I am trying my first Edge Impulse ML task. I’m looking to distinguish left and right. I have only generated about 40 samples/utterance of left and right. I’ve added noise and unknown as per the tutorial video. All seems fine until I get to the NN. It only shows two outputs. Am I doing something wrong? The data is very seriously unbalanced but that is not my concern now. I’m simply trying to learn the mechanics of using Edge Impulse.
what do you mean it only shows 2 outputs?
Hi @StAugustine,
Did you create the NN classifier when you had only 2 classes in your dataset by chance? If you added new classes later on, just remove/add the NN classifier again and it should work.
Aurelien
he said “looking to distinguish left and right” so it seems to be a binary classifier to me. Or there is also the possibility of neither left nor right?
There are also the noise and unknown classes in the dataset.
Hi Aurel:
Yes, I had created the NN classifier when I only had the left and right categories. I will do as you suggest and try it. Much thanks.
Bob
Hi Edge7:
I had added the noise and unknown categories. I will revisit the NN definition by deleting and adding back as Aurelien suggested.
Bob
Aurel, your suggestion was right on. Thanks.
Bob