A Bayesian approach to unsupervised one-shot learning of object categories

By F. Li, Rob Fergus, Pietro Perona. 2003

214 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

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

Citations in the corpus

Probabilistic Model-Agnostic Meta-Learning (2018, methodology): “Hierarchical Bayesian models are a long-standing approach for few-shot learning that naturally allow for the ability to reason about uncertainty over functions [39, 7, 25, 43, 12, 4, 41].”
Lifelong Generative Modeling (2017, background): “latent-variable posterior approximation: a requirement in many learning scenarios such as clustering [18], compression [19] and unsupervised representation learning [20].”
Attentive Recurrent Comparators (2017, methodology): “A bayesian framework for one shot visual recognition was presented in Fe-Fei et al. (2003).”
Towards Intersectionality in Machine Learning: Including More Identities, Handling Underrepresentation, and Performing Evaluation (2022, methodology): “…constant within-group heterogeneity, whether that be through known variances across groups for a variational Bayesian approach to one-shot learning [37], or the homoscedasticity assumption (i.e., that all groups have the same variance) for methods like linear regression and linear discriminant…”
1 On what it means to see , and what we can do about it (2008, unknown): “Such integration is at the core of the most promising current approaches to object and scene vision (Fei-Fei et al., 2003; Freeman, 1993; Torralba et al., 2003), including the explicitly generative “analysis by synthesis” methods (Yuille and Kersten, 2006).”
Write a Classifier: Zero-Shot Learning Using Purely Textual Descriptions (2013, background): “Motivated by the practical need to learn visual classifiers of rare categories, researchers have explored approaches for learning from a single image (one-shot learning [18, 9, 11, 2]) or even from no images (zero-shot learning).”
Measuring and Predicting Importance of Objects in Our Visual World (2007, background): “A state-of-the-art algorithm can classify single object Caltech-101 images, producing one descriptive word per image [9].”
Robust Object Detection with Interleaved Categorization and Segmentation (2008, background): “…have become a popular tool for object categorization recently, and many approaches use variations of this theme (Burl et al. 1998; Weber et al. 2000; Fergus et al. 2003; Li et al. 2003; Agarwal et al. 2004; Borenstein and Ullman 2002; Ullman et al. 2002; Felzen-szwalb and Huttenlocher 2005).”
GOT-10k: A Large High-Diversity Benchmark for Generic Object Tracking in the Wild (2018, methodology): “Unlike existing datasets, GOT-10k follows the standard experimental setting of oneshot learning [35], [36] that the object classes between training and testing videos are non-overlapped, while the support set [35] corresponds to the ”one-shot example” for each testing video.”
Weakly Supervised Localization and Learning with Generic Knowledge (2012, unknown): “Pascal07 -6x2 [Everingham et al, 2007].”
Write a Classifier: Predicting Visual Classifiers from Unstructured Text (2015, background): “Motivated by the practical need to learn visual classifiers of rare categories, researchers have explored approaches for learning from a single image (oneshot learning [6], [7], [16], [17]) or even from no images (zero-shot learning).”
A visual Turing test for computer vision systems (2014, background): “…11, 12]; in discovering latent structure by training multi-layer networks with large amounts of unsupervised data [13]; and in parts-based statistical learning and modeling techniques [14, 15, 16], especially combining discriminative part detectors with simple models of arrangements of parts [17].”
Error-Driven Incremental Learning in Deep Convolutional Neural Network for Large-Scale Image Classification (2014, background): “The second is when there are existing classiﬁers that are related to the new classes to be learned [1, 7, 8, 9, 15, 16, 21, 22].”