01. About ConvNetSharp

ConvNetSharp can be used to train and evaluate convolutional neural networks (CNN) in .NET

ConvNetSharp started initially as C# port of ConvNetJS. This port was done for learning purposes but new features were eventually added:

• Single/double precision
• Other types of layers
• GPU support with Cudnn and custom kernels
• Computation graph (ConvNetSharp.Flow)

Goals of this library:

• Deep learning in C#
• Readable (algorithms implementations should be found easily)
• Reasonable performances: it should fast enough to be used in real applications