Machine learning and deep learning are transforming the way enterprises make decisions. This is especially true in industries like financial services, insurance, healthcare and telecom, where companies are increasingly replacing rules-based systems with AI-based applications.
For instance, banks are using AI-based anti-money laundering solutions to improve accuracy and reduce false positives. Fraudulent behavior is being detected faster and with better accuracy, and product recommendations are now personalized in near real-time for customers.
With the H2O GPU Edition, H2O.ai seeks to build the fastest AI platform on GPUs. For several years, deep learning has been taking advantage of the tremendous performance boost provided by GPUs. Now, many machine learning algorithms can also benefit from the efficient, fine-grained parallelism and high throughput of GPUs.
Importantly, GPUs allow one to complete training and inference much faster than possible on CPUs. The performance improvement is stark.
We see at least a 10x speedup on H2O GPU Edition with one GPU, compared to two CPUs for gradient boosted trees (GBM), one of the most popular machine learning technique used by data scientists. And we see about a 5x speedup on generalized linear models (GLM), a very popular and commonly used machine learning technique that is also highly interpretable.
NVIDIA GPUs enable a quantum leap in machine learning, opening the possibilities to train more models, larger models and more complex models — all in much shorter times.
Iteration cycles can be shortened and delivery of AI within organizations can be scaled with multiple GPU boards with multiple nodes. Enterprises can use this end-to-end solution to operate on large datasets, iterate faster, deploy quickly and gain real-time insights.
The benefit for enterprises is that they can build thousands of models in the time it used to take to build a single model. It’s a huge force multiplier for their highly valuable and scarce data science resources to be able to solve many more problems.
Data scientists can try thousands of experiments and iterations to find the most accurate models for their critical business problems. More than 10,000 organizations and nearly 100,000 data scientists already depend on H2O for critical applications like predictive maintenance and operational intelligence. This also leads to massive optimization in computing footprint at data centers.
The H2O GPU edition captures the benefits from both GPU acceleration and H2O’s implementation of mathematical optimizations. This combination takes the performance of AI to a level unparalleled in the space.
To learn more, watch the on-demand webinar by NVIDIA Director of Applied Solutions Engineering Joshua Patterson and H2O.ai CTO Arno Candel. They explain how enterprises can use this transformative technology to rapidly innovate using machine learning.
We’ll also be at the O’Reilly AI conference at booth 19 in New York, June 27-29. There, we’ll showcase how H2O.ai’s machine learning algorithms can help you discover lightning-fast insights and allow you to build and deploy production-ready interpretable models, optimized and accelerated on the NVIDIA GPU computing platform.