Machine Learning
Machine Learning
Your Progress
Learning goals
Explore ML techniques and innovative ML approaches like deep learning and neural networks as well.
Suggested prerequisites
A curiosity of machine learning tools and technologies.
Intro to ML
Just getting started with Machine Learning? Start building a foundational knowledge of ML concepts here.
What is Machine Learning?
Got lots of data? Machine learning can help! In this episode of Cloud AI Adventures, Yufeng Guo explains machine learning from the ground up. (YouTube)
ESTIMATED TIME: UNDER 10 MINSThe 7 steps of Machine Learning — Part 1
How can we tell if a drink is beer or wine? Machine learning, of course! In this episode of Cloud AI Adventures, Yufeng walks through the 7 steps involved in applied machine learning. (YouTube)
ESTIMATED TIME: 10-20 MINSThe 7 steps of Machine Learning — Part 2
How can we tell if a drink is beer or wine? Machine learning, of course! In this companion article to Cloud AI Adventures, Yufeng walks through the 7 steps involved in applied machine learning. (Medium)
ESTIMATED TIME: UNDER 10 MINSThe Teachable Machine
Teachable Machine allows you to create machine learning models — with no coding required! Learn how to use Teachable Machine to train a computer to recognize your own images, sounds, and poses. (Google)
Some Background
Kaggle is the world's largest data science community with powerful tools and resources to help you achieve your data science goals.
What's Kaggle?
Welcome to Kaggle, a community that helps you achieve your data science goals. You'll find datasets, tutorials, and competitions to help you sharpen your machine learning skills. (YouTube)
ESTIMATED TIME: UNDER 10 MINSGetting started on Kaggle: a quick tour
Not sure how to get started on Kaggle? Check out this short overview of Kaggle's main features, ending in a description of Learn courses, a friendly introduction to data science. (YouTube)
ESTIMATED TIME: UNDER 10 MINSCourses on Kaggle (Python)
Python is one of the most frequently used languages in data science. If you're interested in delving deeper into data science, but are not yet comfortable with Python, check out this course. (Kaggle)
Pretrained Models
Explore several Machine Learning APIs—sets of tools and protocols used for building software and models—which help machine learning developers communicate with each other and share knowledge across various platforms.
Machine Learning APIs by example
Did you know you can make use of Google's machine learning expertise to power your own applications? Check out this video to learn more about how you can use APIs to build and train your models. (YouTube)
ESTIMATED TIME: 30-60 MINSExplore the galaxy of images with Cloud Vision API
Want to better understand the content in your images? Learn more about the Cloud Vision API, which can help you classify images into thousands of categories. (Google Cloud)
ESTIMATED TIME: 20-30 MINSMachine Learning APIs
With a host of APIs, Google Cloud has a tool for just about any machine learning job! Check out this quest to get hands-on practice with detecting labels, classifying text, and more. (Qwiklabs)
ESTIMATED TIME: OVER 2 HOURSUsing Machine Learning to enhance your apps
How does your phone recognize voice or categorize photos? Check out this demo on making requests to the Vision and Speech APIs to learn more about the models that power machine learning capabilities. (YouTube)
ESTIMATED TIME: 30-60 MINS
AutoML
Automated machine learning (AutoML) enables even non-experts to apply machine learning models and techniques to real-world problems.
AutoML Vision — Part 1
In this episode of AI Adventures, Yufeng Guo uses AutoML Vision to build and employ a machine learning model that recognizes different types of….chairs! (YouTube)
ESTIMATED TIME: UNDER 10 MINSAutoML Vision — Part 2
After data preparation, Yufeng Guo shows us how to use AutoML to train a machine learning model — using no programming at all! (YouTube)
ESTIMATED TIME: UNDER 10 MINSAutoML tables
In this episode of AI Adventures, Yufeng introduces AutoML Tables, a tool to automatically build and deploy state-of-the-art machine learning models on structured data. (YouTube)
ESTIMATED TIME: UNDER 10 MINSAutoML vision in action on GCP
Want to see Google AutoML in action? Check out this real-world case study of ramen noodle bowls and learn about the possibilities of using automated machine learning models in data science. (Google)
ESTIMATED TIME: 20-30 MINSClassify images of clouds in the Cloud with AutoML vision
Let's classify images of clouds...using Cloud! By using Cloud with AutoML vision, you can train image recognition models to generate strong predictions via an easy-to-use API. (Qwiklabs)
How is ML Done
How does machine learning work in practice?
A visual intro to ML with decision trees
Don’t miss this enjoyable blog post which brings ML terms and concepts to life through data visualization. Follow along as interactive images show how decision trees sort data and make predictions. (R2D3)
ESTIMATED TIME: 30-60 MINSWrite a decision tree classifier from scratch
Curious about how to write a supervised learning algorithm from scratch? Josh Gordon walks you through how to write a Decision Tree classifier using Python while introducing ML concepts along the way. (YouTube)
ESTIMATED TIME: UNDER 10 MINSMachine Learning tutorial on Kaggle
It's time to build your first machine learning model! Check out the tutorial to learn about core ideas in machine learning, create a model, and measure its performance. (Kaggle)
Beginner Projects
Explore Kaggle competition challenges and solutions for a deeper understanding of machine learning and data science.
