Affiliations: InfraSense Labs, Department of Civil and Environmental Engineering, Imperial College London, London, UK | Fluing, Institute for Multidisciplinary Mathematics, Universitat Politècnica de València, Valencia, Spain
Note: [] Corresponding author: Manuel Herrera, InfraSense Labs, Department of Civil and Environmental Engineering, South Kensington Campus, Imperial College London, SW7 2BU London, UK. E-mail: [email protected]
Abstract: In many real-world applications we have at our disposal a limited number of inputs in a theoretical database with full information, and another part of experimental data with incomplete knowledge for some of their features. These are cases that can be addressed by a label propagation process. It is a widely studied approach that may acquire complexity if new constraints in the new unlabeled data that should be taken into account are found. This is the case of the membership to a group or community in graphs. The proposal is to add the Laplacian matrix as well as another different similarity measures (may be not found in the original database) in the label propagation. A kernel embedding process together with a simple label propagation algorithm will be the main tools to achieve this approach by the use of all types of available information. In order to test the functionality of this new proposal, this work introduces an experimental study of biofilm development in drinking water pipes. Then, a label propagation through pipes belonging to a complete water supply network is approached. These pipes have their own properties depending on their network location and environmental co-variables. As a result, the proposal is a suitable and efficient way to deal with practical data, based on previous theoretical studies by the constrained label propagation process introduced.
Keywords: Label propagation, semi-supervised learning, kernel methods, water supply networks
