The Transform Technology Summits start October 13th with Low-Code/No Code: Enabling Enterprise Agility. Register now!
San Mateo, California-based Superb AI, a startup developing a platform for AI-enabled training data, today announced that it closed a $9.3 million financing round. Cofounder and CEO Hyunsoo Kim said that the financing will allow the three-year-old company to ramp up hiring and scale customer adoption as well as expand and accelerate product development.
Training AI and machine learning algorithms requires plenty of annotated data. But data rarely comes with annotations. The bulk of the work often falls to human labelers, whose efforts tend to be expensive, imperfect, and slow. Moreover, it requires tremendous resources and effort to figure out if the labels themselves are accurate, particularly for companies that use hundreds of thousands — or even millions — of images to train and retrain their AI models.
Superb AI, which counts teams at Qualcomm, LG, and Samsung among its customers, claims to be able to train AI models using incredibly small datasets without requiring human assistance in the workflow. The company’s Auto-Labeling AI can ostensibly annotate thousands of images in a matter of seconds, reducing the time spent on labeling from 80% to 20%.
“At Superb AI, we’ve developed an ‘uncertainty estimation’ technique with which a model can measure how confident it is with its own labeling predictions,” Kim explained. “In other words, our Auto-Labeling AI outputs annotations (i.e. bounding boxes and the corresponding object class) and simultaneously outputs how confident it is with each annotation. Therefore, it requests human verification only in cases the AI is uncertain about, and as a result, reduces the amount of work that goes into human verification of labels.”
Kim says that Superb AI also tailors preexisting Auto-Labeling AI models to new tasks through techniques like Bayesian active learning and transfer learning. Transfer learning, which focuses on storing knowledge gained by a model while solving a problem and applying it to a different but related problem, is particularly advantageous because it requires only a small amount of fine-tuning data, he says.
Superb AI’s platform provides several services aimed at AI engineers, including a tool that slices, partitions, and manipulates training data. On the project management side, Superb AI offers an issue tracker that enables developers to collaborate with data stakeholders and an analytics dashboard from where managers can share datasets, charts, and impose role-base controls for their organization.
With Superb AI, customers can import model predictions and push data to model environments on the fly. When working with computer vision datasets, they gain access to image classification, segmentation, keypoint labeling, video object tracking tools, and more as well as datasets built by Superb AI’s engineering team.
“The financing will allow us to rapidly scale the business while meeting the growing needs of our customers,” Kim said. “Because our deep learning AI can label and analyze images and videos up to 10 times faster than manual processes, we are able to eliminate many of the traditional bottlenecks and allow every industry to more easily adopt AI technology.”
Atinum Investment led the series A round in Superb AI with participation from existing and new investors including Premier Partners, Stonebridge Ventures, Murex Partners, KT Investments, and Duke University’s Angel Network. It brings the company’s total raised to date to $11.3 million.
VentureBeatVentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact. Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:
- up-to-date information on the subjects of interest to you
- our newsletters
- gated thought-leader content and discounted access to our prized events, such as Transform 2021: Learn More
- networking features, and more