Learning Bounds for Open-Set Learning

By Zhen Fang, Jie Lu, Anjin Liu, Feng Liu, Guangquan Zhang. 2021

68 citations recorded by the source corpus.

Traditional supervised learning aims to train a classifier in the closed-set world, where training and test samples share the same label space. In this paper, we target a more challenging and realistic setting: open-set learning (OSL), where there exist test samples from the classes that are unseen during training. Although researchers have designed many methods from the algorithmic perspectives, there are few methods that provide generalization guarantees on their ability to achieve consistent performance on different training samples drawn from the same distribution. Motivated by the transfer learning and probably approximate correct (PAC) theory, we make a bold attempt to study OSL by proving its generalization error-given training samples with size n, the estimation error will get close to order O_p(1/\sqrt{n}). This is the first study to provide a generalization bound for OSL, which we do by theoretically investigating the risk of the target classifier on unknown classes. According to our theory, a novel algorithm, called auxiliary open-set risk (AOSR) is proposed to address the OSL problem. Experiments verify the efficacy of AOSR. The code is available at github.com/Anjin-Liu/Openset_Learning_AOSR.

Subfields: Artificial Intelligence

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Zhen Fang
2. Bo Han
3. Jiahua Dong
4. Feng Liu
5. Jie Lu
6. Gan Sun
7. Guangyao Chen
8. Yonghong Tian

Citations in the corpus

• Federated Class-Incremental Learning (2022, methodology): “We use three datasets: CIFAR-100 [13, 21], ImageNet-Subset [7], and TinyImageNet [35] in our experiments.”
• Is Out-of-Distribution Detection Learnable? (2022, background): “To investigate open-set learning (OSL) without accessing the test data during training, [24] proposes and investigates the almost agnostic PAC learnability for OSL.”
• Where and How to Transfer: Knowledge Aggregation-Induced Transferability Perception for Unsupervised Domain Adaptation (2021, background): “After adversarial learning [53] is first introduced by Hoff-man et al. to perform unsupervised domain adaptation task [10], [14], [15], [52], large quantities of competing variants [13], [27], [41], [42], [43], [46] relying on adversarial framework have been developed to explore shared…”
• Generalized Out-of-Distribution Detection: A Survey (2021, unknown): “Therefore, an investigation of the generalization error bound is conducted and proves the existence of a low-error OSL algorithm under certain assumptions (Fang et al., 2021).”
• Adversarial Reciprocal Points Learning for Open Set Recognition (2021, background): “However, the AUROC evaluates only the distinction between known and unknown classes, and does not consider the accuracy of known classes in open set recognition, which has been however hidden by this gold-standard ”fair” metric [6], [45].”