Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networks
Authors: Shiyu Liang, Yixuan Li, R. Srikant
ICLR 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section, we demonstrate the effectiveness of ODIN on several computer vision benchmark datasets. We run all experiments with Py Torch and we will release the code to reproduce all experimental results. |
| Researcher Affiliation | Collaboration | Shiyu Liang Coordinated Science Lab, Department of ECE University of Illinois at Urbana-Champaign sliang26@illinois.edu; Yixuan Li Facebook Research yixuanl@fb.com; R. Srikant Coordinated Science Lab, Department of ECE University of Illinois at Urbana-Champaign rsrikant@illinois.edu |
| Pseudocode | No | The paper describes the method using prose and equations, but does not include any pseudocode or algorithm blocks. |
| Open Source Code | Yes | We run all experiments with Py Torch1 and we will release the code to reproduce all experimental results2. 2https://github.com/facebookresearch/odin |
| Open Datasets | Yes | Each neural network architecture is trained on CIFAR-10 (C-10) and CIFAR-100 (C-100) datasets (Krizhevsky & Hinton, 2009), respectively. The Tiny Image Net dataset3 consists of a subset of Image Net images (Deng et al., 2009). The Large-scale Scene UNderstanding dataset (LSUN) (Yu et al., 2015). The i SUN (Xu et al., 2015). |
| Dataset Splits | Yes | For each out-of-distribution dataset, we randomly hold out 1,000 images for tuning the parameters T and ε. |
| Hardware Specification | No | The paper does not specify the exact hardware (e.g., GPU models, CPU types) used for running experiments. |
| Software Dependencies | No | The paper states 'We run all experiments with Py Torch' but does not specify a version number for PyTorch or any other software dependencies. |
| Experiment Setup | Yes | For Dense Net, our model follows the same setup as in (Huang et al., 2016), with depth L = 100, growth rate k = 12 (Dense-BC) and dropout rate 0. In addition, we evaluate the method on a Wide Res Net, with depth 28, width 10 (WRN-28-10) and dropout rate 0. [...] All neural networks are trained with stochastic gradient descent with Nesterov momentum (Duchi et al., 2011; Kingma & Ba, 2014). Specifically, we train Dense-BC for 300 epochs with batch size 64 and momentum 0.9; and Wide Res Net for 200 epochs with batch size 128 and momentum 0.9. The learning rate starts at 0.1, and is dropped by a factor of 10 at 50% and 75% of the training progress, respectively. |