Learning from Small Sample Sets by Combining Unsupervised Meta-Training with CNNs

By Yu-Xiong Wang, Martial Hebert. 2016

80 citations recorded by the source corpus.

Subfields: Representation & Self-Supervised Learning

How researchers see this paper

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

In Representation & Self-Supervised Learning

Citations in the corpus

Growing a Brain: Fine-Tuning by Increasing Model Capacity (2017, background): “Transfer learning is related to both multi-task learning [32, 4, 28, 10, 11, 45, 24, 2] and learning novel categories from few examples [49, 19, 21, 35, 51, 22, 5, 13, 50, 47, 31].”
Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need? (2020, methodology): “tant role in few-shot recognition. 4.4. Embeddings from self-supervised representation learning Using unsupervised learning [61,49,16,51] to improve the generalization of the meta-learning algorithms [58] removes the needs of data annotation. In addition to using embeddings from supervised pre-train”
Low-Shot Learning from Imaginary Data (2018, background): “Some work tries to share features between seen and novel classes [1, 36] or incrementally learn them as new classes are encountered [21].”
Boosting Self-Supervised Learning via Knowledge Transfer (2018, background): “Other related work uses clustering in the feature space of a supervised task to extract labels from unlabeled data for novel categories [36] or building hash functions [35].”
The Devil is in the Tails: Fine-grained Classification in the Wild (2017, background): “Our ﬁndings highlight the importance of continued research in transfer and low shot learning [13, 18, 53, 54] and provide baselines for future work to compare against.”
Generalizing from a Few Examples (2019, background): “mated. 6.3 Applications Existing works mainly deal with computer vision applications, such as character recognition [15, 33, 34, 56, 61, 82, 99, 100, 105, 108, 118, 127, 134] and image classification [34, 61, 82, 93, 105, 108, 115, 118, 120, 127, 131, 132, 136]. This is because visual information is”
An Ensemble of Epoch-Wise Empirical Bayes for Few-Shot Learning (2019, background): “ew-shot learning & meta-learning. Research literature on few-shot learning paradigms exhibits a high diversity from using data augmentation techniques [60, 62] over sharing feature representation [3, 61] to meta-learning [16, 58]. In this paper, we focus on the meta-learning paradigm that levera”
Solving Long-tailed Recognition with Deep Realistic Taxonomic Classifier (2020, methodology): “Inspired by meta-learning [58,59], these methods learn how to leverage knowledge from head classes to improve the generalization of tail classes [60].”
Image Deformation Meta-Networks for One-Shot Learning (2019, background): “Typically, meta-learning [25, 24, 20, 28, 32, 31, 22, 33, 17, 5, 13, 37, 15] aims at training a parametrized mapping from a few training instances to model parameters in simulated one-shot learning scenarios.”
Baby steps towards few-shot learning with multiple semantics (2019, methodology): “These approaches include: (i) semi-supervised approaches using additional unlabeled data [8, 13]; (ii) ﬁne tuning from pre-trained models [27, 52, 53]; (iii) applying domain transfer by borrowing examples from relevant categories [28] or using semantic vocabularies [3, 14]; (iv) rendering synthetic…”
Delta-encoder: an effective sample synthesis method for few-shot object recognition (2018, methodology): “fine tuning from pre-trained models [25, 45, 46]; (3) applying domain transfer by borrowing examples from relevant categories [27] or using semantic vocabularies [2, 12]; (4) rendering synthetic examples [32, 7, 40]; (5) augmenting the training examples [23]; (6) example synthesis using Generative A”
Learning Compositional Representations for Few-Shot Recognition (2018, background): “e categorized into the ones based on metric learning [18,40, 36,38] — training a network to predict whether two images belong to the same category, and the ones built around the idea of meta-learning [12,33,43,44] — training with a loss that explicitly enforces easy adaptation of the weights to new ”
Learning to Model the Tail (2017, methodology): “The most related formulation is that of one/few-shot learning [43, 44, 45, 46, 47, 48, 21, 49, 35, 50, 23, 51, 52, 53, 54, 55, 56].”