted to find more efficient ways to perform both unsupervised machine learning, using Lloyd’s algorithm, as well as supervised machine learning, using gradient descent back-propagation applied to an artificial neural network. He created a lambda function in the Eclipse IDE which was then uploaded to AWS, and also developed a mobile application that would perform the machine learning algorithms. From there he ran tests using his code for AWS Lambda within the AWS Cloud, Android OS with the developed mobile application, as well as a MATLAB script running on a PC to compare the computation times.
Dr. Deese is currently working on a paper about his research. His presentation slides, which include his results, as well as a video of his presentation can be found on his website here.