Datasaur snags $3.9M investment to build intelligent machine learning labeling platform – ClearTips
As machine learning has grown, one of the major hurdles has been to label things so that the machine learning application can understand the data it is working with. A member of the Y Combinator Winter 2020 batch, Datsaur today announced an investment of $ 3.9 million to help solve that problem with a platform designed for machine learning labeling teams.
The funding announcement, which included a pre-seed amount of $ 1.1 million from the previous year and a $ 2.8 million seed after graduating from Y Combinator in March, included investments from the starting capital, Y Combinator. And OpenAI CTO Greg Brockman.
Company founder Evan Lee says he has been working in various AI-related capabilities for over seven years. The first time his mobile gaming startup, Loki Studios, was acquired by Yahoo! In 2013, and Lee eventually moved to the AI team, and most recently to Apple. Despite the company, he consistently saw a problem around organizing machine learning labeling teams, one that he felt he was uniquely positioned to solve because of his experience.
“I’ve spent millions of dollars [in budget over the years] And spent countless hours collecting labeled data for my engineers. I found out that it was a thing that was a problem for all the companies I was on. And they were simply reinforcing the wheel and the process. So instead of reinvesting in my most recent company Apple for a third time, I decided to solve it for the industry once and for all. And that’s why we started Datsaur Last year, Lee told ClearTips.
They created a platform to speed up human data labeling with AI supplements involving humans. The platform consists of three parts: a labeling interface, an intelligence component that can recognize the basics, so the labeler is not recognizing the same thing over and over again, and ultimately a team organizing component.
He says the area is hot, but at this point most labeling consulting solutions are included, which exclude contractors from labeling. He points to the figure and sales of figure eight in March 2019, which snagged $ 100 million last year, which are examples of other startups trying to solve the problem in this way, but believes Is that his company is doing something completely different by making software-based solutions
The company currently offers a cloud and on-premier solution based on customer requirements. It has 10 employees with plans to hire in the next year, although it did not share an exact number. As he does, he says that he is working with a partner in Investor to get started on creating a positive and inclusive culture within the organization, and that includes hiring a diverse workforce during the creation of the company. Conversation is included.
“It seems to me that this is just a matter of standard CEO, but it’s something we absolutely value in our top funnel for the hiring process,” he said.
As Lee builds his platform, he also worries about the bias built into the AI system and the detrimental impact it can have on society. He says he has spoken to customers about the role of labeling in prejudice and how to counter it.
“When we talk with our customers, I talk to them about bias from their labels and our product has the ability to delegate the same project to many people. And I explain to my clients that it can be more expensive, but from personal experience I know that it can dramatically improve results to get multiple perspectives on exactly the same data, ”he said .
Lee believes that humans continue to be involved in the labeling process in some way, even making parts of the process more automated. “Nature of our existence [as a company] Humans will always need a loop, […] And moving forward I think it is really important that as we use AI in longer and more and more long tail cases, we will need to continue to educate humans and inform AI , And this is going to be an important part of this technique. “