A.1 Datasets We use two standardized few-shot image classification datasets. Mini-ImageNet: This dataset [58] is a subset of ImageNet [10] and consists of 64 classes for training, 16 for validation, and 20 for testing. There are 600 images per class, with images of size 84 84. Multiple versions of this dataset exist in the literature; we use the version by Ravi and Larochelle [43]. Tiered-ImageNet: A larger subset of ImageNet, Tiered-ImageNet [45] consists of 608 classes split into 351, 97, and 160 for training, validation, and testing, respectively.

CIFAR100 contains 60,000 images in total from 100 classes.

Multiple versions of this dataset exist in the literature; we use the version by Ravi and Larochelle [43]. The original version of the dataset contains43images that are also present in ImageNet. We remove these duplicates to avoid overestimating the transfer capability during evaluation. VGGFlowers: Originally introduced by Nilsback and Zisserman[38], VGGFlowers consists of 102 flower categories with each category containing between40 and 258 images. A.3 Trainingalgorithms For the metric-based family, we use ProtoNet with Euclidean [53] and scaled negative cosine similarity measures [20].

Episodic training is a core ingredient of few-shot learning to train models on tasks with limited labelled data. Despite its success, episodic training remains largely understudied, prompting us to ask the question: what is the best way to sample episodes? In this paper, we first propose a method to approximate episode sampling distributions based on their difficulty. Building on this method, we perform an extensive analysis and find that sampling uniformly over episode difficulty outperforms other sampling schemes, including curriculum and easy-/hard-mining. As the proposed sampling method is algorithm agnostic, we can leverage these insights to improve few-shot learning accuracies across many episodic training algorithms. We demonstrate the efficacy of our method across popular few-shot learning datasets, algorithms, network architectures, and protocols.

Meta-learning has been proposed as a framework to address the challenging few-shot learning setting. The key idea is to leverage a large number of similar few-shot tasks in order to learn how to adapt a base-learner to a new task for which only a few labeled samples are available. As deep neural networks (DNNs) tend to overfit using a few samples only, typical meta-learning models use shallow neural networks, thus limiting its effectiveness. In order to achieve top performance, some recent works tried to use the DNNs pre-trained on large-scale datasets but mostly in straight-forward manners, e.g., (1) taking their weights as a warm start of meta-training, and (2) freezing their convolutional layers as the feature extractor of base-learners. In this paper, we propose a novel approach called meta-transfer learning (MTL) which learns to transfer the weights of a deep NN for few-shot learning tasks. Specifically, meta refers to training multiple tasks, and transfer is achieved by learning scaling and shifting functions of DNN weights for each task. In addition, we introduce the hard task (HT) meta-batch scheme as an effective learning curriculum that further boosts the learning efficiency of MTL. We conduct few-shot learning experiments and report top performance for five-class few-shot recognition tasks on three challenging benchmarks: miniImageNet, tieredImageNet and Fewshot-CIFAR100 (FC100). Extensive comparisons to related works validate that our MTL approach trained with the proposed HT meta-batch scheme achieves top performance. An ablation study also shows that both components contribute to fast convergence and high accuracy.

We propose a method for learning embeddings for few-shot learning that is suitable for use with any number of ways and any number of shots (shot-free). Rather than fixing the class prototypes to be the Euclidean average of sample embeddings, we allow them to live in a higher-dimensional space (embedded class models) and learn the prototypes along with the model parameters. The class representation function is defined implicitly, which allows us to deal with a variable number of shots per each class with a simple constant-size architecture. The class embedding encompasses metric learning, that facilitates adding new classes without crowding the class representation space. Despite being general and not tuned to the benchmark, our approach achieves state-of-the-art performance on the standard few-shot benchmark datasets.