Better, Not Bigger, Data
Performing as a Turnaround CIO Artist, It's Not Magic Anymore!
Building A Strong Team Of IT Generals For Spearheading Campaigns
The Human Capital of High-Performance Computing
New Application Business Models - The Real IIoT Difference
Peter Zornio, CSO, Emerson Process Management
Technology Driving Better Mental Health Outcomes
Wes Williams, Ph.D., VP and CIO, Mental Health Center of Denver
Thank you for Subscribing to CIO Applications Weekly Brief
The Benefits of Bringing Artificial Intelligence to High Performance Computing
By Geetika Gupta, Principal Product Manager, HPC & AI at NVIDIA
HPC and AI complement each other. The early adopters of AI took the best-known methods from the HPC community. For the HPC community, AI is a new tool that can help business leaders get more mileage from all the data they are collecting. AI is a needle mover for innovating faster on solutions, improving customer loyalty, and achieving cost savings or operational excellence that directly benefits the bottom line.
Three benefits of implementing AI into HPC environments
1. Dramatically increase efficiencies
Within AI, deep learning is the fastest growing field and is being used to solve a wide range of problems. In very simple words, deep learning is an algorithm that simulates how a human brain learns to identify patterns by, for example, looking at images, hearing sounds or reading text. It is a powerful tool for analyzing tons of data in a very short time.
Take the pharmaceutical industry. Drug discovery is a long, expensive process. It takes an average of 12 years and $2.6 billion to bring a new drug to market. All drug discoveries require screening millions of candidates by simulating molecular interactions. Running simulations with a high level of accuracy is computationally expensive. It can take five years to simulate 10 million candidates.
Researchers at the University of Florida and the University of North Carolina developed ANAKIN-ME using deep learning algorithms to achieve a similar level of accuracy in a far more efficient manner. They leveraged their domain expertise and tailored the deep learning algorithm to screen drug candidates, reducing the time from years to minutes. This is great example of how the two disciplines came together and found a far more efficient solution to a hard problem.
Using deep learning to make personalized promotions and improve customer service is a good place to start
2. Discover more insights and take actions faster
Choosing where to apply deep learning can vary from enterprise to enterprise. A simple way to start is to identify the value drivers for the business, also known as “following the money.” For consumer-facing businesses, sales and marketing are likely to be the main drivers. Using deep learning to make personalized promotions and improve customer service is a good place to start.
For industries such as manufacturing and oil and gas, where operational excellence creates value, deep learning is well suited for optimizing supply chains, automating inspection and doing predictive maintenance.
Take the example of Baker Hughes, a GE company. BHGE has partnered with NVIDIA to use AI to help distill data in real time and reduce the cost of finding, extracting, processing and transporting oil. Oil-well operators can analyze massive amounts of production and sensor data, such as flow rates and pump pressures, with the help of AI. This gives oil and gas companies better insight into costly issues, such as predicting which machinery might fail and how these failures could affect larger systems.
3. Bring new products and solutions to market faster
Today, customer interactions with products or services happen across multiple platforms like smartphones, digital assistants and social media. All of these touch points provide a ton of real-time data to enterprises. There is opportunity to combine all this customer data and use the capabilities of supercomputing centers to transform research into improvements of everyday products.
For example, Procter and Gamble reached out in 2014 to Oak Ridge National Laboratory to simulate microscopic processes to improve product performance and quality. Fast forward to 2018, and the Summit supercomputer at Oak Ridge will have NVIDIA Tesla V100 Tensor Core GPUs that can be used both for science and deep learning. Under the ACCEL program, enterprises will be able to use deep learning and science to bring new products to market faster. Whether it is improving the efficiency of wind turbines or discovering polymers to improve battery life, there are many opportunities to convert leading-edge research into new products and services that address everyday needs.
The Future of AI and HPC
The future of AI and HPC holds incredible promise. It is the next industrial revolution. With the first industrial revolution, new inventions came about that couldn’t have been conceived of previously. Similarly, AI and HPC will result in new discoveries that we haven’t yet dreamed of.
AI is going to transform HPC in every industry — from pharmaceuticals, manufacturing, and energy exploration to climate-weather forecasting, materials science and more. Today’s visionary CIOs are taking advantage of this industrial revolution underfoot, transforming their businesses with the latest technology available. These trailblazers will deliver incredible benefits to business and improve society in ways we had once not thought possible.