Question regarding find the max/min of a 3-D tensor

[tulyu96]

In one of the exercises, it asks use to find the min and max value of the tensor of shape(224,224,3). By calling tf.argmax()/tf.argmin() and specifying different 'axis' argument, I could get various return results, but I am stuck on how to use the return index to find the actual value. Anyone has any idea? It seems that we have to use some loop, but it would be very inefficient for data with large dimensions.

If 'axis=0', I am getting a 224 * 3, which essentially represent whether the corresponding value in the sub-matrix_1 or sub-matrix_2 is larger. If axis=1, I am getting 224 *3, but the meaning is that for each sub-matrix, the max/min index for each column. If axis=0, I am getting 224 *224, which represents that for each sub-matrix, the max/min for each row.

[mrdbourke]

Hey there!

Do you have a link to the exercise?

What does the code you're trying to run look like?

[tulyu96]

Sorry about the delayed reply. I was refering to the questions below. It's the exercise in part 1

Create a tensor with random values between 0 and 1 with shape [224, 224, 3].
Find the min and max values of the tensor you created in 6. (Specifically this question)

[mrdbourke]

Oh yes, you're right. That could be confusing.

I've updated it:

Find the min and max values of the tensor you created in 6 along the first axis.

Thank you for pointing this out.

[tulyu96]

Thank you for updating it.

Just finish both regression and classification modules. Really look forward to the rest of the course!