PCA train loop when there is zero in the feature vector

[alanssoares]

@nickgillian can you help me? This is related to a development feature in the dev branch.

I don't know why yet, but the PCA stays in loop when there is zero in the feature vector, ie:

...
UseExternalRanges: 0
LabelledTrainingData:
12 0 0.540648 0.841175 1.01087 1.05752 1.04125 0.920698 0.570246 0.194463 0.372667 0.387392 0.370841 0.371541 0.316174 0.380672 0.356918 1.70058 1.70058
...

There are different classes, so if I remove the first column the pipeline works fine. I'm using an example of pipeline:

https://github.com/nickgillian/grt/blob/dev/examples/FeatureExtractionModules/PCAPipelineExample/PCAPipelineExample.cpp

Thanks

[nickgillian]

Hi @alanssoares,

Do you know if the issue is with the data file parsing, or specifically with PCA?

If you still want to resolve the issue then if you share an example training file I can take a look.

Thanks!