This is a PyTorch implementation of LipsNet.
The paper is accepted at ICML 2023 with the title 'LipsNet: A Smooth and Robust Neural Network with Adaptive Lipschitz Constant for High Accuracy Optimal Control'.
LipsNet can serve as actor networks in most actor-critic reinforcement learning algorithms, in order to reduce the action fluctuation. A low level of action fluctuation will protect mechanical components from the wear and tear, and reduce safety hazards.
The overall structure is shown below:
The version of PyTorch should be higher than 1.11 and lower than 2.3, as we incorporate functorch.jacrev and functorch.vmap methods.
We package LipsNet as a PyTorch module.
Practitioners can easily use it just like using MLP.
from lipsnet import LipsNet
# declare
net = LipsNet(...)
# training
net.train()
out = net(input)
...
loss.backward()
optimizer.step()
optimizer.zero_grad()
net.eval()
# evaluation
net.eval()
out = net(input)
More details can be found in lipsnet.py.
@InProceedings{pmlr-v202-song23b,
title = {LipsNet: A Smooth and Robust Neural Network with Adaptive Lipschitz Constant for High Accuracy Optimal Control},
author = {Song, Xujie and Duan, Jingliang and Wang, Wenxuan and Li, Shengbo Eben and Chen, Chen and Cheng, Bo and Zhang, Bo and Wei, Junqing and Wang, Xiaoming Simon},
booktitle = {Proceedings of the 40th International Conference on Machine Learning},
pages = {32253--32272},
year = {2023},
volume = {202},
series = {Proceedings of Machine Learning Research},
month = {23--29 Jul},
publisher = {PMLR}
}
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent