- 2 reviews
- 2 completed
This course is part of the Deep Learning specialization. Week 1: Regularisation and why/how it works. Week 2: Optimization algorithms. It turns out that gradient descent is not the only implementation you can use! Week 3: Guidelines on how to tune your hyperparameters, batch normalization (a technique to accelerate learning), and an introduction to Tensorflow. Assignments are in Numpy (using Jupyter notebooks), with a small Tensorflow exercise in the end. Overall a great course when coupled with the first course in the Deep Learning specialization.
This course is an approachable introduction to machine learning. It gives you tools you can immediately use for practical applications. You should know basic linear algebra (matrix multiplication, transposes, dot products, etc...) and what derivatives are, and be comfortable with mathematical notation. You won't be required to do something like differentiate a function, and in this sense the course isn't very mathematically rigorous. The assignments are educational but on the easier side - implement the rest of some algorithm, mostly. Perhaps this is the best route for an introductory level course though, and its a really good (and fun) introduction. I think you'll get the most value from it by applying it to your own projects. Also, check out Stanford’s official version of the course (CS 229) for a more complete approach. Its also taught by Andrew, and the slides and videos are available freely.