Home / SmartTech / Superb AI generates custom training data for machine learning projects – TechCrunch

Superb AI generates custom training data for machine learning projects – TechCrunch

One of the major challenges in developing a machine learning project is simply to get enough relevant data to train the algorithms. Superb AI, a member of the Y Combinator Winter 2019 class, can help here. The startup helps companies create custom records to meet the needs of each project by using AI to speed up the tagging process.

Hyun Kim, CEO and co-founder of the startup, says one of the biggest obstacles for businesses When you try to include AI and machine learning in their applications, you get a set of appropriate data that trains the models. "Outstanding AI uses AI to create custom AI training data for large technology companies. Customers work with us to develop machine learning in their products many times faster than they could, "Kim told TechCrunch.

Kim and co-founders CTO Jungkwon Lee, mechanical engineers Jonghyuk Lee and Moonsu Cha and Hyundong Lee, head of sales and operations of APAC (based in Seoul, South Korea), all worked in the field when they identified the data problem and decided to start a business to solve this problem.

Traditionally, businesses are working a machine The learning project will hire people to tag data, but this was expensive and error-prone, provided you even had the data you could work with, and Kim and his co-founders who worked on AI projects and the topic The idea was to get AI to work on the day part of the problem.

"Instead of si Leaving things slow and error-prone For manual work, Superb AI uses a proprietary deep-learning AI that helps people identify pictures and videos up to 1

0 times faster, "explains Kim. The company will also help find data sources for companies for which data is initially unavailable.

Kim says that they do not completely remove humans from the process, but the detection accuracy of combining human resources improves with artificial intelligence. He says that this requires a few steps. First, the training data is divided into as many components as possible to automate each piece individually. If the data is too complex and the AI ​​tools can not automate tagging, they use a second approach called "human in the loop". When data is provided by people with data, the AI ​​can learn over time and eventually take over more and more The process.

The co-founders decided to apply to Y Combinator to gain a foothold in Silicon Valley, where they could expand their market beyond their native South Korea. "It was definitely a game changer. The knowledge and experience we have gained from YC partners and co-entrepreneurs is truly incredible. And the huge YC network also helped us find our first customers in the Valley. "The company, which was formed last October, employs up to 13 people, including the co-founders, has raised $ 300,000 in seed investments and Kim has already received the same amount of revenue from the product.

Source link