方法对比

	多语言生成对抗网络 (Multilingual GAN)	迁移学习GAN
领域	深度学习	深度学习
方法族	Machine learning	Machine learning
起源年份	2017–2019	2014–2018
提出者	Goodfellow et al. (GAN); multilingual extensions by various authors from 2017 onward	Goodfellow, I. et al. (GAN); Wang & Ramanan (transfer to GAN)
类型	Generative adversarial model with multilingual conditioning	Generative model with transferred weights
开创性文献	Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., & Bengio, Y. (2014). Generative Adversarial Nets. Advances in Neural Information Processing Systems (NeurIPS), 27.	Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A. & Bengio, Y. (2014). Generative Adversarial Nets. Advances in Neural Information Processing Systems (NeurIPS), 27, 2672–2680.
别名	Multilingual GAN, Cross-lingual GAN, Multilingual Generative Adversarial Network, ML-GAN	TL-GAN, pretrained GAN, GAN fine-tuning, domain-adaptive GAN
相关	5	6
摘要	A Multilingual GAN pairs the generative adversarial framework with cross-lingual components — a shared encoder, language-conditioned generator, and a language discriminator — so that a single model can generate or align representations across multiple languages simultaneously. It is applied to cross-lingual text generation, machine translation, multilingual data augmentation, and language-invariant feature learning.	Transfer Learning GAN initialises a Generative Adversarial Network — or both its generator and discriminator — from weights pretrained on a large source dataset, then fine-tunes the network on a smaller target dataset. This approach allows high-quality generative modelling even when target-domain data are scarce, by reusing low- and mid-level feature representations learned at scale.