コンテンツへスキップ
用語集に戻る
技術

Generative Adversarial Network (GAN)

定義

A framework consisting of two neural networks—a generator and a discriminator—that compete against each other. The generator creates synthetic data; the discriminator tries to distinguish real from fake. GANs produce highly realistic images and are also studied in the context of deepfake detection.

AIの理解にお困りですか?

Physical AI 適合性コールをご予約いただき、これらのAI概念が貴社の業界や課題にどう適用されるかをご相談ください。

Generative Adversarial Network (GAN) | AI用語集 — LLM、RAG、ファインチューニング&主要概念