Machine learning techniques provide an automated solution for reservoir characterization and for the prediction of Petrophysical properties far from well locations. Porosity is one of the essential attributes of reservoir rocks as it determines the amount of fluid a rock can contain. The Hugin sandstone reservoir in the Volve field from North Sea Norway is a thin reservoir with high variation in sand quality. This thickness variation poses challenges in predicting reliable reservoir properties and consequently influences the reserves and production forecasts. This study aims to enhance the characterization of the Hugin Formation reservoir using different seismic interpretation techniques and Machine Learning approaches. To achieve this objective, different porosity prediction approaches are tested. The methods are evaluated and tested at the well locations before applying them on the seismic data. A large number of attributes is prepared for this task. This includes P impedance, S-impedance, Vp/Vs, Lambda-Rho, Mu-Rho, among others. The supervised process started with Single Attributes Regression where individual attributes are assessed to find a relationship with the reservoir porosity. Then, Multi Attribute Regression is used to combine multiple attributes to improve the relationship found between attributes and the porosity. Eventually, Neural Networks namely Deep Feed Forward Neural Network (DFNN) and Probabilistic Neural Network (PNN) are utilized to search for better and more accurate relationships with higher correlation between the input attributes and the porosity. The Probabilistic Neural Network showed a very good capability in predicting the porosity from the input seismic attributes. The predicted porosity map of the reservoir revealed the distribution of the Hugin Formation facies. In addition, the unsupervised Machine Learning helped classifying the reservoir into classes that are found consistent with the facies defined by the PNN.