WORKSHOP "HPC UD ‘17”
"WORKSHOP ON MACHINE LEARNING USING INTEL ARCHITECTURE"
It has been roughly 60 years since the perceptron has been created. Since its early steps the task of training large neural networks has always been a computation intensive task. The recent rise of highly parallelized architectures and evolution of High Performance Computing techniques has dramatically decreased the time to train these networks, making them viable for classifying large amount of data. We have seen the subject gained a lot of thrust, specially in the area of Deep Neural Networks for pattern recognition in images. The goal of this tutorial is to present the main concepts related to Neural Networks and show how to leverage from HPC techniques when defining and training these models.
Todas las fechas
- 20 de Septiembre de 2017 09:50 AM - 01:00 PM
Powered by iCagenda