Liquefaction is a kind of ground failure that occurs during strong seismic shaking. Seismic liquefaction occurs in loose sandy ground that is saturated with water. When pore water pressure rises during shaking, the effective stress decreases with time. The shear strength and shear modulus of the sand decrease as the effective stress decreases. Thus, the sandy ground becomes softer with time. Though the liquefaction had often been witnessed from olden days and well documented in archives, it was only after the year 1964, when the Niigata earthquake in Japan and the Alaska earthquake in the USA caused devastations by the extensive soil liquefaction. Since then, quite a few liquefaction cases have been observed during several earthquakes, and the significance of its effect on geotechnical structures and structural damage has increasingly been recognised.
An attempt has been made as part of the review study to compile the available literature on the liquefaction assessment methods using the machine learning (ML) techniques. The review also consists of several conventional approaches. Initially, the assessment method was started in 1971 by using the stress-based concept and later on, the conventional methods were developed. But, these methods are semi-empirical and region-specific. Of late, several researchers have attempted to solve the liquefaction assessment problem by using several soft-computing techniques by constructing the data-driven models. However, these data-driven models are not capable of addressing all the issues associated with the conventional methods. Therefore, there is a need for further studies that combine the updated liquefaction database with the reliable and computationally efficient soft-computing algorithms after making suitable modifications to the existing models. In the present study, an attempt is made to develop the liquefaction potential assessment procedure for non-cohesive saturated soils using the ML techniques.