Skip to content

Commit

Permalink
Update projects.md
Browse files Browse the repository at this point in the history
  • Loading branch information
sgbaird committed Feb 21, 2024
1 parent eacc5a7 commit bcf89ba
Showing 1 changed file with 41 additions and 12 deletions.
53 changes: 41 additions & 12 deletions projects.md
Original file line number Diff line number Diff line change
Expand Up @@ -21,30 +21,30 @@ We are still refining the list of benchmark tasks. Please contact us if you have

##### Domain

- [Analytical functions]()
- [Molecule Optimization]()
- [Materials Optimization]()
- Analytical functions
- Molecule Optimization
- Materials Optimization

##### Optimization features

- [Multi-fidelity Optimization]()
- [Constrained Optimization]()
- [Multi-objective Optimization]()
- [Real-world noise]()
- Multi-fidelity Optimization
- Constrained Optimization
- Multi-objective Optimization
- Real-world noise

#### Judging Criteria
<!-- #### Judging Criteria -->

### Developing New Benchmarks

Ideally, these tasks will be representative of real-world problems in chemistry and materials science. While remotely accessible automated experiments would be the gold standard, the more pragmatic benchmark tasks typically include surrogate modeling. The new benchmark tasks should lean towards real-world conditions in terms of optimization problem type (objectives, fidelities, constraints) and/or relevance towards chemistry and materials applications (e.g., molecules, materials, reactions, etc.).

#### Judging Criteria
<!-- #### Judging Criteria -->

### Creating Instructional Tutorials

We will provide a set of topics in Bayesian optimization, and your job will be to create a tutorial that introduces the topic conceptually and provides a hands-on example. These are meant to be "gentle introduction" tutorials which assume beginner Python knowledge and Bayesian optimization knowledge (see [resources](_/../resources.md) for more details).

#### Judging Criteria
<!-- #### Judging Criteria -->

### Real-world Chemistry and Materials Tasks

Expand All @@ -54,8 +54,12 @@ The hackathon is also open to proposals for real-world optimization tasks in che

As a teamleader, to initialize your project(s)[<sup>(?)</sup>][faq]{:title="Can I participate in multiple projects?"}, please follow these steps:
1. Accept the GitHub Classroom invitation to a topic above
2. Create a new file in a fork of the hackathon repository named `project-<your-team-name>.md`. For example, if your team name is "Bayes Bandits", the file should be named `project-bayes-bandits.md`.
3. Copy the following contents into the file and fill in the corresponding details. Replace `<your-repo-name>` wit the GitHub repository that was created by GitHub Classroom for you. For example, if your team name is "Bayes Bandits", the repository will be named `bayes-bandits`, and the `github` field should be `AC-BO-Hackathon/bayes-bandits`.
2. Create a new team in the GitHub Classroom interface
3. Create a new file in a fork of the hackathon repository named `project-<your-team-name>.md`. For example, if your team name is "Bayes Bandits", the file should be named `project-bayes-bandits.md`.
4. Copy the following contents into the file and fill in the corresponding details. Replace `<your-repo-name>` with the GitHub repository that was created by GitHub Classroom for you. For example, if your team name is "Bayes Bandits", the repository will be named `bayes-bandits`, and the `github` field should be `AC-BO-Hackathon/bayes-bandits`.
5. Submit a pull request to the hackathon repository with the title "Add project <your-team-name>".
6. Ask your team members to click on the same GitHub Classroom invitation link that you used and join the team you created in step 2.
6. Once the pull request is merged, your project will appear in the list below.

```markdown
---
Expand All @@ -79,4 +83,29 @@ Project 1 description
- ...
```

Here is an example of a filled-in project file called `project-bayes-bandits.md` for the "Bayes Bandits" team:

```markdown
---
number: 1 <!-- leave as-is, maintainers will adjust -->
title: Investigation of Bandit Optimization for Composite Materials Design
pis:
- Jane Doe (University of Invention)
- John Smith (Institute of Discovery)

# Comment these lines to hide these elements
contributors:
- Larry Lab (University of Invention)
- David Data (University of Science)
- Rachel Research (Institute of Discovery)
github: AC-BO-Hackathon/bayes-bandits
<!-- youtube_video: lIanN0DI9R8 -->
---

This project will investigate the application of bandit optimization to the design of composite materials. We will focus on the optimization of the mechanical properties of the composite materials, such as strength, stiffness, and toughness as a function of the fiber types and matrix materials. We will compare the performance of bandit optimization with the performance of Bayesian optimization using featurization tactics for this highly discrete space.

- Aleksandrs Slivkins (2019), "Introduction to Multi-Armed Bandits", Foundations and Trends in Machine Learning: Vol. 12: No. 1-2, pp 1-286. http://dx.doi.org/10.1561/2200000068
- Dimmery, D., Bakshy, E., & Sekhon, J. (2019). Shrinkage Estimators in Online Experiments. arXiv. https://doi.org/10.48550/ARXIV.1904.12918
```

[faq]: {{ site.baseurl }}{% link faq.md %}

0 comments on commit bcf89ba

Please sign in to comment.