Mining is one of the main solid waste generator sectors around the world. Due to increased environmental concerns and recently introduced United Nation Sustainable development goals (UNSDGs) mining industry has been looking for the solutions for tailing management. Therefore, the number of underground mines using cemented paste backfill (CPB) to store tailing in underground rather than surface tailing dams has been increasing. The main expectations from CPB are the strength that needs to be gained time determined by my planning engineers to provide safe working conditions and workability to be able to pipe the paste to any desired location in the mine.  The main design parameters required for obtaining CPB with desired properties are cement to tailing ratio and solid content. Traditionally these are obtained by time and resource consuming trial-and-error process. Therefore, many of research articles published recently focusing on the model development for the prediction of strength and workability based on these key parameters to save time and resources. However, there are limited number of research, focusing on mathematical optimisation using constructed models. The purpose of this paper is to develop an algorithm giving the best possible design parameters to minimise the amount of cement while achieving multiple objectives such as strength in a desired time and workability considering different conditions such as physical properties of tailings.

cemented paste backfill; machine learning; multiple objectives optimization