Full Program »
Articial Neural Networks Interpretation for Breast Cancer Diagnosis
Breast Cancer (BC) is the most common type of cancer among women. Thankfully early detection and treatments improvements helped decrease its number of deaths. Data Mining techniques (DM), which discover hidden and potentially useful patterns from data, partic- ularly for breast cancer diagnosis, are witnessing a new era, where the main objective is no longer replacing humans or just assisting them in their tasks but enhancing and augmenting their capabilities and this is where interpretability comes into play. This paper aims to investigate the Local Interpretable Model-agnostic Explanations (LIME) technique to interpret a Multilayer perceptron (MLP) trained on the Wisconsin Original Data-set. The results show that LIME explanations are a sort of real-time interpretations that help understanding how the constructed neural network "thinks" and thus can increase trust and help oncologists, as the domain experts, learn new patterns.