The topic of machine learning covers a wide range of different technologies with their roots in statistics, neurobiology and IT. A very exciting field for machine learning lies in neural networks which are based on the biological structure model.
In particular, neural networks with ‘deep learning’ topology have enormous potential and have recorded spectacular success in recent times. Given the task of image recognition, deep learning networks already achieve better results than humans.
Configured with GPU hardware customised to neural network training, this technology offers as yet unseen possibilities. Deep learning is rapidly expanding into new sectors and areas of application. It also enables research to develop new topologies and learn algorithms for neural networks which deliver even better performances.
The whitepapers provide an insight into the fundamental principles of neural networks and also present connections to branches of physics.
There is an ever increasing demand of computational power in scientific or industrial research. At the same time, high-performance computing systems consume very much power.
Training neural networks requires high computing power which cannot be delivered by CPUs alone. NVIDIA offers Tesla series GPU models - high-performance modules for workstations which have been specifically designed for deep learning. This culminates in the DGX-1 machine which features 8 Pascal P100 cards.
NVIDIA® DGX-1™ is the world’s first purpose-built system for deep learning. The fully integrated hardware and software can be implemented extremely quickly and easily. The fact that its ground-breaking performance enormously accelerates training times makes it the world’s first deep learning supercomputer in a box.