ODIN: Out-of-Distribution Detector for Neural Networks

This is a PyTorch implementation for detecting out-of-distribution examples in neural networks. The method is described in the paper Principled Detection of Out-of-Distribution Examples in Neural Networks by S. Liang, Yixuan Li and R. Srikant. The method reduces the false positive rate from the baseline 34.7% to 4.3% on the DenseNet (applied to CIFAR-10) when the true positive rate is 95%.

Experimental Results

We used two neural network architectures, DenseNet-BC and Wide ResNet. The PyTorch implementation of DenseNet-BC is provided by Andreas Veit and Brandon Amos. The PyTorch implementation of Wide ResNet is provided by Sergey Zagoruyko. The experimental results are shown as follows. The definition of each metric can be found in the paper. 

Pre-trained Models

We provide four pre-trained neural networks: (1) two DenseNet-BC networks trained on CIFAR-10 and CIFAR-100 respectively; (2) two Wide ResNet networks trained on CIFAR-10 and CIFAR-100 respectively. The test error rates are given by:

Architecture | CIFAR-10 | CIFAR-100 ------------ | --------- | --------- DenseNet-BC | 4.81 | 22.37 Wide ResNet | 3.71 | 19.86

Running the code

Dependencies

• CUDA 8.0

• PyTorch

• Anaconda2 or 3

• At least three GPU

  Note: Reproducing results of DenseNet-BC only requires one GPU, but reproducing results of Wide ResNet requires three GPUs. Single GPU version for Wide ResNet will be released soon in the future.

Downloading Out-of-Distribtion Datasets

We provide download links of five out-of-distributin datasets:

• Tiny-ImageNet (crop)

• Tiny-ImageNet (resize)

• LSUN (crop)

• LSUN (resize)

• iSUN

Here is an example code of downloading Tiny-ImageNet (crop) dataset. In the root directory, run

mkdir data
cd data
wget https://www.dropbox.com/s/avgm2u562itwpkl/Imagenet.tar.gz
tar -xvzf Imagenet.tar.gz
cd ..

Downloading Neural Network Models

We provide download links of four pre-trained models.

• DenseNet-BC trained on CIFAR-10

• DenseNet-BC trained on CIFAR-100

• Wide ResNet trained on CIFAR-10

• Wide ResNet trained on CIFAR-100

Here is an example code of downloading DenseNet-BC trained on CIFAR-10. In the root directory, run

mkdir models
cd models
wget https://www.dropbox.com/s/wr4kjintq1tmorr/densenet10.pth.tar.gz
tar -xvzf densenet10.pth.tar.gz
cd ..

Running

Here is an example code reproducing the results of DenseNet-BC trained on CIFAR-10 where TinyImageNet (crop) is the out-of-distribution dataset. The temperature is set as 1000, and perturbation magnitude is set as 0.0014. In the root directory, run

cd code
# model: DenseNet-BC, in-distribution: CIFAR-10, out-distribution: TinyImageNet (crop)
# magnitude: 0.0014, temperature 1000, gpu: 0
python main.py --nn densenet10 --out_dataset Imagenet --magnitude 0.0014 --temperature 1000 --gpu 0

Note: Please choose arguments according to the following.

args

• args.nn: the arguments of neural networks are shown as follows

  Nerual Network Models | args.nn ----------------------|-------- DenseNet-BC trained on CIFAR-10| densenet10 DenseNet-BC trained on CIFAR-100| densenet100

• args.out_dataset: the arguments of out-of-distribution datasets are shown as follows

  Out-of-Distribution Datasets | args.out_dataset ------------------------------------|----------------- Tiny-ImageNet (crop) | Imagenet Tiny-ImageNet (resize) | Imagenet_resize LSUN (crop) | LSUN LSUN (resize) | LSUN_resize iSUN | iSUN Uniform random noise | Uniform Gaussian random noise | Gaussian

• args.magnitude: the optimal noise magnitude can be found below. In practice, the optimal choices of noise magnitude are model-specific and need to be tuned accordingly.

  Out-of-Distribution Datasets | densenet10 | densenet100 | wideresnet10 | wideresnet100 ------------------------------------|------------------|------------- | -------------- |-------------- Tiny-ImageNet (crop) | 0.0014 | 0.0014 | 0.0005 | 0.0028 Tiny-ImageNet (resize) | 0.0014 | 0.0028 | 0.0011 | 0.0028 LSUN (crop) | 0 | 0.0028 | 0 | 0.0048 LSUN (resize) | 0.0014 | 0.0028 | 0.0006 | 0.002 iSUN | 0.0014 | 0.0028 | 0.0008 | 0.0028 Uniform random noise | 0.0014 | 0.0028 | 0.0014 | 0.0028 Gaussian random noise | 0.0014 |0.0028 | 0.0014 | 0.0028

• args.temperature: temperature is set to 1000 in all cases.

• args.gpu: make sure you use the following gpu when running the code:

  Neural Network Models | args.gpu ----------------------|---------- densenet10 | 0 densenet100 | 0 wideresnet10 | 1 wideresnet100 | 2

Outputs

Here is an example of output.

Neural network architecture:          DenseNet-BC-100
In-distribution dataset:                     CIFAR-10
Out-of-distribution dataset:     Tiny-ImageNet (crop)

                          Baseline         Our Method
FPR at TPR 95%:              34.8%               4.3% 
Detection error:              9.9%               4.6%
AUROC:                       95.3%              99.1%
AUPR In:                     96.4%              99.2%
AUPR Out:                    93.8%              99.1%