Predicting the Accuracy of a Few-Shot Classifier
arXiv:2007.04238
Abstract
In the context of few-shot learning, one cannot measure the generalization ability of a trained classifier using validation sets, due to the small number of labeled samples. In this paper, we are interested in finding alternatives to answer the question: is my classifier generalizing well to previously unseen data? We first analyze the reasons for the variability of generalization performances. We then investigate the case of using transfer-based solutions, and consider three settings: i) supervised where we only have access to a few labeled samples, ii) semi-supervised where we have access to both a few labeled samples and a set of unlabeled samples and iii) unsupervised where we only have access to unlabeled samples. For each setting, we propose reasonable measures that we empirically demonstrate to be correlated with the generalization ability of considered classifiers. We also show that these simple measures can be used to predict generalization up to a certain confidence. We conduct our experiments on standard few-shot vision datasets.
References in corpus (10)
- Fast unfolding of communities in large networks
- Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
- Prototypical Networks for Few-shot Learning
- The Emerging Field of Signal Processing on Graphs: Extending High-Dimensional Data Analysis to Networks and Other Irregular Domains
- Manifold Mixup: Better Representations by Interpolating Hidden States
- Invariant Risk Minimization
- SimpleShot: Revisiting Nearest-Neighbor Classification for Few-Shot Learning
- Learning Disentangled Representations for Recommendation
- A Theory of Usable Information Under Computational Constraints
- Graph-based Interpolation of Feature Vectors for Accurate Few-Shot Classification