1. Advanced Machine Learning with TensorFlow on Google Cloud Platform

This specialization is provided by Google itself. It aims to train you on Google Cloud Platform (GCP) and Tensorflow. I've been doing machine learning every day for a few years now so it went pretty smoothly for me. But this one is way bigger than the previous one I did (covered in my previous article). And even though you can totally do it as a beginner, it would be a bit hard. So this specialization is recommended for the "intermediate" level.

The curriculum consists of 5 courses:

• End-to-End Machine Learning with TensorFlow on GCP

  • This MOOC is covering the overall Machine Learning (ML) process from data exploration to deployment on GCP. The goal is twice: giving you an idea of how ML is done and mapping the GCP services to this process

  • 3 weeks of study, 8-10h per week

• Production Machine Learning Systems

  • This time, you learned some more specific concepts of how to create ML systems for production

  • 2 weeks and 5-7h/week expected

• Image Understanding with TensorFlow on GCP

  • If you are attracted by computer vision, you will like this one. It covers deep learning from the foundations of the actual models. Explanations are not the best and the pre-2.0 TensorFlow (the Qwiklabs are not updated) is not the easiest to start with but it's still interesting

  • 2 weeks and 5-7h/week expected

• Sequence Models for Time Series and Natural Language Processing

  • This one was the most interesting for me since I'm not that familiar with sequence data. It's introducing RNN and how to use them. Some simple NLP cases and explanations about tokenizing texts

  • 2 weeks and 5-7h/week expected

• Recommendation Systems with TensorFlow on GCP

  • As the title says, here you will focus on recommendation systems using collaborative filtering and then neural nets

  • 2 weeks of study and 6-8 hours/week expected

2. Feedback

Like the previous one, I enjoyed it but it's really time-consuming. Especially if you want to take notes! The explanations about neural networks are not the easiest to understand and TensorFlow is not super beginner-friendly. About Qwiklabs, although they are very interesting, some are quite old now. This might cause you some frustrations.

3. Is it worth it?

If your company is using GCP and you are planning to do some ML, I totally recommend this specialization. However, if you are interested in taking this course for specifically ML-related knowledge, I would not recommend it. The way to teach ML things is quite fast and if you don't have some solid understanding before, it might not help you.

Unlike the previous specialization, there is no Google certification associated with this course.

That's it! The next article will be about the last specialization that I followed "Architecting with Google Kubernetes Engine".