ucd4ids/ldbexperiments

Tutorials and experiments for signal classification using Local Discriminant Basis.

Feature Extraction for Signal Classification using Local Discriminant Basis

Tutorials and experiments using Local Discriminant Basis (LDB). In this Pluto notebook we describe the algorithm behind LDB and test its classification capabilities by applying it on two standard signal classification datasets.

Authors

This note book is authored by Shozen Dan and Zeng Fung Liew under the supervision of Professor Naoki Saito at the University of California, Davis.

Table of Contents

1. Setup
2. Results
3. Pluto notebook containing results and report

How to Open and Run Pluto Notebook

Method 1 (Recommended): Opening notebook by cloning this repository

1. Clone the repository by typing the following:

git clone https://github.com/UCD4IDS/LDBExperiment.git

2. Navigate to the LDBExperiment directory and open up the Julia REPL.
3. Ensure Julia is working on the current directory. This can be checked using the following commands:

# shows the current working directory
julia> pwd() 

# change to the directory containing all the files from this repository. Eg:
# Windowns
julia> cd("C:/Users/USER/Documents/LDBExperiments")

# Linux, Mac
julia> cd("~/Documents/LDBExperiments")

4. Enter the package manager in the REPL by typing ]. The following should be observed:

(@v1.7) pkg> 

5. Activate the current environment by typing the following.
   Note: Steps 3-4 has to be done correctly for this step to work!

(@v1.7) pkg> activate .
(@v1.7) pkg> instantiate

6. Exit the package manager mode by hitting the backspace key. Then, type in the following commands:

julia> import Pluto; Pluto.run()

7. Pluto should open up in the default browser. Open up the file by keying in the file path.

Method 2: Opening notebook directly without downloading any files from this repository

1. Open up the Julia REPL.
2. Manually install the required packages for running the notebooks. The list of required packages can be found in the Project.toml file under the notebook directory.
   Install the packages in Julia using either the REPL or through the package manager. The package manager can be activated by hitting the ] key. Example:

# install on REPL
julia> using Pkg; Pkg.add("Pluto")
# install on package manager
(@v1.7) pkg> add Pluto

3. Return to the REPL and type the command below. If you are currently at the package manager mode, you can return to the REPL by hitting the backspace key.

julia> import Pluto; Pluto.run()

4. Pluto should open up in the default browser. Copy-paste the following URL into the file path:
   https://github.com/UCD4IDS/LDBExperiments/blob/main/LDBexperiment.jl

Note: When opening the notebooks using this method, Julia automatically downloads the notebook into the ~/.julia/pluto_notebooks folder in your local machine. You may want to delete them once you are done.

ldbexperiments's Issues

CI/CD for repo

Follow the instructions in https://github.com/JuliaPluto/static-export-template

LDBExperiments notebook in gh-pages does not instantiate environment based on Project.toml

Issue as shown in screenshot below.

Make comparisons for better LDB parameter selections

Create results and guidelines comparing the type of LDB algorithms (LDBK, LDBKASH, LDBEMD), discriminant measures (Asymmetric Relative Entropy, Symmetric Relative Entropy, etc), etc.

Goal: Be able to provide suggestions on the best type of LDB algorithm based on scenarios and the how to choose parameters.