A Seoul-based startup building cloud-based, automated machine learning tool is coming out of stealth mode with a new product and $1 million investment.
While working on a security project with machine learning (ML) at Postech, CEO Jinyoung Choi recognized the gap between the ML theories taught at school and ML practices for production data in the industry.
Building upon this realization, Jinyoung and Anthony saw that there will be a strong need for a universal ML tool rather than an end-to-end solution for specific problems.
XBrain’s cloud-based, automated ML system, Daria, intends to be just that, allowing data science experts and novices alike to build predictive models and deploy them to action.
Customers can achieve ML pipeline automation through Daria in all 4 phases of ML pipeline: 1. Data construction and refinement, 2. Deciding and training ML model, 3. Deployment and integration of ML model and 4. Model performance validation and maintenance.
After proving its strength in ML automation technology at 2016 Cha-Learn Automated Machine Learning Challenge (3rd Place) and 2017 ICML (International Conference on Machine Learning) AutoML Workshop, XBrain has been gaining a customer traction through its closed beta program with around 30 companies across various industries.
XBrain’s main customers include UBASE, a business process outsourcing company, Penta Security Systems, a security solutions company, and a social dating application, Amanda (Nextmatch). Customers share that Daria’s biggest strength is the immediate integration of production business operations, regardless of the industry.
Nextmatch’s data analytics team wanted to build a customer churn prediction model to strategically distribute promotion codes and utilize gained insights for future feature development efforts. However, processing accumulated data to build an optimal feature set was time consuming, including having to, make model selections and tune parameters to reach targeted performance. Nextmatch vastly reduced hours of iterative work through Daria by training multiple feature sets in parallel to compare the performances and immediately applying the best fit model in its production service. Today, Daria is used to predicting customer churn of 1 million users daily.
Xbrain raised 1 million USD from private investors and TIPS (Tech Incubator Program for Startup – Korean Government Fund). Through this investment, XBrain “plans to grow the team and accelerate product development. It will focus on creating success stories with small and medium IT companies in Korea where mobile and internet service developments are vibrant, building a strong foundation for global market expansion,” said CEO Jinyoung Choi.
“Machine learning is not the ultimate solution, but a tool to solve various problems businesses have. Many businesses are currently unable to fully utilize ML due to the lack of resources and technical understanding. Although XBrain will initially focus on small and medium IT companies, we aim to become the default enabler of ML for any business organization,” commented Co-founder Anthony Cha.