Share this page!

Harpreet Sahota πŸ₯‘ Profile picture
Twitter logo
Jan 6 β€’ 7 tweets β€’ 5 min read
The convolution operation q fundamental concepts in deep learning.

Most people know how to perform it.

But far fewer understand what it means.

These five πŸ–πŸ½ resources will help you grok one of the most important concepts in deep learning πŸ‘‡πŸ½

#deeplearning #machinelearning
1) Convolution vs cross correlation

You must understand the difference between convolution and cross-correlation in order to understand backpropagation in CNNs.

This will help.

glassboxmedicine.com/2019/07/26/con…
2) Convolutions in image processing

This 30ish minute lecture is from the MIT course on computational thinking.

Oh, and it’s taught by Grant from @3blue1brown

3) Convolution and cross correlation in neural networks

This one is from pyimagesearch and drives the point home with hands on coding

pyimagesearch.com/2021/05/14/con…
4) Convolutional Neural Networks

This one is from the Code Emporium guy which puts all the pieces together.

5) Andrew Ng on convolutions

Hear the legend himself explain why the convolution is the right operation for the job.

What else would you add to this?

1. Follow me @DataScienceHarp for more of these
2. RT the tweet below & share w/ your friends
3. Also follow: @nevrekaraishwa2 @SanthoshKumarS_ @GiftOjeabulu_ @Sumanth_077 @Saboo_Shubham_ @_jaydeepkarale @Pauline_Cx

β€’ β€’ β€’

Missing some Tweet in this thread? You can try to force a refresh
γ€€

Keep Current with Harpreet Sahota πŸ₯‘

Harpreet Sahota πŸ₯‘ Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @DataScienceHarp

Harpreet Sahota πŸ₯‘ Profile picture
Jan 6
Skip connections are one of the most important innovations in the history of deep learning.

You need to understand the problems it solves and how it helps us build deep networks.

Let me break it down for you in this thread πŸ‘‡πŸ½πŸ§΅

#deeplearning #machinelearning
Skip connections are a common feature in modern CNN architectures. They create an alternative path for the gradient to flow through, which can help the model learn faster.
In a neural network, the gradient measures how much a change in one part of the network affects the output. We use the gradient to update the network during training to recognize data patterns better.
Read 8 tweets
Harpreet Sahota πŸ₯‘ Profile picture
Jan 6
The hardest part of data science is keeping track of and measuring the right things.

And you do that with metrics.

Most data scientists don't know what makes a good metric.

In this thread, I'll teach you what makes a good metricπŸ“ˆ
A good metric has four qualities; it's...
1) Comparative
2) Understandable
3) A ratio or rate
4) Changes in behaviour
1) A good metric is comparative

Comparing a metric to other periods, groups of users or competitors helps you understand which way things are moving.
Read 9 tweets
Harpreet Sahota πŸ₯‘ Profile picture
Jan 5
Accuracy, precision, recall, and F1-score are a few ways you can measure the performance of your classification model.

Most people get them confused.

Use this thread as a cheat sheet and remind yourself what they meanπŸ‘‡πŸ½πŸ§΅
Accuracy: This is the number of correct predictions made divided by the total number of predictions.
True/false positives/negatives: For binary classification, these refer to correctly/incorrectly predicted positive/negative class samples.
Read 10 tweets
Harpreet Sahota πŸ₯‘ Profile picture
Jan 5
Here's how I would study deep learning if I had to do it all over again.

#deeplearning #machinelearning
πŸ‘‡πŸ½ 🧡
1) Skip the math

I’d ignore the math when first starting out.

Looking at equations will demotivate you.

Instead, look for applications of deep learning.

Clone the
@ultralytics
Yolov5 repo and run the usage on the command line.

See the magic happen.

Get inspired.
2) Go through @AndrewGlassner’s DL crash course.

It’s 3.5 hours long but will give you an intuition for how it all works under the hood.

Great return on time investment.

Read 12 tweets
Harpreet Sahota πŸ₯‘ Profile picture
Jan 4
Back propagation is the secret sauce for training neural networks.

You need to understand how it works.

I break it down for you in this thread πŸ‘‡πŸ½πŸ§΅

#deeplearning
Here's the lowdown: the output of a neural network is calculated with the πŸ‹πŸ½ weights πŸ‹πŸ½ of the edges that connect the nodes in the network.

So, you gotta find the optimal values of weights to minimize the final error on training data examples.
1️⃣ We start by assigning random values to all weights in the network

2️⃣ Then, for every input sample, we perform a feedforward operation to calculate the final output and the prediction error.
Read 10 tweets
Harpreet Sahota πŸ₯‘ Profile picture
Jan 4
You shouldn't evaluate the performance of a deep learning model solely on accuracy.

You're missing the whole picture if that's all you're looking at.

There are 3 other factors you should consider when evaluating the performance of your model πŸ‘‡πŸ½

#deeplearning #ai
Flops (floating point operations)

This measures of the amount of computation required to train and run a model.

More complex models often require more flops, which can make them more expensive to use.
Parameters

The number of parameters in a model can also impact its performance.

Models with more parameters may fit the training data better, but they may also be more prone to overfitting and may not generalize well to new data.
Read 6 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

Send Email!

:(