Class notes
Week 4 Review
- Institution
- Arizona State University
Week 4 lecture notes
[Show more]
Seller
Follow
Content preview
Week 4 ReviewConvolutional neural networks are better suited for object recognition in digital
images than the traditional programming
We have an image for an input, followed by convolutional layers.
Each convolutional layer is followed by a pooling layer.
RGB Images - The image is a grid of pixels and can be described using three
matrices/ channels (red, green, blue)
Convolutions
In the convolutional layer, the CNN learns a useful kernel for each feature
map
involves procedurally running kernels (masks) over regions of pixels in
a digital image
Convolutional neural networks are much more efficient by learning
kernels that can be universally applied to the entire image instead of
weights for every individual pixel and channel.
.The convolution is the sum of the multiplication of the weights by the pixel
values.
we can interpret the convolution as a neuron and the weights in the
kernel as the weights in the neuron
Use backpropagation for automatically train the weights
Activation function - ReLU is usually used
Convolution + ReLU = one activation map
Kernels
Where do kernels(masks) come from?
In the past, they were human - engineered
Week 4 Review 1
, With a convolutional neural network, we can use machine learning to
find useful kernels instead.
How does a kernel transform an image?
By linearly combining pixel regions
If we look at the kernel and pixels as a vector, the new pixel value
becomes their dot product.
Common kernels
Gaussian Blur - Gives nearby pixels a larger weight than ones that are
further away
Sharpen filter - The sum of pixels in the mask is one, so no need to
normalize
Edge detection- Use two convolutional kernels together
Stride length determines the step size of the kernel across the input
(horizontal and vertical)
Padding
Pooling
Go through the CNN and apply dimensionality reduction
Max pooling
Take the max value from a local neighborhood of the activation map
Only focus on the strongest activations
Week 4 Review 2
The benefits of buying summaries with Stuvia:
Guaranteed quality through customer reviews
Stuvia customers have reviewed more than 700,000 summaries. This how you know that you are buying the best documents.
Quick and easy check-out
You can quickly pay through credit card or Stuvia-credit for the summaries. There is no membership needed.
Focus on what matters
Your fellow students write the study notes themselves, which is why the documents are always reliable and up-to-date. This ensures you quickly get to the core!
Frequently asked questions
What do I get when I buy this document?
You get a PDF, available immediately after your purchase. The purchased document is accessible anytime, anywhere and indefinitely through your profile.
Satisfaction guarantee: how does it work?
Our satisfaction guarantee ensures that you always find a study document that suits you well. You fill out a form, and our customer service team takes care of the rest.
Who am I buying these notes from?
Stuvia is a marketplace, so you are not buying this document from us, but from seller 4point0. Stuvia facilitates payment to the seller.
Will I be stuck with a subscription?
No, you only buy these notes for $2.99. You're not tied to anything after your purchase.
Can Stuvia be trusted?
4.6 stars on Google & Trustpilot (+1000 reviews)
50064 documents were sold in the last 30 days
Founded in 2010, the go-to place to buy study notes for 14 years now