Landing AI Inc., a startup led by artificial intelligence pioneer Andrew Ng, has raised $57 million in funding from a group of investors that included the venture capital arms of Intel Corp. and Samsung Electronics Co. Ltd.

The investment was announced this morning. Landing AI will reportedly use the proceeds from the round to double its 75-person workforce, an effort set to include hiring more data scientists. The startup said it will also invest in growing its engineering, product, sales and marketing teams.

Landing AI is led by founder and Chief Executive Officer Andrew Ng (pictured). Ng previously co-founded Google Brain, Google LLC’s deep learning research group. Before launching Landing AI, Ng was also chief scientist at Baidu Inc. and co-founded Coursera Inc., a publicly traded online educational company.

Landing AI has developed a software platform called LandingLens that enables manufacturers to build AI models for automatically detecting product defects. According to the startup, companies using its platform to identify manufacturing issues often achieve “substantial improvements in system accuracy.” LandingLens also promises to address several of the technical challenges historically involved in developing AI applications. 

One common source of complexity in deep learning projects that Landing AI is working to address with its platform is a shortage of large-scale training datasets. The process of building AI software involves training a neural network on a large volume of information so that it can learn to identify patterns and become more accurate. But manufacturers often have only a limited amount of training data available for AI projects, which makes development more complicated. 

“You don’t always need big data to win with AI. You need good data that teaches the AI what you want it to learn,” Ng said in a statement today. “AI built for 50 million data points doesn’t work when you only have 50 data points. By bringing machine learning to everyone regardless of the size of their data set, the next era of AI will have a real-world impact on all industries.”

A second challenge Landing AI is tackling is the complexity of carrying out AI training. There are some cases where training a neural network on a dataset requires writing multiple pages of custom code. According to Landing AI, its platform makes it possible to perform the task with a few mouse clicks.

According to Landing AI, there are numerous manufacturing use cases that can be streamlined with deep learning. 

In the electronics market, manufacturers must ensure that the chips they use in their products don’t contain any flaws. But catching every defective chip is difficult. There’s a large number of potential manufacturing flaws that can prevent a processor from working as intended and tracking them all is difficult using traditional production line monitoring systems. Deep learning can provide a more practical and versatile means of spotting issues, Landing AI says. 

In the auto sector, meanwhile, detecting components that have been welded together incorrectly often represents a challenge as well. Slight component variations are easily mistaken for welding flaws by traditional monitoring systems, which creates false positives. Landing AI says deep learning can more effectively differentiate manufacturing defects from false positives.

The startup also lists numerous other use cases on its website across the auto and electronics industries. Moreover, Landing AI says, its technology can be applied by organizations in the healthcare sector as well. 

Photo: Landing AI