From HPC to Deep Learning: IBM PowerAI Vision
IBM PowerAI Vision is a complete ecosystem on the PowerAI distribution to analyze images and videos for actionable insights. Based of Deep Learning models for accuracy and Power Systems to leverage the acceleration from GPUs, support for Large Memory scaling and Distributed Deep Learning, AI Vision offers a competitive solution that delivers highly accurate and performing platform.
The session will present latest IBM PowerAI Vision on IBM Minsky Server Systems accelerated by NVIDIA Pascal GPU with related performance for Deep Learning frameworks
Intelligence on the Edge – where machine meets the reality
Intel Romania GM, Co-founder of Movidius – an Intel company
Advances in computational power have made machine learning the technology of choice for many future applications that require image/pattern recognition, while processing sensor data fusion. Low power processors have made possible pushing the machine towards the edge. Network availability, network latency, privacy are a few reasons to make full use of distributed computation power in the real field, away from centralized cloud infrastructure. While machine training based on neural networks is carried out only in the cloud for the moment, inference is possible with neural network accelerators such as the ones designed on Movidius processors. The machine future looks great! However, have we crossed the frozen bridge from cloud to reality in the field? What applications are likely to crack the ice open?
Amazon Web Services for Machine Learning
Amazon Development Center, Iaşi, Romania
Amazon Web Services (AWS) is a subdivision of Amazon which provides cloud services to thousands of businesses all over the world. One of its focus in the past years has been to ensure support for the growing field of Machine Learning (ML) in both academia and the private sector.
This presentation will go over some of the facilities AWS provides for ML – such as EMR clusters with Apache Spark and easy access to S3, which go hand in hand with the contributions to open source ML related projects – such as the deep learning dedicated library MXNet. Finally, we will look at AWS programs supporting the access of students and academia.