Learning from one example through shared densities on transforms

By Erik G. Miller, Nicholas E. Matsakis, Paul A. Viola. 2000

479 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Media Technology, Representation & Self-Supervised Learning, Meta & Continual Learning

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

In Computer Vision and Pattern Recognition

In Media Technology

In Representation & Self-Supervised Learning

In Meta & Continual Learning

Citations in the corpus

CrossTransformers: spatially-aware few-shot transfer (2020, background): “Few-shot learning [25, 35, 46, 53] has recently been primarily addressed in the meta-learning framework [57, 71, 80], where a model learns an update rule for the parameters of a base-learner model [6, 7, 72] through a sequence of training episodes [79, 88].”
Meta-learning with differentiable closed-form solvers (2018, background): “Such a scenario, in which just one or a handful of training examples is provided, is referred to as one-shot or few-shot learning (Miller et al., 2000; Fei-Fei et al., 2006; Lake et al., 2015; Hariharan & Girshick, 2017) and has recently seen a tremendous surge in interest within the machine…”
Prototypical Networks for Few-shot Learning (2017, background): “Few-shot classification [22, 18, 15] is a task in which a classifier must be adapted to accommodate new classes not seen in training, given only a few examples of each of these classes.”
Learning Generative Visual Models from Few Training Examples: An Incremental Bayesian Approach Tested on 101 Object Categories (2004, background): “It is important to point out that the idea of using prior information from unrelated categories was proposed independently by Miller et al. [15].”
Low-Shot Learning from Imaginary Data (2018, background): “If one has a probability density on transformations, then one can generate additional examples for a novel class by applying sampled transformations to the provided examples [20, 5, 13].”
Zero-Shot Learning by Convex Combination of Semantic Embeddings (2013, background): “Zero-shot learning is closely related to one-shot learning [13, 5, 1, 8], where the goal is to learn object classiﬁers based on a few labeled training exemplars.”
Domain Adaptation for Visual Applications: A Comprehensive Survey (2017, methodology): “used for the supervised learning task of interest. On the other hand, the relation between DA and domain generalization [151, 254, 97, 94, 155], multi-task learning [26, 76, 178] or few-shot learning [149, 80] is much stronger. 39 Code available at https://github.com/jhoffman/lsda/zipball/master 32 ”
Generalizing from a Few Examples (2019, methodology): “transformation. The learned transformation itself is the prior knowledge in E, while neither its training samples nor learning procedure are needed for the current FSL task. The earliest paper on FSL [76] uses exactly this strategy to solve FSL image classification. A set of geometric transformation”
Diverse Few-Shot Text Classification with Multiple Metrics (2018, background): “Few-shot learning (FSL) (Miller et al., 2000; Li et al., 2006; Lake et al., 2015) aims to learn classifiers from few examples per class.”
Infinite Mixture Prototypes for Few-Shot Learning (2019, background): “Few-shot classiﬁcation is the problem of learning to recognize new classes from only a few examples of each class (Lake et al., 2015; Fei-Fei et al., 2006; Miller et al., 2000).”
CG-BERT: Conditional Text Generation with BERT for Generalized Few-shot Intent Detection (2020, background): “Few Shot Learning (Miller et al., 2000; Li et al., 2006) is a task that aims to learn classiﬁers for new classes with only a few training examples per class.”
One-shot learning by inverting a compositional causal process (2013, methodology): “The full HBPL model is compared to a transformation-based approach that models the variance in image tokens as just global scales, translations, and blur, which relates to congealing models [23].”
Cross-generalization: learning novel classes from a single example by feature replacement (2005, background): “Similar bias towards the common classes is present in [10].”
Learning to share visual appearance for multiclass object detection (2011, methodology): “These methods can be applied for object classes with well defined sets of transformations such as letters [20, 28], or faces [28].”
Learning with Imprinted Weights (2017, background): “rovides better generalization than using nearest-neighbor instance embeddings. 1. Introduction Human vision can immediately recognize new categories after a person is shown just one or a few examples [10,8]. For instance, humans can recognize a new face from a photo of an unknown person and new obje”
Learning to Align from Scratch (2012, methodology): “This framework of iteratively transforming images to reduce the entropy of the set is known as congealing [6], and was originally applied to specific types of images such as binary handwritten characters and magnetic resonance image volumes [7] [8] [9].”
PARN: Position-Aware Relation Networks for Few-Shot Learning (2019, background): “Fewshot learning [11, 20, 15] aims to acquire some transferable knowledge like humans, where a classifier is able to generalize to new classes when given only one or a few labeled examples of each class, i.”
Pose pooling kernels for sub-category recognition (2012, background): “While attribute-based models are suitable for addressing the one-shot learning problem (previously considered in [13, 14, 15, 24] among others), they typically focus on relatively coarse grained attributes.”