This study introduces a novel approach for designing fork antennas optimized for the Ultra-Wideband (UWB) frequency band using machine learning. The aim is to develop a parameterized model predicting antenna dimensions based on given performance indices. Leveraging parameter sweep techniques, a dataset capturing the relationship between antenna design (dimensions) and the corresponding performance metrics was generated in CST Microwave Studio. Then the generated dataset was used to train a fully connected Artificial Neural Network (ANN) model where the inputs are the performance metrics and the outputs are the corresponding antenna dimensions. It is demonstrated that the trained ANN model can determine the optimal antenna dimensions very efficiently given the desired performance indices comparing to existing antenna design software. This approach accelerates innovation in 5G communication systems and contributes to ongoing efforts in 6G communication research.