Wave Computing Joins Industry and Academic Leaders to Define Next-Generation Machine Learning Industry Benchmark
THE O’REILLY AI CONFERENCE, New York, May 2, 2018 – Wave Computing®, the Silicon Valley company that is revolutionizing artificial intelligence (AI) and deep learning with its dataflow-based solutions, announced today it is joining forces with innovators from Baidu, Google, and more, to help define a new machine learning benchmark for data scientists. The MLPerf benchmark will enable more accurate evaluation of the performance and scalability of neural network workloads, as well as the ability of compute systems to train and inference production-sized datasets. MLPerf aims to set a new standard for the AI industry by using real life scenarios, unlike many benchmarks used today which are typically based on outdated methodologies or biased to traditional hardware architectures. When finalized, MLPerf will be openly available in the public domain for general use and evolution.
As the demand for AI grows, data scientists need a more reliable way to anticipate how their models and acceleration platforms will behave while performing tasks spanning natural language understanding to image recognition. The difference of seconds between predicted and actual results can mean millions of dollars in lost revenue opportunities for companies using AI in industries spanning retail, financial services and public cloud. By drawing on lessons from the 40-year history of computer benchmarking and leveraging some of the brightest minds in academia and industry, MLPerf will “close the gap” while enabling a fair comparison of AI algorithms and solutions, helping improve AI innovation.
“Wave Computing is honored to have been invited by the organizers of the MLPerf benchmark consortium to participate in this exciting and much-needed effort,” said Dr. Chris Nicol, CTO of Wave Computing. “The AI industry clearly needs a better benchmark solution to keep pace with the quickly evolving nature of neural networks. We look forward to contributing to this important collaboration between industry professionals and academic researchers.”
About the MLPerf Benchmark
MLPerf will be a new benchmark for measuring the speed of machine learning software and hardware based on the time it takes to train deep neural networks to perform tasks including recognizing objects and translating languages. MLPerf will evaluate metrics such as quality, accuracy, execution time, power, and cost to run the suite. The effort is supported by a broad coalition of organizations including Baidu, Google, Intel, Sambanova, and Wave Computing, as well as researchers from Stanford University, University of California, Berkeley, University of Minnesota and University of Toronto.
Wave Computing and the Wave Computing logo are trademarks of Wave Computing, Inc. All other trademarks are used for identification purposes only and are the property of their respective owners. ©2018 Wave Computing, Inc. All rights reserved.