features vs columns in conv1d input shape

21 views Asked by At

I am working on building Conv1d for CNN multiclass classifiaction. I had readings from different sensors like my csv file consists of many rows and each row is readings from those sensors for certain time stamp. I have two questions: 1- when I used autoencoder to generate features and make dimensionality reduction I used Conv1d layer, but the input shape was (1,100) where 100 is the number of sensors, was that correct or it should be (100,1) and if not correct does that mean the generated features are corrupted? 2- when using COnv1d for multiclass classification with number, my y_label consists of integer numbers in this case do I need to process it or use it directly as target? thanks

extensive search on the internet

0

There are 0 answers