Explication of a logistic regression driven hypothesis to strengthen derivative approach driven classification for medical diagnosis

Supervised learning based classification depends on learning from previously known data set. Here, these data sets governs training for classification of new data points. This training is mainly driven by two fundamental approaches. First one is derivative based approach and another centers around heuristics or direct search based methodologies. Both approaches have their pros and cons depending upon realization and formulation of problem. Heuristic based approaches gains an edge over derivative based approaches while solving a non-deterministic problem model in terms of accuracy. Derivative based approaches are faster whose reliability is only limited to deterministic problem model. This paper explicate a logistic regression driven hypothesis to convert non-deterministic neural network model into a deterministic model to support derivative based approaches. Experiments were performed and extensive comparison has been done with other algorithms are tabulated. High accuracy of result supports the use this new method.