[MUSIC] >> We've now seen how cool
deep learning can be. We cannot apply it in
a wide range of areas and yet we need high accuracy by learning
really detailed features of our data, and how your networks can support that. We've seen how it can be applied for various tasks in image
analysis in computer vision. Let's now revisit the block diagram
that we saw summarizing regression and classification and other machinery tasks. How can be applied here for
computer vision for deep learning? So in particular, let's talk about
deep features for classifying images. So deep features for image classification. The input here are pairs of
images with their labels. So, the labels that we've looked at were
things like, whether there is a cat, the dog, a house or
some other object in the image. And now, we feed that through
the feature extractor. In this case, we're using a deep
learning model as a feature extractor. So the output here,
what we call the Deep features for this particular image for every image. And now we feed in this images, representative features
through a machinery model. Where we use a simple classifier
like logistic regression, in here. Say logistic regression as an example. And the output is our predicted labels. Predicted labels. And so we're going to feed in
our predicted labels y hat and the true labels,
y into our measure of quality. So y and y hat, and the measure
of quality depends on your task. For this task we use
classification accuracy. And so the parameters w hat are really the parameters of
the weights of the logistic regressive. So these are our weights on features. And what the machinery algorithm is gonna
do, is take the classification accuracy, try to make a little better by changing
those weights w hat and updating them. We've now seen how deep learning
can give you really cool and exciting results for
a various tasks in computer vision. And we saw those applied
both two classification and to retrieving of new images using raw
neural networks as well as deep features. And our notebooks that we
explored showed us that's really easy to build such
deep learning models and actually apply to really cool machine
learning tasks in computer vision. Both in classification and image retrieval
which allows me to find exactly the kind of tools that I'm really excited about. And with this you'll be able to build
a really exciting intelligent application that uses one of the most sought after
techniques today in machine learning, deep learning. [MUSIC]