Hello fastai people,
I’m starting my journey into deep learning, and at this point, I’m kinda confused.
What do we mean by “control parameters” and “trainable parameters” and “learning parameters” in neural networks?
Thank you all.
Hi ! The parameters you learn in a neural net are the weights that are applied to make linear combinations of the neutrons. Basically, you initialise those parameters randomly, compute your loss (difference between expected results and targets) and computing the gradients by backpropagation you are able to iteratively adjust those weights.
Best,
Charles
Thank you.
I’m actually getting into speech feature extraction with MLP, and there seem to be many kind of parameters. I’m mainly confused by “control parameters” in audio processing.
Much appreciated.