Using deep learning (DL) with the Adam optimizer, I detected 15 previously undiscovered exoplanets in the Kepler data collected by the National Aeronautics and Space Administration (NASA). Some of the exoplanet transit signals were evident, but others were not as strong. Further evaluation is needed. By using my own code and DL libraries including TensorFlow, I built a Python program to search for exoplanets. Among the new candidate exoplanets detected by my program, 13 of them are ultra-short-period (USP) exoplanets with orbital periods shorter than a day. Moreover, I experimented on this Python DL program with current candidate and confirmed exoplanets in the NASA database and was able to detect more than 200 candidate exoplanets and 94 of the 116 confirmed exoplanets in the NASA database. These findings show that DL can be an effective tool to detect objects of interest, such as exoplanets, in astronomy big data.