using gen_weight_File

[MichaelB0312]

Hi,

I'm missing what are you doing with all the weights in the csv file which gen_weight_File has created.

How do you use them afterward?

Thanks ;)

[YuanGongND]

Hi,

The weight file is used here:

psla/src/run.py

Lines 82 to 85 in 46a53b9

	if args.bal == True:
	print('balanced sampler is being used')
	samples_weight = np.loadtxt(args.data_train[:-5] + '_weight.csv', delimiter=',')
	sampler = WeightedRandomSampler(samples_weight, len(samples_weight), replacement=True)

Basically, in gen_weight_file, the weight of each sample is sum(1/class_frequency), note for audioset, each sample has multiple classes, that's why sum is needed. The idea is to give rare class higher weight.

Then in

psla/src/run.py

Line 85 in 46a53b9

sampler = WeightedRandomSampler(samples_weight, len(samples_weight), replacement=True)

, the weight is used to sampling in dataloading, i.e., a sound clip with rare class is more likely to be seen by the model. The algorithm can be found in the paper.

-Yuan