Microsoft this week released an updated version of its Cognitive Toolkit, previously known as CNTK.
Continue Reading Below
Developed internally by computer scientists who wanted a tool to help conduct research more quickly and effectively, the platform has become an essential tool for customers conducting various deep-learning tasks.
"We've taken it from a research tool to something that works in a production setting," Frank Seide, a key architect of Microsoft Cognitive Toolkit (pictured, left), said in a statement.
The latest version, available now on GitHub via an open-source license, includes new functionality allowing developers use the Python or C++ programming languages. It can run on computers with traditional CPUs or GPUs (including Azure's offering, currently in preview), and handles anything from small datasets to "very, very large ones," the company said.
Aimed at everyone from small startups to major technology companies, Microsoft Cognitive Toolkit could power the next Skype Translator or Cortana digital assistant.
Cooling specialist Liebherr is already using it: The firm installed in its refrigerators cameras that recognize individual food items and automatically incorporate information into an inventory shopping list.
The Bing relevance team also employs it in an effort to find better ways to discover hidden connections in search terms.
The field of deep learning has exploded in recent years as more researchers run machine-learning algorithms using deep neural networks, designed to mimic the way the human brain works. Redmond has used this approach to create systems that can do everything from translate conversations to identify the objects in a photograph or video.
Late last month, the tech titan unveiled a new division focused entirely on AI: 5,000 computer scientists and engineers working to standardize artificial intelligence.
Microsoft is also one of six industry groups to join forces for a new nonprofit, the Partnership on Artificial Intelligence to Benefit People and Society. Along with Amazon, DeepMind, Google, Facebook, and IBM, Redmond wants to "maximize [the] potential [of AI] and ensure it benefits as many people as possible."
Here's to hoping that doesn't include the company's failed chatbot attempt: Tay went from happy-go-lucky, human-loving friend to full-on racist in one day.