Machine learning by tom mitchell pdf download - think
God!: Machine learning by tom mitchell pdf download
Dell inspiron 15 5000 series i3 drivers download | |
Free 60s and 70s music downloads | |
Expressvpn download file | |
E michael jones Tolkiens Failed Quest pdf download | |
D&d 5e equipment downloadable pdf |
Tim van Erven
Instructor: Tim van Erven
Teaching assistent: Rogier van het Schip
This is the main website for the machine learning course / at the vrije Universiteit. The course introduces the most important problems, ideas and methods from machine learning, which aims to design computer algorithms that improve automatically from their experiences.
In particular the interrelated problems of classification, regression and prediction are discussed, and we present the main machine learning approaches that address them. In analysing machine learning methods, the ideas of overfitting and the necessity of bias are of central importance. They form a recurring theme throughout the course.
Quicklink
Organisation (short version)
Course Location | |
---|---|
Sept. 6 Oct. 18 | Thursday , room 04A05 (hgb) |
Oct. 31 Dec. 12 | Wednesday , room KC |
Course materials: Machine Learning by Tom M. Mitchell, McGraw-Hill, and lecture notes, papers and tutorials. Dont forget to study the slides and extra materials!
There used to be a more elaborate web page about the organisation of the course, but I did not port it when moving to a new system to manage my website.
Course Schedule
Date | Topics | Assignment | Study |
---|---|---|---|
Sept. 6 | Basic concepts in ML: supervised vs unsupervised learning, prediction, regression, classification (e.g. concept learning) | - | Ch. 1, Ch. 2 up to sect. |
Sept. 13 | Math: Introduction to matrices and vectors; Basic concepts in ML: data representation | 1 | Ch. 1, Ch. 2 up to sect. |
Sept. 20 | Basic concepts in ML: hypothesis spaces, necessity of bias; Methods: least-squares linear regression, list-then-eliminate algorithm | - | Ch. 2 |
Sept. 27 | Math: directed graphs, trees, probability distributions; Method: decision trees | 2 | Ch. 2, Ch. 3 |
Oct. 4 | Information theory: entropy, mutual information; Math: probability distributions and random variables; Methods: decision trees continued | - | Ch. 3 |
Oct. 11 | Basic concepts in ML: train and test sets, overfitting; Methods: least squares regression with polynomials, pruning decision trees | 3 | Ch. 3 |
Oct. 18 | Methods: neural networks (perceptron) | - | Ch. 4 up to sect. |
Oct. 25 | Intermediate exam: - in Q | - | Ch. 1, 2, 3 with slides |
Oct. 31 | Methods: neural networks (perceptron); Math: Gradient descent. Note the change to Wednesday and KC! | 4 | Ch. 4 up to sect. |
Nov. 7 | Cancelled due to illness. | - | |
Nov. 14 | More about probability theory: Bayes' rule, independent events, random vectors; Methods: Instance-based learning (k-nearest neighbour), naive Bayes | 5 | Ch. 8 up to sect. , Sect. |
Nov. 21 | Weka demonstration and practical homework assignment description; Method: naive Bayes; Probability theory: distributions on the reals; Basic concepts in ML: models | 6 | Sect |
Nov. 28 | Methods: maximum likelihood parameter estimation, Bayesian learning | - | Sect. , Ch. 6 up to sect. |
Dec. 5 | Special guest lecture about Minimum Description Length (MDL) learning by Peter Grünwald, author of a comprehensive book about MDL. | - | (just the slides) |
Dec. 12 | Course summary, opportunity for questions; Student discussion about machine learning, in particular with respect to ethics (what about privacy?). Please read TDoML before the lecture! | - | TDoML |
Dec. 20 | Final exam: - in KC 07 | - | |
Feb. 14 | Resit of the final exam: - in G | - |
Slides
The slides should be available on the morning before each lecture, except for the slides of the guest lecture.
Slides | Lecture | Comments |
---|---|---|
www.cronistalascolonias.com.ar | Sept. 6 | To avoid confusing you, I've removed the slides at the end that we did not cover in class. The original slides are still available, although I don't see why you would need them. |
www.cronistalascolonias.com.ar | Sept. 13 | We did not cover the last example on these slides (EnjoySport) in class. I recommend that you study it on your own. |
www.cronistalascolonias.com.ar | Sept. 20 | Corrected `Warm' to `Cloudy' on slides 19 and The original slides are still available. |
www.cronistalascolonias.com.ar | Sept. 27 | Corrected `Warm' to `Cloudy' on slide 5. The original slides are still available. |
www.cronistalascolonias.com.ar | Oct. 4 | Changed `log' to `log2' everywhere for clarity. The original slides are still available. |
www.cronistalascolonias.com.ar | Oct. 11 | |
www.cronistalascolonias.com.ar | Oct. 18 | We only covered part of the slides in this lecture, because we spent our time discussing the answers to the second set of homework exercises. Therefore the things we did not cover are repeated in www.cronistalascolonias.com.ar We did not cover "Linear functions as inner products", "Implementing Boolean functions with a perceptron", "Convex functions" or "Gradient descent (part 1)". |
www.cronistalascolonias.com.ar | Oct. 31 | I suggest that you study the slides about optimizing perceptron weights, which we did not have time to cover during the lecture, on your own. They will help you understand the corresponding parts of Mitchell. |
www.cronistalascolonias.com.ar | Nov. 14 | |
mlslidespdf | Nov. 21 | In the lecture we only got to slide Studying the rest is optional. Some of it is covered in the next lecture. |
mlslidespdf | Nov. 28 | We got to slide 28, and along the way I already told you what is on slide Studying the other slides is optional. |
www.cronistalascolonias.com.ar www.cronistalascolonias.com.ar | Dec. 5 | No slides originally, because Peter did his lecture on the blackboard. You might have taken notes, but to help you study I have made slides after his lecture based on Peter's notes. |
mlslidespdf | Dec. 12 | Added missing curly braces on slide 8. |
Homework Assignments
You have to submit assignments using Blackboard: Under Course Documents click on the name of the assignment and you will get a form that allows you to upload your file.Exams
Intermediate Exam
The small group of students who took the intermediate exam at got a different version: version B. The other students got the normal version.
Final Exam
A few exchange students took the final exam one week earlier. They got a different version.
Resit Exam
Further Reading
If you want to read more on the topics from class, here are some pointers. None of this is mandatory.Reference | Description |
---|---|
Leo Dorst, Vectors for Beginners | This is a slightly more extensive introduction to vectors than we had time for in class. Note that it uses different notation. In class we have introduced vectors at the symbolic and computational level, as Leo calls them. To gain further intuition it is necessary to also give them a geometric interpretation. This (still very brief) introduction discusses all three interpretations. If you have not seen vectors before, reading this will require serious effort. |
Chapter 1 from the MDL book | Chapter 1 (and 17) of Peter Grünwald's book are available for free as sample chapters. If you want to learn more about MDL, I recommend that you read chapter 1, although it may require serious effort. Peter also has a tutorial about MDL available if you really want to get into it. |
-