Feature Labs, a data science startup born out of the computer science and artificial intelligence lab at MIT, is launching on Thursday following the completion of a $1.5 million seed round of financing. The round was led by Boston's Flybridge Capital Partners, which also backed big data company DataXu in its early days. Local VC firm First Star Ventures and San Francisco-based 122 West Ventures also participated in the round.
The announcement includes two moves in the company's advisory board and C-suite. As part of the funding round, Chip Hazard, a general partner at Flybridge, will be part of the company’s board of directors. Also, software industry veteran John Donnelly III, a former senior vice president at Crimson Hexagon, will be joining Feature Labs as COO after having served in an advisory role to several early stage companies coming out of MIT.
Founded in 2015 by Max Kanter, Kalyan Veeramachaneni and Ben Schreck, Feature Labs positioned itself at the convergence of big data, machine learning and artificial intelligence. It's not a secret that enterprise companies are increasingly willing to pay top dollars to extract knowledge from unstructured data, as proved by the "much faster than average" projection of growth of jobs for research scientists certified by the Bureau of Labor Statistics.
Allowing data scientists to do their job, however, poses a challenge: raw data needs to be prepared before feeding it to machine learning algorithms, a process - called "feature engineering" - that Feature Labs said it can automate and make more efficient. In a nutshell, the preparation process of the data includes creating new features, and that's where the company comes into the picture.
"Feature Labs is unique because we automate feature engineering, which is the process of using domain knowledge to extract new variables from raw data that make machine learning algorithms work," co-founder and CEO Max Kanter said in an interview with Techcrunch. "By giving data scientists this automated process, they can spend more time figuring out what they need to predict.”
Specifically, Feature Labs uses a software called "Deep Feature Synthesis," which can be let loose on any kind of dataset, including credit card transactions. As it explores the relationships between rows and tables in a dataset, the software makes recommendations of the top features to use for specific prediction problems, therefore enabling machine learning algorithms to work their magic.
Feature Labs said that organizations including Kohl’s, NASA, BBVA, DARPA, Monsanto and MIT have been using Feature Labs software to automate their feature engineering processes.
"Every industry leading enterprise company is trying to deploy machine learning applications into production, however all are facing the challenge of a shortage of data scientists to do the work,” Chip Hazard said in a statement. “With software tools that can automatically identify predictive patterns in data more than 10 times faster than traditional approaches and a rock star founding team, Feature Labs is well positioned to fundamentally change the way organizations approach using their data.”
