Cardiovascular disease is the leading cause of death for U.S. men and women. Doctors diagnose someone with the condition every 43 seconds.
Finding new and better ways to speed the diagnosis of heart disease couldn’t be more important.
So, we’re pleased to take part in a program aimed at applying technology to help — potentially improving the care of heart disease patients, and helping them live longer, healthier lives.
In the competition, data scientists from around the world will work to help transform the diagnosis of heart disease.
During the 90-day competition, teams will develop machine learning algorithms to analyze a thousand MRI images of existing patients. The goal: systems that automatically identify early indicators of heart disease.
The team with the most accurate algorithm will win $125,000. The second- and third-place teams get $50,000 and $25,000, respectively.
Winning is great. Solving real-world problems is even better.
GPU-Accelerated Deep Learning the Ideal App
Machine learning techniques are enabling major advances across science and industry. And deep learning, the fastest growing segment of the machine learning field, promises to make much more possible.
Deep learning excels at problems like face detection and recognition, speech recognition and image classification. It’s already delivering better than human levels of accuracy for some tasks.
It’s an ideal approach for tasks such as identifying heart disease indicators from MRIs. Deep learning algorithms can be applied for medical image analysis for computer-aided diagnosis, image segmentation and registration.
But training the deep neural networks for this task is time consuming. It can take days or weeks.
GPU acceleration can slash this time to hours. This lets data scientists more quickly design, train and improve deep neural networks. As a result, they can build larger, more sophisticated — and ultimately, more accurate — networks.
Joining the Fray
We’re so confident in the power of GPU-accelerated deep learning that we’re competing, too. We’re forming a team with Booz Allen Hamilton to play in the Data Science Bowl.
Our goal isn’t to win. We’re not submitting our work for prize consideration.
Instead, by showcasing the power of GPU-accelerated deep learning — and sharing our progress throughout the competition — we hope to help other teams learn and advance their work. At the same time, we’ll understand what we can build in the future to better suit this task.
NVIDIA to Highlight Contest Results, Winners at GTC
As part of our participation in the Data Science Bowl, NVIDIA is contributing $25,000 in prize money. We’re also providing free access to online deep learning educational materials to all participants.
We’ll also provide the three winning teams with free passes to our GPU Technology Conference next year (April 4-7, 2016, in Silicon Valley). It’s the premier showcase for GPU technology in deep learning, medical imaging and other fields.
At GTC, we’ll host the Data Science Bowl winning teams and give them an opportunity to share their work, and the positive impact of GPU-accelerated deep learning.