Intelligent Typing Assistance with Deep Neural Networks

This feature suggests complete commands on the Presentation window of the green screen, based on the first 3-4 characters typed by the user using the Deep Neural Network.

The Intelli service Enables to start intelli service in Admin Console, enable AUTO-START for intelli service in HOD_Distribution/private/NSMprop.

Note: Make sure intelli service port is available. Once the HOD service is started, you can see intelli service status in Adminconsole -> Services

Train: Upon clicking the train button, it will fetch the data and tokenize each command and convert it into the tensor form. Once training starts, current Epoch, current accuracy, current loss, and progress can be seen in the status box. While training, train and Fetch button will be disabled.

Fetch: Upon clicking the fetch button, Latest model.json and weights.bin file will be fetched in the WAR file.

Back-end port: Port number on which Node backend server is running

Front end URL: Front end where HOD Next Client is running.

Batch size: The number of training samples processed together in one forward/backward pass Larger batches speed up training but use more memory.

hidden units: The number of neurons/memory cells in the LSTM layer. More units increase model capacity but risk overfitting.

Final accuracy: The percentage of correct predictions made by the model on a dataset.

Save: This button will update the properties in the file.