Publications

Abstract : Breast cancer is one of the major causes of mortality rate increase among women, in both developed and under-developed countries. Hormone-dependent cancers are possibly caused by organ chlorine. High-quality images and skilled mammographic interpretation are required to detect breast cancer at an early stage during breast screening. Radiography combined with computer-aided diagnosis (CAD) is the most promising field to diagnose breast cancer and other diseases. Nevertheless, when a cancer-affected area is surgically removed, a few cancer cells remain hidden with the stem cells and begin to grow after a period of time. To avoid this, chemotherapy is advisable, but tends to weaken the patient, even with the first dose. Hence, early detection and treatment are the key features. Neural networks have a number of important properties that make the screening possible. This chapter describes prototypical joining neural networks via logic technical stimulated programming in a novel method for solving multi-hop reasoning tasks over natural language. Specifically, we propose to use a human aided system that we extend to utilize a similarity function over pertained word encoders. The feature of resemble function is fine-tuned via back propagation. This points the way toward a system that can involve rule-based reasoning using natural language, and stimulate domain-specific rules in detection training data.