IBM breaks image recognition software speed record
|Richard Harris in Artificial Intelligence Thursday, August 10, 2017|
Deep learning image recognition software by IBM breaks record for speed of recognition.
The software and new record represents a milestone in making Deep Learning much more practical at scale. Namely, IBM researchers have figured out how to address the bottleneck that is occurring as GPUs become faster - and are too quick to communicate with one another.
The bottom line is that the record IBM broke slashes Deep Learning training time from days to hours, which will enable their customers to more easily address larger technical challenges significantly faster. The software responsible is available to customers through IBM Systems' colleagues via PowerAI.
- Record communication overhead and 95% scaling efficiency on the Caffe deep learning framework over 256 GPUs in 64 IBM Power systems.
- Previous record held by Facebook AI Research: scaling efficiency 89% for a training run on Caffe2, at higher communication overhead.
- Previous record by Facebook: 1 hour
· Using this software, IBM Research achieved a new image recognition accuracy of 33.8% for a neural network trained on a very large data set (7.5M images).
- Previous record published by Microsoft: 29.8% accuracy.
Learn the best ways to organize your app development projects, and keep code straight, clients happy, and breathe a easier through launches.
Bridging the silos of crowdsourcing research to build better solutions, faster.
Write and run code every step of the way, using Android Studio to create apps that integrate with other apps, download and display pictures from the web, play sounds, and more. Each chapter and app has been designed and tested to provide the knowledge and experience you need to get started in Android development.
How to create a profitable, sustainable business developing and marketing mobile apps.