This paper presents a novel algorithmic approach for accurate and reliable photometric redshift determination, which is a key factor in wide-field photometric surveys. Instead of conventional machine learning and artificial intelligence techniques, we propose a method to determine the photometric redshift of galaxies using a conditional generative adversarial network (CGAN). The method provides both point estimates and probability density estimates of photometric redshifts, and is tested using the Dark Energy Survey (DES) Y1 data and compared with the Mixed Density Network (MDN). Although the MDN outperforms the CGAN, its performance metrics are similar to those of the MDN, suggesting the potential application of CGAN for photometric redshift estimation.
