What’s New: A new report from analyst firm Kisaco Research positions SigOpt, an Intel company, as a leader in artificial intelligence (AI) software acceleration, due to the quality of its optimization platform, breadth of its market penetration and strength of its go-to-market strategy. By making customers’ AI workloads more efficient, SigOpt helps them get the most out of their hardware investment.
“SigOpt, an Intel company, achieved an impressive score across all dimensions of our KLC evaluation to be ranked as Leader. Customers using SigOpt benefit from productivity gains, faster time to train, and efficient utilization of compute. In combination, this empowers modelers to bring more models into production on a faster timeline, boosting the overall return of AI for enterprises.”
– Michael Azoff, Kisaco Research
What the Study Says: SigOpt is prominent in the machine learning community for its automated, scalable and high-performing hyperparameter optimization solution that can be applied to any parameterized system but is most often used in the context of deep learning tasks. Kisaco Research says the platform can take “operational tasks that are time intensive for machine learning and [make] them easy and seamless, using AI to assist in this task.” For customers, this means a boost in productivity and the ability to tune algorithms faster and utilize compute more efficiently.
The report details how two of SigOpt’s Fortune 500 customers, Two Sigma and an unnamed large global technology consulting firm, use the advanced features that are only available in its platform to solve entirely new business problems. In the case of the consulting firm, SigOpt helped the firm yield productivity gains of over 30% and projected value of nearly $50 million annually.
Primary SigOpt functionalities are designed to help manage, optimize and scale AI applications. (See graphic for a representation of how SigOpt fits into the machine learning technology stack.)
Why It’s Important: The complexity, size and variety of AI models is exploding, making it resource-intensive to efficiently fine-tune models to meet the particular needs of a given project. SigOpt’s platform eases this burden by introducing a more efficient approach to model development: integrating best-in-class automation and productivity tools for model development, including model artifact tracking, training visualization, and sample-efficient hyperparameter optimization. For developers, data scientists, researchers and machine learning engineers, this means a highly intuitive and efficient modeling experience that ultimately leads to much better AI results.
SigOpt’s fully hardware and software-agnostic platform also provides the same user experience throughout – including deep learning, machine learning, reinforcement learning, or simulations. When combined with Intel AI hardware and the consistent developer experience afforded by Intel® oneAPI Toolkits, SigOpt will continue to offer developers the most efficient way to develop and deploy AI models. Individuals and enterprises can access the SigOpt platform by signing up on the SigOpt website.