EISNOVA
Emulation Interface using Symbolic Neural Object for Visualization and Analysis
Visual System C provides a set of CNTK library (based on the CNTK C++ API) that could be used to model a wide variety of Neural Network. The CNTK library includes important blocks such as MiniBatch Reader, Convolution, Pooling and many other processing blocks. There are also blocks that train and generate trained model for various experimentation. Intermediate results can be inspected by collecting them using Signal Collection block.
In this example, the CNTK library set is used to model a convolutional neural network (CNN, or ConvNet). It is a type of feed-forward artificial neural network made up of neurons that have learnable weights and biases, very similar to ordinary multi-layer perceptron (MLP) networks. The model uses the MiniBatch Reader to read a training file Train-28x28_cntk_text.txt to train the CNN model. The CNN model uses two convolution layers, followed by a deep learning layer to recognize handwritten numbers (from 0 to 9). The trainer block trains the set of parameters in these neural network layers. Classification error block can detect errors. Signal collection blocks are used to collect various intermediate results. After the model is trained, the model can be tested using the Test-28x28_cntk_text.txt to verify the effectiveness of the model.
To compile and run this model, CNTK C++ API is required. VSC provides a cloud computing interface to Amazon Web Services EC2. Choose the image Ubuntu with CNTK to compile and run this example.
..