https://github.com/ziyuw/rembo

Bayesian optimization in high-dimensions via random embedding.
https://github.com/ziyuw/rembo

Last synced: 3 months ago
JSON representation

Bayesian optimization in high-dimensions via random embedding.

• Host: GitHub
• URL: https://github.com/ziyuw/rembo
• Owner: ziyuw
• Created: 2013-08-04T20:52:59.000Z (almost 11 years ago)
• Default Branch: master
• Last Pushed: 2013-08-04T20:55:23.000Z (almost 11 years ago)
• Last Synced: 2024-01-20T13:37:17.142Z (5 months ago)
• Language: Matlab
• Size: 207 KB
• Stars: 108
• Watchers: 10
• Forks: 24
• Open Issues: 3
• Metadata Files:
  • Readme: README.md

Lists

• awesome-AutoML - REMBO - dimensions via random embedding (Tools and projects / LLM)
• awesome-automl - ziyuw/rembo - dimensions via random embedding. (Libraries)
• awesome-open-source-mlops - REMBO - Bayesian optimization in high-dimensions via random embedding. (Scheduling)
• awesome-llmops - REMBO - dimensions via random embedding. |  | (AutoML / Profiling)
• my-awesome-stars - ziyuw/rembo - Bayesian optimization in high-dimensions via random embedding. (Matlab)
• Awesome-LLMOps - REMBO - Bayesian optimization in high-dimensions via random embedding. (Profiling)

README

        
## REMBO

This package contains code for the paper "Bayesian Optimization in a Billion 

Dimensions via Random Embeddings". The paper tries to solve the problem of 

doing Bayesian optimization in high dimensions. For more details please read 

the paper (http://www.cs.ubc.ca/~ziyuw/papers/rembo.pdf). 

### INSTALL

The package is written in Matlab. Thus to install, just run startup.m. This 

package could make use of the function "fminsearch" from Matlab. 

This, however, is not necessary.

To run the lpsolve example from the paper, please download the data files from:

http://www.cs.ubc.ca/~ziyuw/AC_blackbox_eval.tar.

And untar the data files to the folder /demos/lpsolve/.

### Configuration files

This package allows one the specify the parameters settings in 

a configuration file. In the configuration file, one can specify

the type of parameters to be optimized. 

Currently, the package supports continuous, discrete, and categorical

parameters. It also allows one to use log scales. 

For an example of such a configuration file, please checkout 

"demos/lpsolve/lpsolve.yaml".