Deep learning is the future of artificial intelligence (AI). Although research has been going on for decades, AI has exploded in the last few years. Organizations have begun to get a handle on collecting and storing massive amounts of unstructured data. Technology and analytics used to process big data have improve dramatically, making it possible for machines to learn from large, complex datasets and perform specific tasks with greater speed, accuracy and efficiency than ever.
Deep learning is a form of AI that falls under the umbrella of machine learning. Deep learning relies on multilayered neural networks, which are modeled after neurons in the human brain, to process and correlate large volumes of data and solve complex problems. Unlike machine learning, which can hit an analytics performance wall after processing a certain volume of data, deep learning algorithms continue to become smarter as they consume more data.
Deep learning’s neural networks are capable of identifying more subtle patterns and correlations than machine learning. Also, machine learning algorithms produce strictly numerical outputs. Deep learning outputs can be just about anything, from numbers to text to audio.
The use cases for deep learning are virtually unlimited. Voice recognition technology can be used in contact centers to not only authenticate callers but to analyze sentiment. Audio analysis can detect flaws in machinery, cars and airplanes. Deep learning can be used in healthcare to aid patient assessments. For example, a neural network can help a dermatologist recognize skin cancer. Mayo Clinic researchers have used deep learning to identify the genomic data of a brain tumor without performing a biopsy.
Critical to the growth of AI is the ongoing development of graphics processing units (GPUs), the computer chips that deliver the power and speed to support deep learning software. GPUs were first developed to create smooth graphics in gaming systems, but are now widely used for AI. In fact, Popular Science has referred to the GPU as “the workhorse of modern AI.”
NVIDIA is the unquestioned leader in the GPU market, with more than 200 percent growth since last year. The NVIDIA Tesla P100 GPU, based on NVIDIA Pascal architecture, delivers more than 12 times the neural network training performance than the previous generation. Up to eight Tesla P100 GPUs can be connected with NVIDIA NVLink to maximize application performance and scalability, enabling organizations to expand uses of deep learning to overcome business, scientific and technical challenges.
HPE recently introduced innovative deep learning capabilities and workload-optimized solutions for large-scale, high-density GPU environments. The new HPE SGI 8600 supports liquid-cooled GPU performance with NVIDIA Tesla GPU accelerators for optimal scalability and power efficiency.
HPE supports NVIDIA’s next-generation Tesla GPUs in the Apollo 2000, Apollo 6500 and Proliant DL380 servers. HPE also announced enhanced collaboration with NVIDIA to advance benchmarking, code modernization and proof-of-concept initiatives at Centers of Excellence around the world.
These solutions are creating new possibilities for deep learning in the enterprise. Let us show you how to leverage the latest innovations from HPE and NVIDIA to drive your AI strategy.