Adaptive Information Processing

Bert de Vries, Tjalling Tjalkens and Marco Cox
Eindhoven University of Technology, Dept. of Electrical Engineering
Corr. to [email protected]

This site contains materials for course 5SSB0 (Adaptive Information Processing) at TU/e.

Read-only versions

You can view the lecture notes through the links below:

• 0 - Introduction
• 1 - Machine Learning Overview
• 2 - Probability Theory Review
• 3 - Bayesian Machine Learning
• 4 - Working with Gaussians
• 5 - Density Estimation
• 6 - Linear Regression
• 7 - Generative Classification
• 8 - Discriminative Classification
• 9 - Clustering with Gaussian Mixture Models
• 10- The EM Algorithm
• 11- Continuous Latent Variable Models - PCA and FA
• 12- Factor Graphs and Message Passing Algorithms
• 13- Dynamic Latent Variable Models
• 14- EM as a Message Passing Algorithm

Opening the lecture notes locally

To open the lecture notes in IJulia, download the .ipynb files to your computer and start a Jupyter notebook by

$ jupyter notebook

You will now get a new page in your browser with a list of available notebooks. Alternatively, if you don't have Julia/Jupyter installed on your system, you can use JuliaBox to run the notebooks (see exact instructions below).

To run the interactive code examples in the lecture nodes, the following Julia packages are required: Cubature, DataFrames, Distributions, Interact, PyPlot. To install the required packages, execute:

map(Pkg.add, ["Cubature", "DataFrames", "Distributions", "Interact", "PyPlot"])

Running the lecture notes on JuliaBox.org

Follow these instructions to run the code examples from the lecture notes online through JuliaBox.

1. (Create account) Go to https://www.juliabox.org/, create an account and log in.

2. (Necessary JuliaBox hack) Click the 'IJulia tab', and then create a new text file using New -> Text File. Paste the following content in the new file:

   VERSION >= v"0.4" && splice!(Base.LOAD_CACHE_PATH, 3)
   

   Then, press ctrl+s to save the file and close the browser tab. Back in the IJulia file explorer, click the checkbox in front of untitled.txt, click Rename, and rename the file to .juliarc.jl.

3. (Install required packages) Go to the Console tab, and then start a Julia v0.4.3 shell by typing

   /opt/julia-0.4.3/bin/julia
   

   In the Julia shell, execute the following command to install all required packages:

   map(Pkg.add, ["Cubature", "DataFrames", "Distributions", "Interact", "PyPlot"])

   Afterwards, type exit() to quit Julia.

4. (Import lecture notes into JuliaBox) Go to the Sync tab, and add the lecture notes git repository through the following actions:

5. Paste https://github.com/bertdv/AIP-5SSB0.git in the Git Clone URL field

6. Click with the mouse in the branch field. You should get master in the branch field and AIP-5SSB0 in the juliabox field.

7. Press the plus-sign.

You can now open the lecture notes by going to the IJulia tab (press the refresh button if the folder AIP-5SSB0 does not show up). Navigate to a specific lesson and click the .ipynb file to open the notebook.

Creating a PDF bundle of all lessons

This procedure will only be able to generate the PDF bundle for all lessons if ForneyLab is available as it is used for the later chapters. This is not publicly available software, so the functionality is limited for those who do not have access to it.

Get ForneyLab.jl. Either clone it in the root of the project or modify the path in the Dockerfile to point at your own installation. If you do not have access to it, but do want to generate a PDF of the remaining lessons, just create an empty ForneyLab.jl directory.

Install Docker from https://www.docker.com.

Finally from the root directory of the project issue

$ docker build -t aip-5ssb0-bundler .
$ docker run --rm \
             --volume ${PWD}/lessons:/aip-5ssb0-bundler/lessons \
             --volume ${PWD}/output:/aip-5sbb0-bundler/output \
             aip-5ssb0-bundler

to obtain a bundle.pdf file containing all lessons in the output directory.

License

Adaptive Information Processing (5SSB0) by Bert de Vries, Tjalling Tjalkens and Marco Cox is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Based on a work at https://github.com/bertdv/AIP-5SSB0.