Neural networks, computational systems conceptually modeled after the brain, are an increasingly effective tool in modern data analysis. One challenge facing scientists using neural networks is the nature of their data. Different types of data require the use of different neural networks. Michela Taufer, Dongarra Professor of electrical engineering and computer science, hopes her JDRD project will make identifying the correct neural networks easier.
Humans have always found ways to complete tasks more quickly and effectively. Beginning around the 1980’s factories began to use more and more automation in order to generate products more quickly. The volume and pace of production outstripped human capability so humans turned to computerized operation. Neural networks do a similar job for data analysis. Continue reading