A framework for modeling positive class expansion with single snapshot

Yu, Yang; Zhou, Zhi‐Hua

doi:10.1007/s10115-009-0238-7

Cited by 5 publications

(3 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Looking past the task of predicting classes and only considering instances and attributes we can view ZSL as a special case of transfer learning problem. Here attribute distributions differ from seen to unseen classes and we propose to view this as a domain adaptation under target shift [27,42,26], where attribute marginals for training set P tr φ and that for test set P te φ are different but conditionals P X|φ remains the same. Correcting for target shift one needs to have P te φ with training data.…”

Section: Main Ideamentioning

confidence: 99%

Addressing target shift in zero-shot learning using grouped adversarial learning

Chemmengath¹,

Paul²,

Bharadwaj³

et al. 2020

Preprint

View full text Add to dashboard Cite

In this paper, we present a new paradigm to zero-shot learning (ZSL) that is trained by utilizing additional information (such as attribute-class mapping) for specific set of unseen classes. We conjecture that such additional information about unseen classes is more readily available than unsupervised image sets. Further, on close examination of the underlying attribute predictors of popular ZSL algorithms, we find that they often leverage attribute correlations to make predictions. While, attribute correlations that remain intact in the unseen classes (test) benefits the prediction of difficult attributes, change in correlations can have an adverse effect on ZSL performance. For example, detecting an attribute 'brown' may be the same as detecting 'fur' over an animals image dataset captured in the tropics. However, such a model might fail on unseen images of Arctic animals. To address this effect, termed target-shift in ZSL, we utilize our proposed framework to design grouped adversarial learning. We introduce grouping of attributes to enable the model to continue to benefit from useful correlations, while restricting cross-group correlations that may be harmful for generalization. Our analysis shows that it is possible to not only constrain the model from leveraging unwanted correlations, but also adjust them to specific test setting using only the additional information (the already available attribute-class mapping). We show empirical results for zero-shot predictions on standard benchmark datasets, namely, aPY, AwA2, SUN and CUB datasets. We further introduce to the research community, a new experimental train-test split that maximizes target-shift to further study its effects.

show abstract

Section: Main Ideamentioning

confidence: 99%

Addressing target shift in zero-shot learning using grouped adversarial learning

Chemmengath¹,

Paul²,

Bharadwaj³

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…In order to solve it, several approaches were proposed. In (Lin et al , 2002), authors assumed that the shift in the target distribution was known a priori, while in (Yu & Zhou, 2008), partial knowledge of the target shift was supposed to be available. In both cases, the assumption of prior knowledge about the class proportions in the target domain seems quite restrictive.…”

Section: Introductionmentioning

confidence: 99%

Optimal Transport for Multi-source Domain Adaptation under Target Shift

Redko¹,

Courty²,

Flamary³

et al. 2018

Preprint

View full text Add to dashboard Cite

In this paper, we tackle the problem of reducing discrepancies between multiple domains, i.e. multi-source domain adaptation, and consider it under the target shift assumption: in all domains we aim to solve a classification problem with the same output classes, but with different labels proportions. This problem, generally ignored in the vast majority of domain adaptation papers, is nevertheless critical in real-world applications, and we theoretically show its impact on the success of the adaptation. Our proposed method is based on optimal transport, a theory that has been successfully used to tackle adaptation problems in machine learning. The introduced approach, Joint Class Proportion and Optimal Transport (JCPOT), performs multi-source adaptation and target shift correction simultaneously by learning the class probabilities of the unlabeled target sample and the coupling allowing to align two (or more) probability distributions. Experiments on both synthetic and real-world data (satellite image pixel classification) task show the superiority of the proposed method over the state-of-the-art.

show abstract

“…It is possible to improve the video retrieval level with a large number of semantic concepts. Many methods have been addressed by [3][4][5] on semantic learning and video classification. However, the key barrier for robot learning is lack of training dataset, both positive dataset and negative dataset as indicated by [10][11].…”

Section: Introductionmentioning

confidence: 99%

Unstructured robot perception through Internet semantic concept learning

Wang

Chen

Yuan

2014

2014 International Conference on Multisensor Fusion and Information Integration for Intelligent Systems (MFI)

View full text Add to dashboard Cite

Intelligent robot is one of the most important ongoing technologies both in industry and social life. Smart perception is the key technology for intelligent robots. Lack of training data, there has been many barriers for intelligent robot to learn the unstructured environment. In this paper, an automatic data mining method for smart robots to learn semantic concepts from videos crawled to known Internet video/image websites (e.g. video-Baidu, Bing, Youku) is presented. An updated novel Internet video-mining method is addressed. An automatic graph model generator is addressed as well as the weight assignment for concepts-relationship learning based on known ontology and an automated video source discovery method in concepts detection from the massive Internet videos is proposed. Experimental results with Tera-bytes level videos show that the method is effective and efficient to solve the smart perception for intelligent robots.

show abstract

A framework for modeling positive class expansion with single snapshot

Cited by 5 publications

References 19 publications

Addressing target shift in zero-shot learning using grouped adversarial learning

Addressing target shift in zero-shot learning using grouped adversarial learning

Optimal Transport for Multi-source Domain Adaptation under Target Shift

Unstructured robot perception through Internet semantic concept learning

Contact Info

Product

Resources

About