Practical AI Accelerating ML innovation at MLCommons
Jan 19, 2021
David Kanter, the Executive Director of MLCommons, shares insights on accelerating machine learning innovation. He discusses the organization’s three key pillars: benchmarks like MLPerf, the People's Speech dataset, and best practices via MLCube. Kanter emphasizes the importance of equitable standards and collaborations in ML to enhance accessibility, particularly for underserved communities. He also draws parallels between AI's evolution and early aeronautics, showcasing how standardized components can drive innovation and community engagement.
AI Snips
Chapters
Transcript
Episode notes
MLPerf's Role
- MLPerf provides common metrics for measuring progress in machine learning, like a barometer.
- It facilitates a shared language among engineers, researchers, and marketers.
Accuracy and Standardization
- Accuracy in machine learning is achievable with sufficient compute and data, as shown by Greg Diamos's research.
- David Kanter compares current ML to early aeronautics, highlighting the need for standardization.
MLCommons' Mission
- MLCommons aims to advance ML innovation and broaden its access, focusing on benchmarks, data sets, and best practices.
- They want to help analog companies adopt AI by providing standardized components like TensorFlow and PyTorch.