Microsoft Cognitive Toolkit, a system for deep learning that is used to speed advances in areas such as speech and image recognition and search relevance on CPUs and NVIDIA GPUs today gets an update. It’s for now run on multiple GPUs, including Azure’s GPU offering, which is currently in preview.
Previously known as CNTK, the beta version of the toolkit, include more programming flexibility, advanced learning methods like reinforcement learning and extended API support for training and inference from Python, C++ and BrainScript.
So developers can use popular languages and network. it also help researchers do a type of artificial intelligence work called “reinforcement learning.”
The toolkit was initially developed by computer scientists to do their own research more quickly and effectively. “We’ve taken it from a research tool to something that works in a production setting,” said Frank Seide.
It’s used to develop commercial grade AI in popular Microsoft products like Skype, Cortana, Xbox and Bing. “This is an example of democratizing AI using Microsoft Cognitive Toolkit,” said Xuedong Huang, Microsoft distinguished engineer.
Also, Object detection capability, based on deep learning, is now a part of the toolkit. The approach is based on a method called Fast R-CNN, which was demonstrated to produce state-of-the-art results for Pascal VOC, one of the main object detection challenges in the field.
Fast R-CNN takes a “deep neural network (DNN), which was originally trained for image classification by using millions of annotated images, and modifies it for the purpose of object detection.”
Pre-trained image classification DNNs are generic and powerful, allowing this method to also be applied to many other use cases, such as finding cars or pedestrians or dogs in images.
Developers and researchers can start training with the Microsoft Cognitive Toolkit for free by visiting https://aka.ms/cognitivetoolkit.