Today’s artificial intelligence (AI) technologies are usually run on machine learning algorithms. These technologies operate on a neural network — a system designed to mimic the human brain inner functioning —that is called deep learning. Currently, most of AI advances are basically due to deep learning, with developments like AlphaGo, the Go-playing AI produced by Google’s DeepMind.
Now, IBM has developed an AI that makes the entire machine learning process faster. Instead of running complex deep learning models on just a single server, the team at IBM led by Research’s director of systems acceleration and memory Hillery Hunter, to manage efficiently and scale up distributed deep learning (DDL) method using multiple servers.
“The idea is to transform the rate of how fast you can train a deep learning model and boost their productivity,” Hunter told Fortune. Beforehand, it was difficult to implement DDL setups because of the complexity needed to keep the processors in-synchronisation. The IBM Research team has managed to use 64 of its Power 8 servers to facilitate data processing. Each processor was linked using Nvidia graphical processors and a fast NVLink interconnection, ensuing in what Hillery’s team calls PowerAI DDL.
BOOSTING PROCESSING POWER
A deep learning network to process models before takes days to but now it could take only hours. “Our aim is to reduce the latency time associated with deep learning training from days or hours to minutes or seconds, and enable superior accuracy of these AI models,” Hunter wrote in an IBM Research blog.
In their recent study published online, the IBM team claimed that they managed a 95 percent scaling efficiency transversely 256 processors while setup using a deep learning framework developed at the University of California Berkeley. They also recorded a 33.8 percent image recognition accuracy rate, processing 7.5 million images in a tiny over seven hours, beating Microsoft’s record of 29.8 percent in 10 days.
Some, however, are skeptical of the achievement. Patrick Moorhead, president and founder a Texas-based technology research company told Fortune that 95 percent seemed too good to be true. Still, IBM’s achievement could potentially boost the capabilities of deep learning networks. It could lead to improvements in how Artificial Intelligence (AI) helps in medical research and in independent systems, cutting down the time necessary that makes progress slower.