Deep Learning for Big Data Analytics in High-Performance Computing Environments
Md. Zahangir Alom
Deep Learning (DL) has been showing huge success for analysis the big data problem. However, this large scale implementation of deep learning algorithms for Big Data analytics requires huge computing resources, leading to a high power requirement and communication overhead. Recently, IBM has developed a new non von Neumann architecture called TrueNorth Cognitive System which allows for a new direction of research of in the neuromorphic computing. We have implemented deep learning approach with different optimizer on the IBM’s TrueNorth system using Caffe, Tea and Corelet Programming Environment (CPE-2.1) which is experimented on MNIST dataset. The experimental results are analyzed for different optimization functions. In addition, we also implemented Intrusion detection for cyber security which being considered another big data problem. The experimental results show promising recognition accuracy for anomaly detection and classification.
Graduate Research - Graduate
Tarek M Taha
Primary Advisor's Department
Electrical and Computer Engineering
Stander Symposium poster
"Deep Learning for Big Data Analytics in High-Performance Computing Environments" (2017). Stander Symposium Posters. 1075.