Comparer des méthodes

	GAN multimodale	Transformeur Multimodal
Domaine	Apprentissage profond	Apprentissage profond
Famille	Machine learning	Machine learning
Année d'origine	2014–2016	2019–2021
Auteur d'origine	Reed et al. (text-to-image GAN); foundation by Goodfellow et al.	Lu et al. (ViLBERT); Radford et al. (CLIP)
Type	Generative adversarial model	Cross-modal attention-based deep learning model
Source fondatrice	Reed, S., Akata, Z., Yan, X., Logeswaran, L., Schiele, B., & Lee, H. (2016). Generative adversarial text to image synthesis. Proceedings of the 33rd International Conference on Machine Learning (ICML), PMLR 48, 1060–1069.	Lu, J., Batra, D., Parikh, D., & Lee, S. (2019). ViLBERT: Pretraining Task-Agnostic Visiolinguistic Representations for Vision-and-Language Tasks. Advances in Neural Information Processing Systems (NeurIPS), 32.
Alias	MM-GAN, multimodal generative adversarial network, cross-modal GAN, multi-modal GAN	multimodal attention model, cross-modal transformer, vision-language transformer, multi-modal fusion transformer
Apparentées	4	5
Résumé	A Multimodal GAN is a generative adversarial network conditioned on — or jointly learning across — more than one data modality (e.g., text descriptions, images, audio, or structured data). By fusing information from multiple sources, the generator can synthesize realistic outputs that respect cross-modal constraints, enabling tasks such as text-to-image synthesis, image-to-audio generation, and joint modality imputation.	A Multimodal Transformer extends the standard Transformer architecture to process and jointly reason over two or more input modalities — most commonly text and images, but also audio, video, or structured data. Cross-modal attention layers allow information from one modality to inform representations in another, enabling tasks such as visual question answering, image captioning, and multimodal sentiment analysis.