Skip to content

Latest commit

 

History

History
108 lines (82 loc) · 4.42 KB

CONTRIBUTING.md

File metadata and controls

108 lines (82 loc) · 4.42 KB
Raw

Contributing

We have adopted the Contributor Code of Covenant. Please read and follow these!

  • Did you notice a paper is missing?
  • Would you have found a paper sooner if this was organized in another way?
  • Is there a feature that would improve this list?

Please help us expand this list by opening Issues or forking/sending a Pull Request with a .bib file.

Updating README.md

Making Changes

Do not modify the README.md at the base of the repository.

  • To change a citation, adjust the corresponding file in the bib/ folder.
  • To change anything else, adjust the static/README.md file.

static/README.md is a template that the main file is built from. This is done by injecting citations built from .bib files.

Building

Once you have added .bib files or modified static/README.md, make sure you have the necessary Python dependencies:

$ pip install -r src/requirements.txt

Then build a new copy using the Makefile:

make clean
make

Additional notes on the build process are located in the src/ directory.

Mission

The original Bayesian Networks Repository laid out the mission statement below.

We hope to incorporate many of these elements if possible. But our primary goal is to organize the landscape of research on Bayesian Networks.

The intention is to construct a repository that will allow us to do empirical research within our community by facilitating (1) better reproducibility of results, and (2) better comparisons among competing approaches. Both of these are required to measure progress on problems that are commonly agreed upon, such as inference and learning.

A motivation for this repository is outlined in "Challenge: Where is the Impact of Bayesian Networks in Learning?" by N. Friedman, M. Goldszmidt, D. Heckerman, and S. Russell (IJCAI-97).

This will be achieved by several progressive steps:

  1. Sharing domains. This would allow for reproduction of results, and also allow researchers in the community to run large scale empirical tests.

  2. Sharing task specification. Sharing domains is not enough to compare algorithms. Thus, even if two papers examine inference in a particular network, they might be answering different queries or assuming different evidence sets. The intent here is to store specific tasks. For example, in inference this might be a specific series of observations/queries. In learning, this might be a particular collection of training sets that have a particular pattern of missing data.

  3. Sharing task evaluation. Even if two researchers examine the same task, they might use different measures to evaluate their algorithms. By sharing evaluation methods, we hope to allow for an objective comparison. In some cases such evaluation methods can be shared programs, such as a program the evaluates the quality of learned model by computing KL divergence to the "real" distribution. In other cases, such an evaluation method might be an agreed upon evaluation of performance, such as space requirements, number of floating point operations, etc.

  4. Organized competitions. One of the dangers of empirical research is that the methods examined become overly tuned to specific evaluation domains. To avoid that danger, it is necessary to use "fresh" problems. The intention is to organize competitions that would address a specific problems, such as causal discovery, on unseen domains.

New Topics

Because of the architecture, creating new topics is as simple as listing a topic in the .bib file for a paper.

But before you do, make sure you ask yourself: are the existing topics truly insufficient to describe what this paper does?

Is this actually an awesome-list?

Not technically. The Requirements for Awesome list(s) specifically require:

Non-generated Markdown file in a GitHub repo.

We generate a significant portion of our list from .bib files.