One of many huge challenges of creating a machine studying undertaking might be merely getting sufficient related information to coach the algorithms. That’s the place Excellent AI, a member of the Y Combinator Winter 2019 class, can assist. The startup helps firms create custom-made information units to satisfy the necessities of any undertaking, utilizing AI to hurry up the tagging course of.
Hyun Kim, who’s CEO and co-founder on the startup, says one of many huge obstacles for firms making an attempt to include AI and machine studying into their purposes is arising with a set of appropriate information to coach the fashions. “Excellent AI makes use of AI to make custom-made AI coaching information for giant tech firms. Purchasers work with us to develop machine learning-based options of their merchandise a number of occasions quicker than they may themselves,” Kim advised TechCrunch.
Kim and his co-founders (CTO Jung Kwon Lee, machine studying engineers Jonghyuk Lee and Moonsu Cha and Hyundong Lee, head of APAC gross sales and operations, who is predicated in Seoul, South Korea) all have been working within the discipline after they recognized the information drawback and determined to launch an organization to unravel it.
Historically, firms engaged on a machine studying undertaking will rent human employees to tag information, however this has been costly and error susceptible, assuming you even had the information to work with. Kim and his co-founders, who labored on AI tasks and studied the topic in school, got here up with the concept of placing AI to work on the tagging a part of the issue.
“As a substitute of counting on gradual and error-prone guide labor, Excellent AI makes use of proprietary deep studying AI that assists people to attain as much as 10x quicker labeling of photographs and movies,” Kim defined. The corporate will even assist discover information sources for firms that don’t have any information to start with.
Kim says that they don’t take people out of the method utterly, however they do improve tagging accuracy by combining human employees with synthetic intelligence underpinnings. He says that this includes a few steps. First, it splits coaching information into as many parts as attainable to be able to automate each bit separately. If the information is just too advanced, and the AI instruments can’t automate the tagging, they use a second strategy referred to as “human within the loop.” As people label information, the AI can study over time and ultimately take over an increasing number of of the method.
The co-founders determined to use to Y Combinator to realize a foothold in Silicon Valley, the place they may develop their market past their native South Korea. “It’s positively been a sport changer. The quantity of data and expertise we gained from the YC companions and fellow entrepreneurs is actually unbelievable. And in addition the huge YC community helped us discover our early prospects within the Valley,” Kim mentioned.
The corporate, which launched final October, is as much as 13 staff, together with the co-founders. It has raised $300,000 in seed funding and has already generated the identical quantity in income from the product, in line with Kim.