UDACITY is an online education well known by many experts from science and industry has a set of 3 courses with the topic “Machine Learning”. The first one is calles “Machine Learning: Supervised Learning”.
They are hold by
Michael Littman, Professor of Computer Science at Brown University and
Charles Isbell, Professor and Senior Associate Dean at the School of Interactive Computing at Georgia Tech
[jwplayer player=”1″ mediaid=”82″]
The first part is giving a short introduction, some definitions
e.g. if (Machine Learning) ML is considered to be “computational statistics” (MLitt) or a “broader notion of building computational artefacts that learn over time based on experience” (CI).
Generally it’s said, that Supervised Learning is Functional Approximation: Generalisation, Inductive Bias -> Deduction aka ‘Reasoning’.
Unsupervised Learning: has not those input, therefore creating an own structure and classes to create a compact Description of the data sets.
Reinforcement Learning is Learning from delayed reward, even if it will still be unclear what exactly was the mistake (e.g. at a lost tick tack toe game).
They are going to show Tools and Techniques.
And talk about : supervised learning: labels data well – reinforcement learning: behaviour scores well – unsupervised learning: cluster scores well.
http://www.udacity.com/course/ud675 preview – (www)