Deep Learning

Notes from Coursera Specializations

In this section of the website, you’ll find my detailed notes and insights from the Coursera specializations I completed:

Coursera Machine Learning Specialization: https://www.coursera.org/specializations/machine-learning
Coursera Deep Learning Specialization: https://www.coursera.org/specializations/deep-learning

These specializations have greatly helped me learn and understand more about these exciting fields.

Machine Learning Specialization vs. Deep Learning Specialization on Coursera

Specialization	Focus	Duration	Instructor	Prerequisites	Audience
Machine Learning Specialization	Fundamentals of machine learning	3 courses	Andrew Ng	None	Beginners
Deep Learning Specialization	Deep learning techniques	5 courses	Andrew Ng	Familiarity with machine learning	Intermediate to advanced learners

Common points between the Machine Learning Specialization and the Deep Learning Specialization on Coursera:

Both specializations are taught by Andrew Ng, a leading expert in machine learning.
Both specializations are self-paced and can be completed at your own pace.
Both specializations are offered by Coursera, a leading online learning platform.
Both specializations cover the use of Python for machine learning (especially Numpy and Tensorflow)

Machine Learning Specialization groups 3 lessons:

Deep Learning Specialization groups 5 lessons:

There is an overlap between the two specializations, or more specifically, the Machine Learning specialization provides more detailed explanations in certain sections compared to the Deep Learning specialization. This is the case for the following points:

Concept	Machine Learning	Deep Learning
Gradient Descent	c1-supervised-ml/week1	c1-neural-networks-and-deep-learning/week2
Vectorization	c1-supervised-ml/week2	c1-neural-networks-and-deep-learning/week2
Forward Propagation	c2-advanced-learning-algorithms/week1	c1-neural-networks-and-deep-learning/week3
Computation Graph	c2-advanced-learning-algorithms/week2	c1-neural-networks-and-deep-learning/week2
Back Propagation Intuition	c2-advanced-learning-algorithms/week2	c1-neural-networks-and-deep-learning/week3
Activation functions	c2-advanced-learning-algorithms/week2/	c1-neural-networks-and-deep-learning/week3
Why Non-Linear Activation	c2-advanced-learning-algorithms/week2/	c1-neural-networks-and-deep-learning/week3

Andrew Ng

Andrew Ng is a pioneering figure in machine learning. His career includes co-founding Google Brain, where he advanced AI research, and Stanford University’s AI Lab.

He co-founded Coursera, an online platform that provides accessible education in AI and machine learning to a global audience.

Andrew Ng is a clear, concise, and engaging teacher who uses a variety of teaching methods to help students learn about machine learning. His courses are well-organized and easy to follow, and he encourages students to ask questions and participate in discussions.

Mathematic level

Andrew Ng’s Machine Learning Specialization on Coursera assumes a foundational understanding of basic calculus and linear algebra. It expects students to be familiar with concepts like derivatives, integrals, matrices, and vectors.

Here are some of the specific math concepts that are covered in the Machine Learning Specialization:

Calculus: derivatives, integrals, limits
Linear algebra: matrices, vectors
Probability and statistics: probability distributions

If you are not comfortable with these concepts, you may find it difficult to follow some lectures and complete the assignments. However, the specialization does provide a good introduction to these concepts, so you should be able to learn them as you go along.

Python level

A beginner to intermediate Python knowledge is required. You should be comfortable with the basics of Python, such as variables, data types, functions, and loops.

The assignments are implemented using Python Jupyter notebooks, which provide a standardized environment. However, it’s worth mentioning that this choice of environment can sometimes hide abstractions.

In terms of coding, all assignments provide a lot of help and hints. The code is pre-written, and you only fill in a few lines of code in each assignment. This approach allows learners to focus on specific implementation aspects while leveraging existing code foundations. But that why sometime, you leave the assignment with the feeling of something too simple.

Overall, the assignments are great, thank you Coursera and Andrew for the work of popularization.

Machine Learning Specialization content

This specialization is a great introduction to machine learning for beginners. It covers the basics of machine learning, such as supervised learning, unsupervised learning, and reinforcement learning.

My notes from the Machine Learning Specialization cover topics such as:

Supervised learning algorithms (linear regression, logistic regression, etc.)
Unsupervised learning algorithms (clustering, dimensionality reduction, etc.)
Recommender systems
Deep learning and neural networks
Advice for applying machine learning techniques in practice

My notes from the Machine Learning Specialization

Deep Learning Specialization content

This specialization is a more advanced specialization that focuses on deep learning techniques. Deep learning is a subset of machine learning that uses artificial neural networks to learn from data.

My notes from the Deep Learning Specialization cover a wide range of topics, including:

Neural networks and deep learning
Convolutional neural networks
Recurrent neural networks
Structuring machine learning projects
Natural Language Processing (NLP) and sequence models
Generative adversarial networks (GANs)
Special applications of deep learning

My notes from Deep Learning Specialization

Which specialization is right for you?

If you are new to machine learning, the Machine Learning Specialization is a great place to start. It will give you a solid foundation in the basics of machine learning.
If you are already familiar with machine learning and want to learn more about deep learning, the Deep Learning Specialization is a great option. It will teach you the latest deep learning techniques and how to use them to build real-world applications.
Ultimately, the best specialization for you will depend on your individual needs and goals. If you are not sure which specialization is right for you, I recommend starting with Machine Learning Specialization checking out the course descriptions and reviews on Coursera.

Conclusion

My experience with these two Coursera specializations was really great. I took the two specializations in succession, which took me about 7 months, spending a lot of hours every evening and on weekends. This allowed me to grasp the basics of machine learning, which are actually quite straightforward to understand, and to experiment with Python and TensorFlow.