Harnessing Out-Of-Distribution Examples via Augmenting Content and Style
Authors: Zhuo Huang, Xiaobo Xia, Li Shen, Bo Han, Mingming Gong, Chen Gong, Tongliang Liu
ICLR 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section, we first describe the implementation details. Then, we experimentally validate our method on three applications, namely OOD detection, open-set SSL, and open-set DA. Finally, we present extensive performance analysis on our disentanglement and intervention modules. Additional details and quantitative findings can be found in the supplementary material. |
| Researcher Affiliation | Collaboration | Zhuo Huang1, Xiaobo Xia1, Li Shen2, Bo Han3, Mingming Gong4, Chen Gong5, , Tongliang Liu1, 1Syndey AI Centre, The University of Sydney; 2JD Explore Academy; 3Department of Computer Science, Hong Kong Baptist University; 4The University of Melbourne; 5Key Laboratory of Intelligent Perception and Systems for High-Dimensional Information of Ministry of Education, School of Computer Science and Engineering, Nanjing University of Science and Technology |
| Pseudocode | Yes | Algorithm 1 Training process of HOOD |
| Open Source Code | No | The paper does not contain an explicit statement about the release of source code or a link to a code repository for the methodology described. |
| Open Datasets | Yes | In OOD detection task, we use SVHN (Netzer et al., 2011) and CIFAR10 (Krizhevsky et al., 2009) as the ID datasets, and use LSUN (Yu et al., 2015), DTD (Cimpoi et al., 2014), CUB (Wah et al., 2011), Flowers (Nilsback & Zisserman, 2006), Caltech (Griffin et al., 2007), and Dogs (Khosla et al., 2011) datasets as the OOD datasets... In open-set SSL task, we follow (Guo et al., 2020) to construct our training dataset using two benchmark datasets CIFAR10 and CIFAR100 (Krizhevsky et al., 2009)... In open-set DA task, we follow (Saito et al., 2018) to validate on two DA benchmark datasets Office (Saenko et al., 2010) and Vis DA (Peng et al., 2018). |
| Dataset Splits | Yes | OOD detection. The labeled set is used for training, and the unlabeled set is used as a test set... we only sample 100 labeled data and 20,000 unlabeled data from each class and conduct semi-supervised training, then we test the trained model on the unlabeled OOD dataset. ... Open-set SSL. The labeled set Dl and unlabeled set Du are both used for training... The constructed dataset has 20,000 randomly sampled unlabeled data and a varied number of labeled data. Here the number of labeled data is set to 50, 100, and 400 per class in both CIFAR10 and CIFAR100. |
| Hardware Specification | Yes | Each trial of our experiments is conducted in one single NVIDIA 3090 GPU. |
| Software Dependencies | No | The paper mentions specific network architectures (Wide Res Net-28-2, Res Net50), data augmentation techniques (Rand Augment), and optimizers (SGD), but does not provide version numbers for any software libraries or frameworks (e.g., PyTorch, TensorFlow, scikit-learn). |
| Experiment Setup | Yes | In experiments, we choose Wide Res Net-28-2 (Zagoruyko & Komodakis, 2016) for OOD detection and Open-set SSL tasks, and follow (You et al., 2020; Cao et al., 2019) to utilize Res Net50 pre-trained on Imagenet (Russakovsky et al., 2015) for Open-set DA. For implementing HOOD, we randomly choose 4 augmentation methods from the transformation pool in Rand Augment (Cubuk et al., 2020), to simulate different styles. The pre-training iteration Augmentation_Iter is set to 100,000, and the perturbation magnitude ϵ = 0.03, following (Volpi et al., 2018) in all experiments. ... The employed Stochastic Gradient Descent (SGD) optimizer starts with an initial learning rate 3e 2 which is decayed by following the cosine function cos(const current_iteration / 500,000 ) without warm-up, in which const is constant, we follow (Saito et al., 2021) to set it as 7/16π. The momentum factor is set to 0.9, which is also the same as (Saito et al., 2021). For choosing the pseudo labels of unlabeled data, we follow (Lee, 2013) to set the pseudo label threshold as 0.95. |