[11a-W810-1~10] 31.1 Focused Session "AI Electornics"

[11a-W810-10] Generative Adversarial Network based on Chaotic Time Series: CHAOS GAN

〇Makoto Naruse¹, Takatomo Mihana², Kazutaka Kanno², Atsushi Uchida² (1.NICT, 2.Saitama Univ.)

Keywords:generative adversarial network

We experimentally demonstrate that a generative adversarial network generates qualitatively different outcomes when the input data was given by chaotic time series generated by a semiconductor laser, indicating that the temporal structure inherent in laser chaos is inherited to artificially generated outputs.