Deep learning startup Skymind raises $3 million, launches Intelligence Layer distribution

Skymind, a startup that promotes the use of the Deeplearning4j open-source software for deep learning — a type of artificial intelligence (A.I.) — is announcing today that it has raised a $3 million seed round.

There are many open-source frameworks for deep learning, which generally involves training artificial neural nets on lots of data, such as labeled photos, and then getting them to make inferences about new data. Deeplearning4j is primarily written in Java and is meant to be integrated into companies’ existing architectures.

[aditude-amp id="flyingcarpet" targeting='{"env":"staging","page_type":"article","post_id":2066296,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"business,entrepreneur,","session":"C"}']

Now the team has built an enterprise distribution called the Skymind Intelligence Layer (SKIL) that packages up Deeplearning4j and other open-source tools from Skymind and that features a graphical user interface. It can be run in public clouds or in on-premises data centers, on standard x86 chips, optionally those with graphics processing units (GPUs), or Intel Xeon Phi chips or IBM Power8 chips.

Deep learning has been hot in the past few years, as Apple, Google, Facebook, and Microsoft have unveiled smart features that take advantage of the technology. A.I. has gotten hot and then cold a few times over the past several decades. Nicholson believes that this time is different because algorithms are better than they’ve ever been, because lots more data is available for training, and because hardware is better (and changing rapidly).

“Deep learning is getting human-level accuracy, sometimes super-human — a quantum leap over traditional machine-learning algorithms,” Skymind cofounder and chief executive Chris Nicholson told VentureBeat in an interview. With that level of performance, things like self-driving cars become more realistic. “Whole new product lines and whole new businesses will be founded upon that,” he said.

Earlier this year, one deep learning startup, Nervana, was acquired by Intel, and another, MetaMind, was acquired by Salesforce. Skymind partly stands out because it pushes open-source software (but now also a commercial distribution) and doesn’t dabble in hosted services or custom hardware.

San Francisco-based Skymind was founded in 2013 and is based in San Francisco, with 15 geographically dispersed employees. Skymind’s investors include Tencent, GreatPoint Ventures, Hemi Ventures, Liquid 2 Ventures, Mandra Capital, SV Angel, and Y Combinator. After launching in 2014, Skymind now has half a dozen customers, including Canonical and Orange.