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 👇🏽🧵

#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.

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 metric📈 A good metric has four qualities; it's...
1) Comparative
2) Understandable
3) A ratio or rate
4) Changes in behaviour

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 🖐🏽 resources will help you grok one of the most important concepts in deep learning 👇🏽

#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…

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 mean👇🏽🧵 Accuracy: This is the number of correct predictions made divided by the total number of predictions.

Here's how I would study deep learning if I had to do it all over again.

#deeplearning #machinelearning
👇🏽 🧵 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.

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 👇🏽🧵

#deeplearning Here's the lowdown: the output of a neural network is calculated with the 🏋🏽 weights 🏋🏽 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.

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 👇🏽

#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.

🤯 Say goodbye to lifeless textbooks and hello to an exciting way to learn statistics! 💪

I have a masters degree in statistics, but these 11 books taught me more about how statistics in the real world than any course I've taken.

Have you read any of them?👇🏽🧵

#66DaysOfData 📚🧠Who says learning statistics has to be boring?!

🤓 The Manga Guide to Statistics makes it fun and easy to learn all the basic concepts, with entertaining examples and applications.

Get it here:nostarch.com/mg_statistics.…

You don't need a bootcamp to get started in machine learning.

All you need are the right resources, discipline, and time.

Here are 6 of my favourite FREE resources to get you started. 1/ edX's Machine Learning with Python: A Practical Introduction

This course will give you all the tools you need to get started with supervised and unsupervised learning.

Time commitment: 4 hours a week and you're done in 2 weeks.

edx.org/course/machine…

The curse of dimensionality is a major roadblock for machine learning practitioners.

But most don't fully understand it.

Don't be left in the dark - join me in this thread as I clarify and demystify this concept 👇🏽🧵 The Curse of Dimensionality (let's just call it "The Curse") refers to problems that occur when you try to use statistical methods in high-dimensional space.

If I had to learn data analysis with Python in 2023, here's how I would do it.

Get ready to transform your data analysis skills with these highly recommended resources.

#66DaysOfData 👇🏽🧵 1/ Cognitive Class: Data Analysis with Python

You'll learn how to prepare data for analysis, perform statistical analyses, create data visualizations, predict trends from data, and more!

Spend 30 minutes a day, everyday, and you'll be done in 3 weeks.

cognitiveclass.ai/courses/data-a…

Feature selection is a crucial part of building a good machine learning model.

But most data scientists don't think before they select features.

The fact is: feature selection in machine learning is not always necessary.

Here are 5 situation when you don't need it 👇🏽🧵 1. You have a small dataset that doesn't have many features.

If the data you're using is small and doesn't have many features, you don't need to do feature selection.

Machine learning and Python go hand in hand.

Ready to take the first step towards a rewarding career in machine learning?

These 4 resources will help you learn Python and get started 👇🏽🧵

#100DaysOfCode #66DaysOfData #DeepLearning 1/ Python Principles

I've never seen anything like this course.

This is a text based course with an interactive coding environment that will teach you all the basics of Python.

There's lots of challenges and exercises, too.

This should take 2 weeks.

pythonprinciples.com/lessons/

The number one cause of machine learning model failure is data set drift.

Yet most data scientists and machine learning practitioners don't know why their data sets are drifting.

Here are 6 of the most common reasons for data set drift in machine learning 👇🏽🧵 What is dataset drift? It's when the statistical properties of a dataset change over time, which can negatively impact the performance of a machine learning model.

Calculus is the foundation of deep learning.

If you don't understand it, you won't get far in your journey.

Here are 3 resources that will quickly get you up to speed on the topic.

👇🏽🧵 1/ The Essence of Calculus by 3B1B

This will take you three hours to get through and you'll develop an intuition for the subject and a lay of the land.

youtube.com/playlist?list=…

Linear algebra is the backbone and foundation of machine learning (especially deep learning).

If you don't know this stuff, you won't get taken seriously in the field.

You don't need to spend hours searching for the best resources, I have 4 of them here for you 👇🏽🧵 1/ Essence of Linear Algebra by 3B1B

Grant has a way of explaining topics in an easy to understand way. The animations make this intuitive and easy to grasp.

Spend 3 hours to go through the playlist and you'll have a good lay of the land

youtube.com/playlist?list=…

Want to get into machine learning in 2023?

Then you need to know statistics and probability.

If you don't know this stuff, you're just a data charlatan.

Here are four resources that will get you up to speed:

👇🏽🧵 First up is Khan Academy Statistics and Probability

Spend 7 hours a week on this you'll be done in 2 weeks and have a solid overview of the basics.

khanacademy.org/math/statistic…

import os or import pathlib...

That is the question.

Let's talk about what to use and why

👇🏽🧵 Ultimately using the os library or the pathlib library depends on your personal preference and the requirements of your project.

Both libraries provide similar functionality for working with file paths and directories, but they have some key differences.

Python is the most important language to learn for deep learning.

I learned it all wrong when I first started.

Here’s the path I’d take if I was to do it all over again…

🧵 1) Python principles

This is where I would go to first. It teaches the absolute fundamentals and basic principles of Python.

Lessons teaching it from the ground up + challenges that will help prep you for more difficult things.

Register for free: pythonprinciples.com

If I had to learn to code again, here’s how I would do it

🧵 /s 1) Skip all the languages and go straight to the source

Pick up a book on logic. Study it.

Then grok Boolean algebra.

In June of 2021, I read two books that lit a fire inside of me and changed the direction of my career forever 📚

And since today is Book Lovers Day, I’d like to share those books with you.

👇🏽 📖 One book made a complex subject – one that I was afraid of – come alive in a way that was accessible and understandable

It taught me the history of the ideas in deep learning and introduced me to the names of the people behind them🧍🏻‍♂️