In the past five decades, there has been a growing trend in employing compacted stone columns as a method to reinforce weak cohesive soil by improving bearing capacity, increasing the rate of consolidation, and enhancing the settlement response of foundation soils. The present study presents a predictive model for the cohesion (C) and angle of internal friction (ϕ) between aggregates in cemented stone columns, employing an Artificial Neural Network (ANN) approach. To achieve this, preliminary numerical analyses were performed on stone column samples under triaxial loading conditions using the Discrete Element Method (DEM). In this method Cemented stone column was modeled using a wide variety of inputs that defines the properties of aggregates in the stone column, these include Modulus of elasticity (E), Normal to Shear stiffness ratio (K = kn/ks), Particle radius range (R), Friction coefficient (f), Parallel bond Tensile strength (Pa), Parallel bond Cohesive strength (Pa), Porosity, Plate velocity. The results obtained from the model provide the C and ϕ between particles. A dataset with varying particle properties was subsequently generated. Artificial neural network (ANN) models were then developed using an optimization algorithm known as Bayesian Regularization. Multiple models with different hyperparameters were created, and their performance was evaluated based on three statistical metrics: Root Mean Squared Error (RMSE), R-squared (R2), and Mean Absolute Error (MAE). The Bayesian Regularization algorithm demonstrated strong performance in predicting the cohesion and friction values of particles in cemented stone columns. The final developed model enables the determination of the C and ϕ between particles for any given input values within the dataset's range. Thus eliminating the need for additional modeling in software, simply by inputting the relevant values into the trained ANN model.