Superb AI generates relevant data sets for machine learning

A major challenge for machine learning is the collection of relevant data. In order to train an algorithm, a large dataset must be available. Superb AI has now been set up for this purpose, a startup that can help companies put together datasets for a project. Artificial intelligence helps to tag these datasets and therefore makes them suitable for use.

Hyun Kim is the CEO and co-founder of the startup and states that one of the stumbling blocks for companies that want to apply AI and machine learning in their apps is to find suitable datasets with which they can train their models. Superb AI uses AI to create suitable training data for large tech companies, says Kim to the TechCrunch site. Together with customers, the company develops a suitable dataset, so that these customers can get to work much faster.

Quick tagging

Usually, companies working on a machine learning project hire people to tag data. This is expensive, but also error-prone and difficult process. It is also often difficult for a company to find a very suitable dataset. Kim and his co-founders worked on artificial intelligence before and came up with the idea to use AI to tag datasets.

As a result, Superb AI can also find and tap into new data sources that companies have not yet thought of. In addition, it works faster by not relying on the slow and error-prone handiwork. The company’s software can tag images and videos up to ten times faster than human counterparts can.

People continue to be needed

Although Superb AI uses artificial intelligence to tag data sets, it still relies on people to supplement this software. This works as follows: the company splits data sets into as many separate components as possible. This allows the tagging to take place automatically. But if the datasets turn out to be too complex for that, it engages a human being to do this work. The AI watches, learns from its decisions and takes over at some point.