Deep Metric Transfer for Label Propagation with Limited Annotated Data

By Bin Liu, Zhirong Wu, Han Hu, Stephen Lin. 2018

68 citations recorded by the source corpus.

We study object recognition under the constraint that each object class is only represented by very few observations. Semi-supervised learning, transfer learning, and few-shot recognition all concern with achieving fast generalization with few labeled data. In this paper, we propose a generic framework that utilize unlabeled data to aid generalization for all three tasks. Our approach is to create much more training data through label propagation from the few labeled examples to a vast collection of unannotated images. The main contribution of the paper is that we show such a label propagation scheme can be highly effective when the similarity metric used for propagation is transferred from other related domains. We test various combinations of supervised and unsupervised metric learning methods with various label propagation algorithms. We find that our framework is very generic without being sensitive to any specific techniques. By taking advantage of unlabeled data in this way, we achieve significant improvements on all three tasks. Code is availble at http://github.com/Microsoft/metric-transfer.pytorch.

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

S4L: Self-Supervised Semi-Supervised Learning (2019, methodology): “Semi-supervised learning algorithms are typically [28, 22, 2, 39, 1, 21] evaluated on small-scale datasets such as CIFAR-10 [17] and SVHN [23].”
MixMatch: A Holistic Approach to Semi-Supervised Learning (2019, neutral): “, “transductive” models [14, 22, 21], graph-based methods [49, 4, 29], generative modeling [3, 27, 41, 9, 17, 23, 38, 34, 42], etc.”
ReMixMatch: Semi-Supervised Learning with Distribution Alignment and Augmentation Anchoring (2019, background): “…research on “transductive” or graph-based semi-supervised learning techniques which leverage the idea that unlabeled datapoints should be assigned the label of a labeled datapoint if they are sufﬁciently similar (Gammerman et al., 1998; Joachims, 2003; 1999; Bengio et al., 2006; Liu et al., 2018).”
In Defense of Pseudo-Labeling: An Uncertainty-Aware Pseudo-label Selection Framework for Semi-Supervised Learning (2021, background): “…for interested readers which includes: “transductive” models (Gammerman et al., 1998; Joachims, 1999; 2003), graph-based methods (Zhu et al., 2003; Bengio et al., 2006; Liu et al., 2019), generative modeling (Belkin & Niyogi, 2002; Lasserre et al., 2006; Kingma et al., 2014; Pu et al., 2016).”
RETRIEVE: Coreset Selection for Efficient and Robust Semi-Supervised Learning (2021, unknown): “Due to space constraints, we do not talk about generative [50, 48, 26, 11, 19, 30, 3] and graph-based [62, 33] methods for SSL in this work.”
Semi-Supervised Semantic Segmentation With Cross-Consistency Training (2020, background): “Recently, many efforts have been made to adapt classic SSL methods to deep learning, such as pseudo labeling [31], entropy minimization [19] and graph based methods [35, 27] in order to overcome this weakness.”
CAPE: Encoding Relative Positions with Continuous Augmented Positional Embeddings (2021, methodology): “We reproduce results from [31, 32] by training a sinpos -based model with 6L-6L, 18L-18L for DE and 6L-6L for FR encoder-decoder layers with ADMIN.”