Abstract
Finite Gamma mixture models have proved to be flexible and can take prior information into account to improve generalization capability, which make them interesting for several machine learning and data mining applications. In this study, an efficient Gamma mixture model-based approach for proportional vector clustering is proposed. In particular, a sophisticated entropy-based variational algorithm is developed to learn the model and optimize its complexity simultaneously. Moreover, a component-splitting principle is investigated, here, to handle the problem of model selection and to prevent over-fitting, which is an added advantage, as it is done within the variational framework. The performance and merits of the proposed framework are evaluated on multiple, real-challenging applications including dynamic textures clustering, objects categorization and human gesture recognition.
Highlights
The amount of multimedia data available in the world is increasing at an astounding rate
We address the problem of model selection in finite Gamma mixtures using a component-splitting approach, which has been successfully applied for the case of Gaussian and Dirichlet mixtures in [9,24]
Dynamic textures (DT) plays a substantial role in many applications and the modeling of DT has been addressed by many researchers to solve different problems, including motion synthesis or retrieval, motion classification, recognition and segmentation [32]
Summary
The amount of multimedia data available in the world is increasing at an astounding rate. Analyzing these heterogeneous and multimodal data automatically and extracting knowledge instantly through machine learning techniques has become a substantial problem for various decision-making fields. Clustering techniques aim at grouping items having the same features and this process can assist businesses, for instance, in identifying separate groups within their customer base. These problems have great practical applications in multimedia information retrieval, machine learning, data security and pattern recognition, to name a few. How to build an accurate model of high-dimensional data in a compact and reliable way is one of the most difficult issues
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.