yaozhong / urnano Goto Github PK

I want to train the model, but was informed

 train_urnet.py: error: the following arguments are required: -cf/--cacheFile

From the script I can see that this arg. is required, but the default value is "".

In the training part of your github page, I can see that

## If no data cache is given, provide the signal and labels as used in Chiron.

As I understand, -cf will not be necessary if the signal and label files are provided, but the program cannot run, saying that -cf is essential.

My question is: if I want to train a model using the E. coli dataset, where can I find the cacheFile, or how can I generate these files from the signal and label files?

Thanks~

In your paper, you use median shift and median abs deviation to normalize signal. However, in your code ,you use mean value and std value normalize signal. I try both methods and only mean operation has the same waveform with your paper. So actually don't you use normalize method proposed in your paper?

Thank you.

In section 2.2, you use D={X,Y} to train cnn model. I have a question, raw signal's length may longer than label. For example, signal data 12,30,10 has a correspond label A. So how do you make label has the same length as signal data when training network? Thank you.