The start-up company’s technology allows for faster AI model training.
Hewlett Packard Enterprise (HPE) has expanded its artificial intelligence (AI) portfolio by acquiring Determined AI, a start-up that has developed a technology to quickly train AI models using the company’s open-source machine learning platform. The technology allows for faster AI model training at any scale.
HPE will add Determined AI’s software to its AI and high performance computing (HPC) offerings—a suite of technologies that will enable machine learning (ML) engineers to deploy and train ML models. These models will allow ML engineers and data scientists to provide fast and accurate insights from their data in a wide variety of industries. For example, Determined AI claims that its platform accelerated a drug company’s efforts to train an AI from three days to three hours.
“AI-powered technologies will play an increasingly critical role in turning data into readily available, actionable information to fuel this new era,” said Justin Hotard, senior vice president and general manager of HPC and Mission Critical Solutions (MCS) at HPE. “Determined AI’s unique open-source platform allows ML engineers to build models faster and deliver business value sooner without having to worry about the underlying infrastructure.”
One of the most demanding and important stages of ML development is the building and training of optimized ML models at scale. To do so effectively, engineers and scientists need to navigate challenges often found in HPC, such as properly establishing and managing highly parallel software processes and infrastructure with specialized computing, storage and accelerators. And to take full advantage of that infrastructure, engineers also need to program, schedule and train their models efficiently. The resulting complexity can noticeably slow down productivity.
Determined AI’s open-source ML training platform aims to address these issues and remove the complexity and cost that can be associated with ML development. The technology simplifies setup, configuration, management and sharing workstations or AI clusters running on-site or in the cloud—enabling researchers and scientists to focus on innovation and accelerate delivery time. The platform includes features such as accelerator scheduling, fault tolerance, and high-speed parallel and distributed model training. It also allows for advanced optimization of hyperparameters, which are variables whose values are used to control the learning process, and neural architecture search—a technique for automating AI model design.
“Over the last several years, building AI applications has become extremely compute, data, and communication intensive,” said Evan Sparks, CEO of Determined AI. “By combining with HPE’s industry-leading HPC and AI solutions, we can accelerate our mission to build cutting edge AI applications and significantly expand our customer reach.”
Determined AI demonstrates distributed training and hyperparameter search on Amazon Web Services (AWS).
Hewlett Packard’s acquisition of Determined expands the company’s presence in the growing AI market—a market where more and more sectors and industries are adopting AI training and technologies.
HPC demand is expected to grow at 7.1 percent CAGR and hit $55 billion in revenue by 2024. HPC computing power is becoming increasingly in demand as a tool to train and optimize AI models—and deploy them to augment workloads such as simulation and modeling.
Acquiring Determined AI will enable Hewlett Packard to deploy intelligent specialized computing to scalable projects and innovations—helping the company to expand its footprint in this dynamic and rapidly growing market.
Read more about developments in AI at Google Launches Vertex AI, Bringing Machine Learning to the Masses.