Unlocking Transfer Learning for Open-World Few-Shot Recognition
This addresses a critical real-world challenge in AI for categorizing known classes while identifying unknown ones in few-shot settings, representing an incremental advancement in transfer learning for open-world recognition.
The paper tackles the problem of Few-Shot Open-Set Recognition (FSOSR) by proposing a two-stage method combining open-set aware meta-learning and transfer learning, achieving state-of-the-art performance on benchmarks like miniImageNet and tieredImageNet with only a 1.5% increase in training effort.
Few-Shot Open-Set Recognition (FSOSR) targets a critical real-world challenge, aiming to categorize inputs into known categories, termed closed-set classes, while identifying open-set inputs that fall outside these classes. Although transfer learning where a model is tuned to a given few-shot task has become a prominent paradigm in closed-world, we observe that it fails to expand to open-world. To unlock this challenge, we propose a two-stage method which consists of open-set aware meta-learning with open-set free transfer learning. In the open-set aware meta-learning stage, a model is trained to establish a metric space that serves as a beneficial starting point for the subsequent stage. During the open-set free transfer learning stage, the model is further adapted to a specific target task through transfer learning. Additionally, we introduce a strategy to simulate open-set examples by modifying the training dataset or generating pseudo open-set examples. The proposed method achieves state-of-the-art performance on two widely recognized benchmarks, miniImageNet and tieredImageNet, with only a 1.5\% increase in training effort. Our work demonstrates the effectiveness of transfer learning in FSOSR.