SaliencyMix: A Saliency Guided Data Augmentation Strategy for Better Regularization
Authors: A F M Shahab Uddin, Mst. Sirazam Monira, Wheemyung Shin, TaeChoong Chung, Sung-Ho Bae
ICLR 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We present extensive experiments on various standard CNN architectures, benchmark datasets, and multiple tasks, to evaluate the proposed method. In summary, Saliency Mix has obtained the new best known top-1 error of 2.76% and 16.56% for Wide Res Net (Zagoruyko & Komodakis, 2016) on CIFAR-10 and CIFAR-100 (Krizhevsky, 2012), respectively. Also, on Image Net (Olga et al., 2015) classification problem, Saliency Mix has achieved the best known top-1 and top-5 error of 21.26% and 5.76% for Res Net-50 and 20.09% and 5.15% for Res Net-101 (He et al., 2016). |
| Researcher Affiliation | Academia | A. F. M. Shahab Uddin uddin@khu.ac.kr Mst. Sirazam Monira monira@khu.ac.kr Wheemyung Shin wheemi@khu.ac.kr Tae Choong Chung tcchung@khu.ac.kr Sung-Ho Bae shbae@khu.ac.kr Department of Computer Science & Engineering, Kyung Hee University, South Korea. |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Source code is available at https://github.com/SaliencyMix/SaliencyMix. |
| Open Datasets | Yes | We present extensive experiments on various standard CNN architectures, benchmark datasets, and multiple tasks, to evaluate the proposed method...Saliency Mix has obtained the new best known top-1 error of 2.76% and 16.56% for Wide Res Net (Zagoruyko & Komodakis, 2016) on CIFAR-10 and CIFAR-100 (Krizhevsky, 2012), respectively. Also, on Image Net (Olga et al., 2015) classification problem...Moreover, Saliency Mix trained model has proved to be more robust against adversarial attack and improves the top-1 accuracy by 1.96% on adversarially perturbed Image Net validation set. All of these results clearly indicate the effectiveness of the proposed Saliency Mix data augmentation strategy to enhance the model performance and robustness. |
| Dataset Splits | Yes | Image Net (Olga et al., 2015) contains 1.2 million training images and 50, 000 validation images of 1000 classes. |
| Hardware Specification | Yes | All experiments were performed on Py Torch platform with four NVIDIA Ge Force RTX 2080 Ti GPUs. |
| Software Dependencies | No | The paper mentions "Py Torch platform" but does not specify a version number for PyTorch or any other software dependencies with their versions. |
| Experiment Setup | Yes | We train the networks for 200 epochs with a batch size of 256 using stochastic gradient descent (SGD), Nesterov momentum of 0.9, and weight decay of 5e-4. The initial learning rate was 0.1 and decreased by a factor of 0.2 after each of the 60, 120, and 160 epochs. The images are normalized using per-channel mean and standard deviation. |