Learning is enhanced when you get necessary practices along with real-world case studies. This Machine Learning Course by Google Developers is an absolutely free course comprising of 30+ exercises, 25 lessons, and lectures from top Google Researchers with interactive visualizations. In short, it features the best practices from Google Experts on key machine learning concepts.
The main page of this course contains the following four menus:
| Courses | Practica | Guides | Glossary |
|---|---|---|---|
| 1. Crash Course 2. Introduction to ML 3. Problem Facing 4. Data Prep 5. Clustering 6. Recommendations 7. Testing & Debugging 8. GANs | 1. Image Classification 2. Fairness in Perspective API These are some practices available. You have to complete them before stepping into the next course. | 1. People + AI Guidebook 2. Rules of ML 3. Text Classification 4. Good Data Analysis Above are some of the guidebooks available to enhance your learning. | It contains a detailed explanation of highlighted keywords or terms used in between the course or mentioned at the end of every course. |
So the above-mentioned categories can be explored simultaneously with the courses you will pursue. As none of the sub-courses contains certificates of completion, you can access them as per your preference of learning. Though I’ll share the prerequisites of each course. Now let’s get some brief introduction of the sub-courses individually:
1. Crash Course
What You will learn:
This will enable you to recognize the practical benefits of mastering machine learning and understand the philosophy behind ML.
Duration: Approx 20 hours.
Lessons:
- ML Concepts
- Introduction to ML
- Framing
- Descending into ML
- Reducing Loss
- First steps with TF
- Generalization
- training and Test Sets
- Validation Set
- Representation
- Feature Crosses
- Regularization: Simplicity
- Logistic Regression
- Classification
- Regularization: Sparsity
- Neural Networks
- Training Neural Nets
- Multi-Class Neural Nets
- Embeddings
- ML Engineering
- Production ML Systems
- Static vs. dynamic Training
- Static vs, Dynamic Inference
- Data Dependence
- Fairness
- ML Systems in the Real-World
- Cancer prediction
- Literature
- Guidelines
- Conclusion
- Next Step
Prerequisites:
You should have sufficient knowledge in topics such as Algebra, Trigonometry, Statistics, Calculus, Python programming, Bash terminal/ Cloud Console, etc.
2. Introduction to ML
What You will learn:
It’s a simple introductory course where you will learn about its features and types in detail.
Duration: 20 minutes.
Lessons:
- Introduction to Machine learning
- Machine Learning
- What is Machine Learning?
- Supervised Learning
- Conclusion
- Test Your Understanding (MCQ-based exercise)
Prerequisites:
You can complete the Machine Learning Crash Course though there are no such prerequisites for this course.
3. Problem Framing
What You will learn:
This course deals with the right skill to determine if a machine learning approach is good for a problem and simultaneously teaches you to outline an ML solution.
Duration: 45 minutes
Lessons:
- Overview
- Introduction
- Problem Framing
- ML Problem Framing
- Understand the Problem
- Framing an ML Problem
- Conclusion
- Implementing a Model
- Summary and Next Steps
Prerequisites:
Must be well-practiced with Machine Learning Fundamentals with companion TensorFlow exercises. For this, you can complete the above Machine Learning Crash Course and Introduction to ML.
4. Data Preparation & Feature Engineering in ML
What You will learn:
Data Preparation and feature engineering in ML help us to find patterns in data that one uses to make predictions about the new data point. Basically, this module covers two key steps – construction of the data set and transformation of the data correctly.
Duration: 3 hours
Lessons:
- Overview
- Introduction
- The Process
- Construct Your Dataset
- Introduction to Constructing Your Dataset
- Collecting Data
- Sampling and Splitting
- Transform Your Data
- Introduction to Transforming Data
- Transforming Numeric Data
- Transforming Categorial Data
- Check Your Understanding (MCQ-based exercise)
- Programming Excercise
- Programming exercise – Data modeling
- Conclusion
- Summary
Prerequisites:
Should have completed the Machine Learning Crash Course.
5. Clustering
What You will learn:
You will learn to define Clustering for ML applications, compare manual, and supervised similarity measures, use the k-means algorithms, and evaluation of the quality of your clustering result.
Duration: 4 hours
Lessons:
- Overview
- Clustering in Machine Learning
- What is Clustering?
- Clustering Algorithms
- Clustering workflow
- Overview
- Prepare Data
- Create a SImilarity Measure
- Run the Algorithm
- Interpret Results
- Summary
- K-means Advantages and Disadvantages
- Implement k-Means
- Clustering Programming Excercise
- Next Steps
Prerequisites:
One should have completed all three courses i.e. Introduction to ML Problem Framing, Machine Learning Crash Course, and Data Preparation & Feature Engineering. Additionally, one should have basic knowledge of Data Ditributions (Like Gaussian and Power Law Distributions) and some programming knowledge in Python.
6. Recommendations
What You will learn:
This course expands your knowledge of recommendation systems and explains different modules used in recommendation including matrix factorization and deep neural networks.
Duration: 4 hours
Lessons:
- Introduction
- Background
- Large-Scale Recommendation System
- Terminology
- Recommendation System Overview
- Check Your Understanding (MCQ Type Excercise)
- Candidate Generation
- Candidate Generation Overview
- Content-Based Filtering
- Collaborative Filtering and Matrix Factorization
- Recommendation Using Deep Neural Networks
- Retrieval, Scoring, and Re-ranking
- Softmax Excercise
- Conclusion
- Summary
Prerequisites:
Must have done Machine Learning Crash Course either in-person or self-study or have equivalent knowledge. Additionally, you must have familiarity with linear algebra and at least a little experience programming with TensorFlow and Pandas.
7. Testing & Debugging
What You will learn:
This course describes how to validate raw feature data and engineered featured data. Also, you will learn to debug and optimize a working ML Model. Furthermore, you will learn to monitor model matrics during development, launch, and production.
Duration: 4 Hours
Lessons:
- Introduction
- Debugging & Testing
- Overview (Excercise)
- Data Debugging
- Model Debugging (Excercise)
- Interpreting Loss Curves
- Model Metrics (Excercise)
- Model Optimization (Excercise)
- Testing in Production
- Overview
- Testing For Deployment
- Testing in Production (Excercise)
- Conclusion
- Implementation
- Summary and Next Steps
Prerequisites:
Should have completed ML Crash Course or have equivalent knowledge and Data Preparation and Feature Engineering Course. Basic programming knowledge in Python is also recommended.
8. GANs
What You will learn:
Generative adversarial networks (GANs) are generative models that create new data instances that resemble your training data. For instance, using GAN you can create images of human faces, which don’t even belong to any real being. So here you’ll learn to differentiate between generative and discriminative models and identify problems that GANs can solve. Furthermore, you’ll become familiar with the advantages and disadvantages of common GAN loss functions, identify possible solutions and use the TF GAN library to make a GAN.
Duration: Not mentioned
Lessons:
- Overview
- Introduction
- Generative Models
- GAN Anatomy
- Overview of GAN Structure
- Discriminator
- Generator
- GAN Training
- Loss Functions
- Check Your Understanding
- Real World GANs
- Common Problems
- GAN Variations
- Using TF-GAN
- TFGAN Colab Tutorials
- Conclusion
- Summary and Next Steps
Prerequisites:
Along with a little experience in programming with TensorFlow, one must have completed the Machine Learning Crash Course.
For more such free courses check: FREE COURSES BY INPROGRAMMER
