This Letter presents a baseband Digital PreDistortion (DPD) techniquebased on Feed-Forward Neural Network (FFNN). The process of MemoryPolynomial (MP) DPD is time-consuming because of the large numberof mathematical calculations. The FFNN is adopted to realize themathematical calculations in MP DPD with direct learning architecture.The training samples of FFNN are derived from MP DPD with directlearning architecture. It guarantees the accuracy of imitating the MP DPD.Although the training of the FFNN is time-consuming, the trained FFNNDPD is less time-consuming than MP DPD. This solution is validatedbased on a PA ZFL-2500 driven by a WCDMA signal with 3.84 MHzbandwidth. The experimental results show that the FFNN can mimicthe behavior of the MP DPD. The proposed DPD achieves a significantimprovement in linearity and is stable.