mean-conc-beta
Beta distribution parameterized by mean and concentration for bounded continuous action spaces in reinforcement learning.
Install
$ pip install mean-conc-beta
Usage
import torch
from mean_conc_beta import Beta
beta = Beta()
# network output: (batch, num_actions, 2) for raw mean and concentration
params = torch.randn(16, 4, 2, requires_grad = True)
# distribution on (-1, 1)
dist = beta(params)
# sample actions
actions = dist.sample()
actions_reparam = dist.rsample()
# log prob and entropy
log_prob = beta.log_prob(dist, actions)
entropy = beta.entropy(dist)
# behavior cloning with mse loss on mean
expert_actions = torch.rand(16, 4)
pred_mean = beta.mean(params)
bc_loss = (pred_mean - expert_actions).pow(2).mean()
bc_loss.backward()
Citations
@article{Ferrari2004BetaRF,
title = {Beta Regression for Modelling Rates and Proportions},
author = {Silvia L. P. Ferrari and Francisco Cribari-Neto},
journal = {Journal of Applied Statistics},
year = {2004},
volume = {31},
pages = {799 - 815}
}
@inproceedings{Chou2017TheBP,
title = {The Beta Policy for Continuous Reinforcement Learning},
author = {Po-Wei Chou and Daniel Maturana and Sebastian Scherer},
booktitle = {International Conference on Machine Learning},
year = {2017}
}
1 maintainer
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
mean_conc_beta-0.0.5.tar.gz
(6.8 kB
view details)
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file mean_conc_beta-0.0.5.tar.gz.
File metadata
- Download URL: mean_conc_beta-0.0.5.tar.gz
- Upload date:
- Size: 6.8 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via:
uv/0.8.17
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
05b5109b4a2fcd7e5ad15f5945fe8898c3665c33518b4a451b534e07ccf824ee
|
|
| MD5 |
210cf51944d1cc1bac6e2bf7b0c7a976
|
|
| BLAKE2b-256 |
af482179a9a2548a9395736024060dba80a9724fcb359341278d0484ea094364
|
File details
Details for the file mean_conc_beta-0.0.5-py3-none-any.whl.
File metadata
- Download URL: mean_conc_beta-0.0.5-py3-none-any.whl
- Upload date:
- Size: 6.1 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via:
uv/0.8.17
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
8a5696f646215110408e4ac16bd3540d551debe4f223dc08d341a12bf582290d
|
|
| MD5 |
5ab06d8b622e4119fd3ba03a09932732
|
|
| BLAKE2b-256 |
9d16381f332d3a74694bf61243aa33ca4637c79fc092f90f8c7f9641346458eb
|
