The exponential growth in use of large, deep neural networks (DNN) has accelerated the need for training these networks in hours—even minutes.
This kind of speed cannot be achieved on a single machine—a single node cannot satisfy the compute, memory, and I/O requirements of today’s state-of-the-art DNNs.
The way to do it is through scalable and efficient distributed training, which is facilitated by deep-learning frameworks.
Join Intel® software engineer and deep-learning expert, Mikhail Smorkalov, for an overview of three Intel®-optimized deep-learning frameworks—Caffe*, Horovod* (for TensorFlow*), and nGraph—that boost communication performance on distributed workloads compared to existing approaches.
Find out more about these optimized frameworks, including how to get them.
- Intel® Optimization for TensorFlow*
- Using Intel® Xeon® Processors for Multi-Node Scaling of TensorFlow with Horovod
Software engineer, Intel Corporation
Mikhail specializes in deep-learning technologies. His responsibilities include defining deep-learning architecture, developing and deploying new features for the Intel® Machine Learning Scaling Library (Intel® MLSL) and scaling deep-learning workloads to some of the fastest supercomputers in the world.
Before joining Intel in 2014, Mikhail spent years developing software and middleware for the telecom industry. He holds a master of science in computational mathematics and cybernetics from the State University of Nizhni Novgorod.