Clay soils can exhibit excessive swelling due to changes in moisture content. The clay swelling threats the long-term stability of structures and foundations, and thus, a correct understanding and prediction of clay swelling properties is essential in many geotechnical projects. We present feed-forward (FF) and cascade-forward (CF) network models trained with Levenberg¿Marquardt (LM) and Bayesian optimisation (BR) algorithms to determine swelling potential of natural and artificial clayey soils. The compiled dataset includes various types of soils covering a wide span of swelling potential, and swelling pressure, ranging from 0.01 to 168.6$\%$, and 25 to 1297.82 $kPa$, respectively. The activity, moisture content, dry unit weight, liquid limit, plastic limit, plasticity index, gravel content, sand content, silt content and clay content (C) were considered as the input parameters of the network model because as are commonly measured during the experimental testing of soil behavior. The developed model showed substantial improvements over previous empirical and semi-empirical correlations in determining the swelling potentials of both natural and artificial soils. We employed feed-forward (FFNN) and cascade-forward neural network (CFNN) models trained with Bayesian regularization (BR) and Levenberg¿Marquardt (LM) algorithms to determine swelling potential of clayey soils over a wide range of conditions. A sufficiently large dataset containing free swell experimental data collected from the literature were used to develop the network models. Moreover, the predicted values were compared with three different empirical correlations. The FFNN-LM achieved the highest overall accuracy among all network models as its predictions showed an acceptable agreement with the experimental data. The model outperforms the tested empirical equations in predicting swelling potential of clayey soil over a span of conditions for which the model was trained.