Data Science projects for beginners
New to data science and unsure where to start? Check out this video to learn about 3 projects that are a good starting point for beginners. (YouTube)
ESTIMATED TIME: UNDER 10 MINSThe Housing Competition — regression techniques
Armed with Python & R, you’re ready to put theory into practice and flex your data analysis skills in a competition environment. Learn about regression techniques while predicting the sales price of houses. (Kaggle)
The Housing Competition — comprehensive data exploration
Curious about initial dataset exploration? Check out this solution to the Housing Regression Analysis competition to learn how to better understand data science problems and test your methods. (Kaggle)
The Housing Competition — fun with real estate data
You may need to try more than one model to get the most accurate prediction! Check out this solution, which tries three models to help get the most accurate prediction for the real estate market. (Kaggle)
The Titanic Competition
It's time to put on your prediction hat! In this competition, you'll analyze who was most likely to survive the Titanic shipwreck using the binary classification method. (Kaggle)
The Titanic Competition — exploratory data analysis
This exploratory data analysis gets high quality predictions on the Titanic dataset. Can you beat the benchmark by adding your own analysis? Note that Python is required. (Kaggle)
ESTIMATED TIME: OVER 2 HOURSThe Titanic Competition — data science solutions
What's the typical workflow like for competitions? This exercise takes you through each step of that workflow using the Titanic Data Science Solutions as an example. (Kaggle)
ESTIMATED TIME: OVER 2 HOURSThe Titanic Competition — walk-through for data analysis
Learn how Kaggle’s top kernel contributor, Heads or Tails, approaches the Titanic competition. You'll also brush up on topics such as data cleaning and feature engineering along the way. (Kaggle)
ESTIMATED TIME: OVER 2 HOURSThe Digit Recognition Competition
Machine learning can be used to recognize images, sounds, and even hand-written text! Learn about digit recognition and computer vision fundamentals with the famous MNIST data. (Kaggle)
TensorFlow and deep learning, without a PhD
Learn to build and train a neural network to recognize handwritten digits in this codelab. As you code in Python and TensorFlow, you'll discover tips and tricks to add to your deep learning toolkit. (Codelabs)
ESTIMATED TIME: OVER 2 HOURSIntro to convolutional neural networks with Keras
Learn to cook up a 5-layer convoluted neural network for digit recognition using this easy-to-follow-along recipe. Note that you'll need to have Python installed to get started. (Kaggle)
ESTIMATED TIME: OVER 2 HOURS
Advanced Topics
With more machine learning knowledge under your belt, now's your chance to dive deeper into more complex topics.
Introduction to Machine Learning problem framing
Interested in how to frame a machine learning problem and propose a solution for it? This course helps you get into that mindset! This course does not cover how to implement machine learning models. (Google Developers)
Machine Learning Crash Course
Are you an aspiring machine learning scientist? Check out this self-study guide that includes lectures, case studies, and hands-on exercises to help you get up to speed on machine learning concepts. (Google Developers)
Data preparation and feature engineering in ML
Machine learning helps us find patterns in data that we can then use to make predictions about new data points. Learn how to prepare your data to get those predictions right. (Google Developers)
Data cleaning
Real-world data can be messy! Learn best practices for data cleaning, such as how to handle missing values or decode characters, in this course. (Kaggle)
Feature engineering
Curious about how to improve the accuracy of your machine learning models? Check out this course on feature engineering and learn how you can train your data to predict the most accurate outcomes. (Kaggle)
Clustering
Clustering—grouping examples into categories—is a first step used in machine learning to understand a dataset. Learn about what clustering is and how it can be used to efficiently group data. (Google Developers)
Recommendation systems
How is a product or app able to provide you with the best possible recommendation? Learn about how models, such as matrix factorization or deep neural networks, help to create recommendation systems. (Google Developers)
Testing and debugging
Machine learning systems need to be tested and debugged, just like software! This course walks you through how you can debug your model and also monitor your pipeline in production. (Google Developers)
Generative Adversarial Networks
You can use Generative adversarial networks (GANs) to generate new data instances, such as images. Learn more about GANs and how it can be used to resemble training data. (Google Developers)
A brief history of deep learning
Explore how deep learning progressed from the perceptron to neural networks. Along the way, you’ll learn about great thinkers in the field and get helpful references to academic sources. (External blog)
ESTIMATED TIME: 1-2 HOURSThe AI Hub
Check out AI Hub, a collection of resources for developers building AI systems. Learn about solutions for creating and managing models while also checking out in-depth tutorials on ML pipelines. (Google Cloud)
Peter Norvig's Statistical NLP
Code along with Google's legendary research director, Peter Norvig, who authored this easy-to-follow Jupyter Notebook introduction to natural language processing in Python. (Peter Norvig)
How to solve (almost) any NLP problem
Score some tips on approaching natural language processing problems in Kaggle competitions using this kernel, which also features info on deep neural networks and ensembling. (Kaggle)
Understanding long short-term memory (LSTM) networks
Increase your understanding of Long Short-Term Memory (LSTM) networks by reading this post by Chris Olah. The clear writing and visualizations help explain recent advances in concepts such as natural language processing. (External blog)
ESTIMATED TIME: 30-60 MINSApplied introduction to LSTMs for text generation
Looking for an interactive way to learn how to implement LSTMs in Python? Fork the kernel that accompanies this article by Kaggle data scientist, Megan Risdal, and you'll soon be able to generate text! (freeCodeCamp)
ESTIMATED TIME: OVER 2 HOURSThe unreasonable effectiveness of recurrent neural networks
Witness the power of a Recurrent Neural Network (RNN)! Follow along with Karpathy, head of AI at Tesla, as he demonstrates how LSTMs can be used to generate writing similar to human speech. (External blog)
ESTIMATED TIME: OVER 2 HOURSSequence Modeling (Chapter 10)
Looking for an explanation of recurrent neural networks and natural language processing? Check out the Sequential Modeling chapter to get up to speed! Note that knowledge of Python is required. (DeepLearningBook.org)
ESTIMATED TIME: OVER 2 HOURS