Learning a Propagable Graph for Classification under the Scenario with Uncertain Labels
- Mr Karan P. Bhatt
- Ms Ompriya V. Kale
Semi-supervised learning, graph-based learning, propagability, graph harmoniousness, uncertainty, feature extraction, multiclass classification
In this paper, we present the classification of uncertain labels using learning by propagability in graph. The entire leaning methodology is based on the data labels and optimality of feature representation that can create a harmonic system. Here data labels are invariant regarding the propagation on the similarity graph constructed based on the optimal feature representation. Using this idea, we can perform classification. This approach deals with multiclass classification problems. Dealing with uncertain labels is the crucial problem in this approach. Using graph-based semi-supervised learning, we can implement the classification procedure for uncertain labels data sets. Different from previous graph-based methods that are based on discriminative models, our method is essentially a generative model in that the class conditional probabilities are estimated by graph propagation and the class priors are estimated by linear regression.
Mr Karan P. Bhatt, Ms Ompriya V. Kale. "Learning a Propagable Graph for Classification under the Scenario with Uncertain Labels".INTERNATIONAL JOURNAL OF ENGINEERING DEVELOPMENT AND RESEARCH ISSN:2321-9939, Vol.3, Issue 2, pp.715-719, URL :https://rjwave.org/ijedr/papers/IJEDR1502128.pdf
Volume 3 Issue 2
Pages. 715-719