The idea of this project is to maintain an overview over the current state of research on Algorithms with Predictions and collect links to further relevant material. This should especially help new researchers in this field to orient faster, but we also try to keep track over and cluster the large amount of results and publications in this field.

The sources and data for this webpage are available on GitHub. It uses Gatsby for static site rendering and is hosted via GitHub Pages.

Everyone is very welcome to contribute:

• Maintain paper references, e.g. adding, updating and labeling references (see papers/ directory and syntax below)
• Add and edit further material. This page is generated from the Markdown file src/markdown-pages/material.md and is thus easy to modify.
• Improve layout or design, add new features.

Most contributions can be done via Pull Requests directly in the repository. For more involved suggestions or discussions, feel free to contact us ([email protected]).

Paper entries are based on YAML files, which are located in the directory papers/.

As an example, this is the data file for the caching paper by Lykouris and Vassilvitskii (LykourisV18competitive.yml):

title: Competitive Caching with Machine Learned Advice
authors: Lykouris, Vassilvitskii
publications:
  - name: ICML
    year: 2018
    url: http://proceedings.mlr.press/v80/lykouris18a/lykouris18a.pdf
  - name: arXiv
    year: 2018
    month: 1 # optional
    day: 4 # optional
    url: https://arxiv.org/pdf/1802.05399.pdf
  - name: J. ACM
    year: 2021
    url: https://dl.acm.org/doi/10.1145/3447579
labels: 
  - online
  - caching/paging 

If you want to add or change the entry of a paper, you can either add/edit the file via a Pull Request or send us the file via e-mail. In case you want to add a paper, please try to find a unique filename (as in the example above; but there are no strict conventions).

It is also possible to only add the title and labels of a paper:

title: Competitive Caching with Machine Learned Advice
labels: 
  - online
  - caching/paging 

We use an automated procedure to fetch the authors last names and all publications which have an (almost) matching title from DBLP and arXiv. In this example, all three publications of this paper can be added automatically.

You can start a local development instance of this webpage as follows:

First, clone this repository and install dependencies via a node package manager, e.g. yarn:

yarn

Then, install the gatsby command line interface

yarn global add gatsby-cli

and finally run

yarn develop