Application of Machine Learning Algorithms in the Mitigation Phase of Disaster Management: A Review

Initial Research Overview

A methodical procedure was devised for a comprehensive search of relevant literature. A set of keywords pertaining to the field were selected and classified into primary and secondary, and were then entered into the Web of Science search tool. The search produced a total of 623 research papers focusing on a variety of topics across all stages of disaster management.

Selecting Research Papers Dealing With Mitigation

Disaster mitigation is commonly defined as reducing or eliminating the cause, impact and after-effect of disasters through actions taken prior to its occurrence. Based on this definition and by looking at examples of mitigation activities from past disasters, a set of keywords and phrases pertaining to mitigation were listed and these were used to separate the papers dealing with mitigation.

Figure 1 contains a generic list of such mitigation related terms. This procedure yielded 68 papers, which were further filtered down to 16 recent papers employing ML Techniques either individually or in combination with other techniques that include but are not limited to optimization algorithms, Geographic Information Systems (GIS) and boosting algorithms. These papers reveal the current arsenal of ML Techniques being used in the mitigation phase of various types of geophysical, meteorological, hydrological, climatological and technological disasters.

Figure 1.

Generic list of mitigation related terms