DL Python Network Creator

I am new to this. Just want to read input data containing text for binary classifications. Looking 1 example from the server called Sentiment Analysis on IMDB movie reviews . I have vector document with 1000 features and 1 label and want to read that data in DL Python Network Creator or any other options of doing it.
Thanks for the suggestion.
Stripathi

Hello Stripathi,

the purpose of the DL Python Network Creator node is to create deep learning networks but you can now also use our Keras nodes to compose your networks in a visual way.
For data loading you can use all the other nodes for data loading in KNIME (in the end you will typically end up with the File Reader node :wink: ).

Once you have your network and your data, you can combine them for training using the Keras Network Learner node, or if you load an already trained network you can apply it to your data using the DL Network Executor node.

Best,

nemad

Thanks. I am still having tough time to take example from server and run on my own data. I have transferred text data into 2557 features (columns) with 1 label (containing 1 and 0). When I run the model the following message are displayed selected column provides more features (2556) than neurons 80?. How to correct this?
see below the message.

Hi @stripathi,

I don’t think that your task is the same as the task done by the sentiment analysis workflow.
The model in this workflow takes sentences encoded as a integer sequence and predicts the sentiment using a LSTM based deep learning network.

Your data looks like you have something like a document vector, or more specifically some features extracted for each sentence.
You can use deep learning for such data but I believe you might actually be better off with a random forest or gradient boosted trees based model.
If you really want to use deep learning than this workflow https://www.knime.com/nodeguide/analytics/deep-learning/keras/wide-and-deep-learning-on-census-dataset is better suited for your problem.
However, I have to warn you that it is more complex than the sentiment analysis workflow.

Best,

Adrian