I have finished the Machine Learning Course offered by Stanford University on Coursera, and since there are already several who have asked me openly and privately about it, I wanted to detail a little more what it seemed to me and that whoever decides to do it knows what they are going to find.
It is a free course on Machine Learning, taught by Andrew Ng. once finished if you want you can have a certificate that endorses the skills achieved for € 68. It is divided into 3 pillars, videos, Exams or Quizz and programming exercises. It is in English. You have subtitles in several languages, but the Spanish are not very good and sometimes they are out of date, much better if you put them in English.
It is quite theoretical. But maybe that's why it seems like a good way to start because you are not only going to learn what to do but why you do it.
- When to choose one algorithm or another.
- How to choose and define the different parameters.
- What problems can arise with the algorithms and especially what measures to take.
It has a lot of algebra and some calculus, and see, as I explain it, you really won't have to operate, you won't have to arrive at those equations, prove them, or modify them, well just vectorize them. So even if your level of mathematics is not good you could do the course, but of course, spending hours watching and listening to videos where they explain each term how it influences and why it is there, is hard.
If you don't know what it is Machine Learning, let's say it is a part of Artificial Intelligence that is dedicated to algorithms that make all this from machine vision, spam classification, etc, etc work.
My vision has changed me. When you thought about these types of problems, you faced them from a programming point of view, thinking about loops, conditions, etc. and they are really all functions, minimization of cost functions, which can be distances between points. Predictions based on regressions, etc, etc
So above these are the main parts of the course, divided into two, the Supervised part and the Unsupervised part
- Model and Cost function
- Gradient descent for linear regression
- Neural Networks
- Large Machine Classification and Kernels
- Principal Component Analysis (PCA)
- Machine learning system design
- Vector Machines Support
- Dimensionality Reduction
- Anomaly detection
- Recommender Systems
- Large Scale Machine Learning
I leave things but come on is the main thing, then everything breaks down.
For practice you use Matlab or Octave that we could say the Matlab OpenSource. I have done the course with Octave. As indicated in the first courses, they have chosen these tools because they allow rapid prototyping of the algorithms. With other tools the student would spend too much time programming.
What is certain is that although it is not easy, they leave everything ready for you to finish it off. You have the whole environment ready for the exercises, the data sets, the plots of the graphs, many functions and variables to use and what the student does is fill in a few lines with the main algorithms.
I repeat, it is not trivial, especially since you spend a lot of time watching how something is done with Octave.
Seeing examples of applications and what can be done I have no doubt that this is the future of the industry. Any company will end up implementing solutions with machine learning, artificial intelligence or whatever we want to call it to improve predictions, control quality and improve different production processes. Attentive that I am not only talking about applications, or the online world, but about physical companies, services, production, logistics, etc.
In addition to the already known ones, voice recognition, OCR, computer vision, language translators,
Recommend systems, predictions
And now that
This year my idea is to try to put into practice what I have learned by creating some tools that would be of great help to us at work. I know that it will not be easy and that I will have to familiarize myself with Python and some framework, well Tensor Flow, PyTorch and a library like Numpy. I have to probe the market.
In addition, I would like to delve into Deep Learning with the free course offered at http://course.fast.ai/ and also start with Big Data, another of the fields related to artificial intelligence and Machine Learning and that would also be very useful to me at my work. I've been looking at specializing in Coursera Big Data there are better ones but much more expensive.
If you have any questions you can leave a comment.