Home

Awesome

PALM

This is official code repository for

Learning with Mixture of Prototypes for Out-of-Distribution Detection

Haodong Lu, Dong Gong, Shuo Wang, Jason Xue, Lina Yao, Kristen Moore

The Twelfth International Conference on Learning Representations (ICLR) 2024

[arXiv]

Method

fig_model

How To Start

Requirements

All experiments were conducted using the following libraries on a single RTX3090 GPU.

Prepare Datasets

The default root directory for ID and OOD datasets is data/.

ID datasets Datasets like CIFAR-10 & CIFAR-100 will be automatically downloaded.

OOD datasets We use SVHN, Textures (dtd), Places365, LSUN-C (LSUN) and iSUN as our primary OOD datasets in our experiments.

OOD datasets can be downloaded via the following links :

Training and Evaluation

The training and evaluation scripts are presented in runner.sh file.

Pretrained weights

Please check out our pretrained weights here, and configure the save_path argument in runner.sh for evaluation.

Experiment Results

fig1

Citation

If you find our work useful, please consider citing our paper:

@inproceedings{
PALM2024,
title={Learning with Mixture of Prototypes for Out-of-Distribution Detection},
author={Haodong Lu, Dong Gong, Shuo Wang, Jason Xue, Lina Yao, Kristen Moore},
booktitle={The Twelfth International Conference on Learning Representations},
year={2024},
url={https://openreview.net/forum?id=uNkKaD3MCs}
}