What is it

Microsoft Cognitive Toolkit, or CNTK, is an open-source deep learning toolkit that was designed for commercial use and distributed deep learning. However, it is important to note that CNTK is no longer actively developed, with the final major release being CNTK 2.7.

Key features

• Describes neural networks as a series of computational steps via a directed graph.
• Allows users to easily realize and combine popular model types such as feed-forward DNNs, convolutional neural networks (CNNs), and recurrent neural networks (RNNs/LSTMs).
• Implements stochastic gradient descent (SGD, error backpropagation) learning with automatic differentiation and parallelization across multiple GPUs and servers.
• Can be included as a library in Python, C#, or C++ programs, or used as a standalone machine-learning tool through its model description language (BrainScript).
• Supports 64-bit Linux or 64-bit Windows operating systems.
• Is one of the first deep-learning toolkits to support the Open Neural Network Exchange ONNX format, an open-source shared model representation for framework interoperability and shared optimization.

Pros

• Easy to use and combine popular model types.
• Supports stochastic gradient descent (SGD, error backpropagation) learning with automatic differentiation and parallelization across multiple GPUs and servers.
• Can be used as a library in Python, C#, or C++ programs, or as a standalone machine-learning tool through its model description language (BrainScript).
• Supports 64-bit Linux or 64-bit Windows operating systems.
• Is one of the first deep-learning toolkits to support the Open Neural Network Exchange ONNX format.

Cons

• No longer actively developed, with the final major release being CNTK 2.7.
• May not be as feature-rich as some other deep learning toolkits.

Summary

CNTK is an open-source deep learning toolkit that was designed for commercial use and distributed deep learning. It is easy to use and combine popular model types, and it supports stochastic gradient descent (SGD, error backpropagation) learning with automatic differentiation and parallelization across multiple GPUs and servers. CNTK can be used as a library in Python, C#, or C++ programs, or as a standalone machine-learning tool through its model description language (BrainScript). It supports 64-bit Linux or 64-bit Windows operating systems, and it is one of the first deep-learning toolkits to support the Open Neural Network Exchange ONNX format. However, it is important to note that CNTK is no longer actively developed, with the final major release being CNTK 2.7.