COSTA: Co-Occurrence Statistics for Zero-Shot Classification

By Thomas Mensink, Efstratios Gavves, Cees G. M. Snoek. 2014

246 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, 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 Meta & Continual Learning

Citations in the corpus

Zero-Shot Learning—A Comprehensive Evaluation of the Good, the Bad and the Ugly (2017, background): “[31], [34], [50], [51], [52], [53], [54] which exploit other auxiliary information that reduces this annotation effort.”
Evaluation of output embeddings for fine-grained image classification (2014, background): “Learning to classify in the absence of labeled data (zeroshot learning) [58, 45, 25, 29, 2, 37, 34, 17] is a challenging problem, and achieving better-than-chance performance requires structure in the output space.”
Visual Relationship Detection with Language Priors (2016, background): “There has been a series of work related to improving object detection by leveraging object cooccurrence statistics [9,10,11,12,13,14].”
Don't just listen, use your imagination: Leveraging visual common sense for non-visual tasks (2015, background): “Co-occurrence statistics of visual concepts mined from the web has been shown to not generalize to images [31].”
Generalizing from a Few Examples (2019, background): “nyms) [1, 50], , word embeddings (real-valued vectors which represents words or phrases learned from text corpus) [120, 126] , co-occurrence statistics (hit-count returned by in searching on the web) [74], and knowledge graphs (a graph implicitly encoding the relationships of entities such as classe”
An Empirical Study and Analysis of Generalized Zero-Shot Learning for Object Recognition in the Wild (2016, methodology): “Existing research on ZSL has been almost entirely focusing on this setting [4,10,11,12,13,14,15,8,16,17,18,19,20,21,22,23,24,25,26,27,28].”
Exploiting Feature and Class Relationships in Video Categorization with Regularized Deep Neural Networks (2015, methodology): “Another two recent works [35], [36] utilized the co-occurrence statistics to help video classiﬁcation, where the co-occurrence of classes was used more as a semantic feature representation.”
Zero-shot recognition with unreliable attributes (2014, methodology): “For example, an unseen object’s classifier can be estimated by combining the nearest existing classifiers (trained with images) in the ImageNet hierarchy [23, 14], or by combining classifiers based on label co-occurrences [13].”
Zero-Shot Event Detection by Multimodal Distributional Semantic Embedding of Videos (2015, background): “In practice, however, we think that typical use of event queries under this setting should be similar to text-search, which is based on few words instead that we model their connection to the multimodal content in videos.”
Classifier and Exemplar Synthesis for Zero-Shot Learning (2018, unknown): “Detailed discussions of SynC: COSTA [44] combines pre-trained classifiers of seen classes to construct new classifiers.”
Small Sample Learning in Big Data Era (2018, methodology): “Likewise, Mensink et al. (2014) used co-occurrence statistics learning concept-concept relationships from texts (between the new label and existing ones), as a weight to combine seen classes classifiers.”
Improving Semantic Embedding Consistency by Metric Learning for Zero-Shot Classiffication (2016, background): “The space of class names can also constitute an interesting embedding, such as in the works of [24,25,11,26] which represent images as mixtures of known classes distributions.”
Beyond Instance-Level Image Retrieval: Leveraging Captions to Learn a Global Visual Representation for Semantic Retrieval (2017, background): “Explicit modeling of a scene can be done with attributes [19, 17, 53, 41], object co-occurrences [47], or pairwise relationship between objects [12, 14, 43].”