From the course: Neural Networks and Convolutional Neural Networks Essential Training

Unlock the full course today

Join today to access over 22,600 courses taught by industry experts or purchase this course individually.

Creating and compiling the model

Creating and compiling the model

From the course: Neural Networks and Convolutional Neural Networks Essential Training

Start my 1-month free trial

Creating and compiling the model

“

- [Instructor] As before, it's probably helpful for us to have a view of what the model we're trying to create looks like. We have our original image which is 28 by 28 with one channel so it's a grayscale image. We then do the convolution operation with the five by five kernel and then there are 32 filters. We then do a pooling so our image drops from 24 by 24 to 12 by 12. We then do another convolution operation with the five by five kernel and this time with 64 filters. We do another pooling. Again, we see a reduction in our image by half. That's from eight by eight to four by four. Finally, there's a flattening so there's a fully connected network. Then we've got the output. We've got all of the 1,024 notes terminating in the ten outputs. The ten outputs correspond to the ten digits, zero to nine. Let's head over to our notebook and create this in Caris. The first thing we want is to create a sequential model so let's call our model 'cnn'. We then want to add our convolution layer.…

Contents