Visual Transformer for Task-aware Active Learning

If you find this code useful, please cite: https://arxiv.org/abs/2106.03801

Requirements:

python 3.6+ torch 1.0+

pip libraries: tqdm, sklearn, scipy, math

Run:

Please have a look over the config file before running. Also, check the args of the code. CUDA-GPU implementation. Different random seed might produce different results.

Active Learning methods implemented:

Active Learning for Convolutional Neural Networks: A Core-Set Approach: https://arxiv.org/pdf/1708.00489.pdf [CoreSet]

Learning Loss for Active Learning: https://arxiv.org/pdf/1905.03677.pdf [lloss]

Variational Adversial Active Learning: https://arxiv.org/pdf/1904.00370.pdf [VAAL]

Contextual Diversity for Active Learning: https://arxiv.org/abs/2008.05723 [CDAL]

Contact

If there are any questions or RaFD synthetica dataset request, feel free to send a message at: [email protected]