https://github.com/davidrusu/ctw

An Implementation of the Context Tree Weighting (CTW) Sequence Prediction Algorithm
https://github.com/davidrusu/ctw

rust sequence-prediction

Last synced: 22 days ago
JSON representation

An Implementation of the Context Tree Weighting (CTW) Sequence Prediction Algorithm

• Host: GitHub
• URL: https://github.com/davidrusu/ctw
• Owner: davidrusu
• Created: 2022-03-05T03:14:31.000Z (over 2 years ago)
• Default Branch: master
• Last Pushed: 2022-03-05T04:17:14.000Z (over 2 years ago)
• Last Synced: 2024-09-14T01:42:12.589Z (about 2 months ago)
• Topics: rust, sequence-prediction
• Language: Rust
• Homepage: http://www.davidrusu.com/blog#ctw
• Size: 5.86 KB
• Stars: 7
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        


# Context Tree Weighting (CTW)

CTW is a lightweight, practical and well performing sequence prediction algorithm discovered by Frans Willems, Yuri Shtarkov and Tjalling Tjalkens (1995).

It has good query and update performance (linear in the context length).

# Useage

The following example demonstrates CTW learning the altenating binary sequence 101010...

```rust

use ctw::CtwTree;

let mut rng = rand::thread_rng();

let mut tree = CtwTree::new(8); // context length is 8

let pattern = [true, false].into_iter().cycle().take(100); // true, false, true, ..

tree.update_batch(&Vec::from_iter(pattern));

let mut predictions = Vec::new();

for _ in 0..10 {

    let prediction = tree.sample(&mut rng);

    tree.update(prediction);

    // convert bool to 1 and 0 for legibility

    let bit = if prediction { 1 } else { 0 };

    predictions.push(bit);

}

println!("predictions: {predictions:?}"); // --> "prediction: [1, 0, 1, 0, 1, 0, 1, 0, 1, 0]"

assert_eq!(predictions, vec![1,0,1,0,1,0,1,0,1,0]);

```

# Resources

1. http://www.hutter1.net/publ/sctw.pdf

2. https://web.stanford.edu/class/ee378a/lecture-notes/lecture_9.pdf

3. http://www.data-compression.info/Algorithms/CTW/

4. https://www.cs.cmu.edu/~aarti/Class/10704_Spring15/CTW.pdf (original paper)