NEC’s AutoML Spinoff Takes Off
A data science spinoff from NEC Corp. has raised additional early stage funding to accelerate development of its automated machine learning platform.
DotData, San Mateo, Calif, was spun off from its Japanese parent company last year. The new company announced this week it raised $23 million in Series A funding 18 months after its launch and seed funding round. To date, DotData has raised $43 million.
The latest funding announced on Wednesday (Oct. 30) was led by the Tokyo-based venture fund JAFCO with participation by Goldman Sachs (NYSE: GS). NEC (TYO: 6701) is an existing investor.
The startup said it would use the new funding to expand its data science automation platform and other efforts that so far include the launch of several products. It also claims a more than 300 percent increase in annual revenues.
DotData bills itself as among the first AutoML vendors to offer a platform covering the entire data science life cycle from automation to production.
The spinoff is led by CEO Ryohei Fujimaki, who worked on more than 100 data science projects at NEC before the Japanese computer giant spun out Fujimaki’s team last year as an independent software vendor. Fujimaki, the youngest research fellow ever appointed in the 119-year history of NEC, said the funding would be used to accelerate product development centered on its flagship AutoML platform.
The startup’s automation strategy is built around reducing the time to move and prepare data from months to days, democratizing data usage across organizations and, ultimately, providing up to a ten-fold increase in operational data science projects.
It’s “AI-powered feature engineering” framework is combined with AutoML tools and a proprietary API to automate manual steps such as data collection and “last mile” ETL. Data science projects are then moved to production via machine learning and reporting tools. DotData’s secret sauce resides in its AI technology that automates feature engineering, described by the startup as “typically the most challenging and the most time-consuming part of data science.”
Feature engineering is used to automate data cleansing, aggregation and other data preparation steps, then combines hundreds of tables with complex relationships and billions of rows into a single feature table, the company said.
The AutoML platform also includes a graphical user interface designed to speed data science efforts by reducing requirements such as coding skills in SQL, Python or R.
DotData’s proprietary algorithm automatically selects data features to be used for downstream machine learning processes. The software also generates a description that explains why it selected specific features.
“We are the only platform that can automate this end-to-end process on raw data, to prepare data for feature engineering in machine learning,” Fujimaki told Datanami earlier this year. “Our features are very transparent and easy to understand by domain experts. This is the key of our automated feature engineering.”
The company announced in early October that it has been added to Microsoft’s (NASDAQ: MSFT) list of startup partners developing AutoML tools for deploying machine learning in production on Azure.