Abstract: Since the transient electromagnetic (TEM) method is susceptible to interference during surveys, denoising is a key step in TEM data processing. For conventional denoising methods, multiple techniques are required to suppress various types of noise, while the performance of these methods depends on manual experience. This study proposes a method for denoising of TEM data using artificial intelligence and deep learning. Specifically, signals bearing noise are input to the self-coding network for denoising, while the denoised data are output after encoding and decoding. Numerical simulation results indicate that the method using deep learning and neural networks enables the effective denoising of TEM data.