Colloquium Computer Science - Andreas Neocleous University of Cyprus

Title: Machine learning and signal processing for the identification of radiocarbon (Δ 14 C) anomalies in the atmosphere, over time

Abstract:

Rapid increment in the concentration of the radiocarbon in the atmosphere (Δ 14 C) has been identified in the years 3372 - 3371 BCE, 665 - 660 BCE, 774 - 775 CE and 993 - 994 CE (Miyake events) using annual measurements on known-age tree-rings. It is known that such events may cause health complications, such as cancer, under some circumstances. Furthermore, it can cause failure in communication systems (Carrington event 1859 CE) and thus are important to be identified in the historical data and to be able to predict them.

In this colloquium, several intelligent computational methods will be discussed on how similar Miyake events can be identified in the past. It will be shown how several state-of-the-art pattern matching techniques as well as feature representations, a procedure that typically is used in machine learning and classification can be used for the task at hand.