4
$\begingroup$

I am recently watching some tutorials for deep learning from Dr Andrew Ng on Youtube. Link is hereThe Youtube video

There is a concept of number of features in convolutional neural network in TensorFlow's tutorial.

I don't quite understand why the feature is 32 or 64 here in conv layer1 or layer2?

Then I came to the video, there is also the concept of Quadratic features. It is calculated as 3 million however. But how is it calculated?

Are the two features related in concept?

| improve this question | |
$\endgroup$
  • $\begingroup$ it's common to use powers of two for sizing filters or dense layers in neural networks $\endgroup$ – Vadim Smolyakov Aug 15 '17 at 18:26
1
$\begingroup$

Sorry for being so late in the response. I have just read your message.

The instructor is using $\frac{x^2}2$. So $\frac{2500\times 2500}2$, and this will get approx $3$ millions features.

BR

| improve this answer | |
$\endgroup$
0
$\begingroup$

What this means is:

1) 50x50 pixels image means 2500 unit pixels in your image. Then each unit is a feature, i.e.,

x1=pixel 1, x2 = pixel 2, .......x2500 = pixel 2500. A quadratic function means that you take all possible combinations of xi.xj, for example x1^2, x1.x2, x1.x3,....,x1.x2500, etc... this will give you a better outcome than considering just a linear function..., just get all the possible combinations and count them and you will get around n^2/2 = 2500x2500/2 = 3M

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.