🎉 🎉 RAPIDMINER 9.10 IS OUT!!! 🎉🎉

Download the latest version helping analytics teams accelerate time-to-value for streaming and IIOT use cases.


Deep learning extension : Error with Conv layer(s)

lionelderkrikorlionelderkrikor Moderator, RapidMiner Certified Analyst, Member Posts: 1,190   Unicorn
Dear all,

I try to perform "time series classification" using the Deep learning extension and I builded a neural network using 
convolutional layer(s).
To describe the project : 
 - I have a collection of around 1200 TimeSeries each including 2 signals of several seconds. Each Time Serie is associated to a label.
There are 6 labels and the collection of datasets is balanced, thus there are around 200 TimeSeries per label.
 - I used the TimeSeries to Tensor and the Deep Learning (Tensor) operators
 - Inside the Deep Learning(Tensor) operator I'm using convolutional layer(s).

When I execute the process, RapidMiner is raising the following error : 

  • Message: Invalid input for Convolution layer (layer name="ConvolutionalLayer"): Expected CNN input, got InputTypeRecurrent(3,timeSeriesLength=278,format=NCW)

Is there an error in my process ? How to fix this error. ?

In attached file(s), you can find  : 
 - the process
 - The folder of the collection of files (Time Series)
in order to reproduce the error.

Thank you for your answer(s),


Sign In or Register to comment.