SAAL: Sharpness-Aware Active Learning
Authors: Yoon-Yeong Kim, Youngjae Cho, Joonho Jang, Byeonghu Na, Yeongmin Kim, Kyungwoo Song, Wanmo Kang, Il-Chul Moon
ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We conduct experiments on various benchmark datasets for vision-based tasks in image classification, object detection, and domain adaptive semantic segmentation. The experimental results confirm that SAAL outperforms the baselines by selecting instances that have the potentially maximal perturbation on the loss. |
| Researcher Affiliation | Collaboration | 1Agency for Defense Development, AI Autonomy Technology Center (ADD, AIA Center) 2KAIST, South Korea 3Yonsei University, South Korea 4Summary.AI. Correspondence to: Il-Chul Moon <icmoon@kaist.ac.kr>. |
| Pseudocode | Yes | Algorithm 1 Sharpness-Aware Active Learning |
| Open Source Code | Yes | The code is available at https://github.com/ Yoonyeong Kim/SAAL. |
| Open Datasets | Yes | We conduct our experiment on Fashion-MNIST (Fashion) (Xiao et al., 2017), SVHN (Netzer et al., 2011), CIFAR-10, and CIFAR-100 (Krizhevsky et al., 2009). We experiment with PASCAL VOC 2007 and 2012 dataset (Everingham et al., 2010), which contains 5,011 images and 4,952 images with 20 object classes, respectively. We experiment the semantic segmentation from a source domain SYNTHIA (Ros et al., 2016) to a target domain City Scapes (Cordts et al., 2016). |
| Dataset Splits | No | The paper describes using initial labeled datasets and selecting additional instances from unlabeled pools (e.g., 'we construct the initial labeled dataset with 20 instances... and we select 10 additional instances with the highest acquisition score...'). While it discusses training and testing, it does not explicitly specify train/validation/test splits (e.g., percentages, counts) or mention a dedicated 'validation set' for hyperparameter tuning or early stopping during training. |
| Hardware Specification | No | The paper does not specify any hardware used for the experiments, such as specific GPU models, CPU types, or memory configurations. It only describes the training process and models used. |
| Software Dependencies | No | The paper mentions using 'Adam optimizer (Kingma & Ba, 2015)' and 'SAM optimizer (Foret et al., 2020)' with specific learning rates. However, it does not provide specific version numbers for these optimizers or other software dependencies (e.g., Python version, deep learning framework like PyTorch or TensorFlow version). |
| Experiment Setup | Yes | We train the network for 50 epochs after each acquisition step, using Adam optimizer (Kingma & Ba, 2015) with a learning rate of 0.001; or SAM optimizer (Foret et al., 2020) with a learning rate of 0.001 for Fashion, SVHN, CIFAR-10, and 0.1 for CIFAR-100. In Image Net experiment, we follow the above settings besides 500 training epochs after each acquisition step by the Adam optimizer with 0.001 learning rate. We train the model for 300 epochs with a batch size of 32. |