Incremental Transductive Learning Approaches to Schistosomiasis Vector Classification

The key issues faced pertaining to collection of epidemic disease data for research and analysis purposes, is that it is often a time consuming and expensive process. This results in availability of sparse sample data from which we aim to develop prediction models. To address this sparse sample data issue, the research carried out in this paper presents novel Incremental Transductive methods. These are used to circumvent the data collection process by applying previously acquired data to provide precise and consistently confident labelling alternatives to the pro- cess of manually retrieving relevant data from areas of interest. We have conducted research and investigated various approaches for semi-supervised machine learning including Bayesian models in terms of reasoning for labelling data. Results in this paper have shown that using the proposed Incremental Transductive methods, we can consistently label instances of data with a class of vector density to a high degree of confidence. By applying the Liberal (LTA) and Strict (STA) Training Approaches, we provide a bespoke labelling and classification process as an alternative to standalone algorithms. All of the methods employed in this paper are components in the process aimed at reducing the proliferation of the Schistosomiasis disease and its effects.