Evaluation of the Performances of a Parallel Algorithm to Recognize the Patterns in Relation with the Sequential Variant

摘要

The rhythm of human biological degradation, due to increased obesity, high blood pressure and the main cardiovascular affections, to cancer extension, to the breathing diseases etc., is more and more alarming. A first aim of this paper is the use of some clinical tests [1], based on medical investigations, on patients to prevent the main diseases that threaten the human species. The pattern recognition algorithms are very useful in analyzing the information contained by the tests. The great volume of the entering data (hundreds of recordings) collected from patients and used by these algorithms have an important part in grouping them according to their genetic heritage. A correct clustering of these inheritances cannot be achieved by the sequential algorithms, having a single processor with limited memory, but by the parallel algorithms using a great number of processors, each having its own memory.