Microsoft Acquires AI Specialist Bonsai
AI services startup Bonsai has been acquired by early investor Microsoft, giving the software giant another automation tool dubbed machine teaching that abstracts machine learning steps so subject experts can train autonomous systems to accomplish specific tasks.
The acquisition would combine Bonsai’s proprietary “deep reinforcement” learning platform with Microsoft Azure cloud tools along with an open source simulator based on Microsoft Research’s AirSim framework. The partners noted that Bonsai’s approach is based on model training in a simulated environment.
Microsoft (NASDAQ: MSFT) said Wednesday (June 20) Bonsai’s platform would be integrated with its simulation tools and reinforcement learning technology to develop an “AI toolchain” for building autonomous systems for industrial control and calibration tasks.
The toolchain would work with Microsoft’s Azure machine learning framework running on Azure cloud with GPUs and Microsoft’s Project Brainwave. The resulting models are to be deployed on the Azure Internet of Things offering, creating what Microsoft claims is a platform for building and operating “brains” for autonomous systems.
Bonsai, Berkeley, Calif., was founded in 2014 by CEO Mark Hammond and Keen Browne with the goal of adding intelligence to hardware and software applications.
AI platform developers argue that many industrial enterprises lack the tools required to move from generic AI platforms to application-specific models that combine advanced machine learning libraries and algorithms to meet specific industry requirements.
In response, Bonsai expanded access to its services last year in a bid to woe customers to jointly develop AI models for individual use cases.
The startup’s platform abstracts the complexity of libraries like TensorFlow, making the programming and management of AI models more accessible to developers and enterprises.
The startup claims large industrial companies as early customers, including those seeking to improve operations via “dynamic control systems” spanning robotics, wind turbines and machine tuning.
“To realize this vision of making AI more accessible and valuable for all, we have to remove the barriers to development, empowering every developer, regardless of machine learning expertise, to be an AI developer,” Microsoft noted in a blog post announcing the deal.