Сравнение методов

	Мультимодальный GRU	Мультимодальный трансформер
Область	Глубокое обучение	Глубокое обучение
Семейство	Machine learning	Machine learning
Год появления	2014–2017	2019–2021
Автор метода	Cho, K. et al. (GRU); adapted to multimodal settings by multiple research groups	Lu et al. (ViLBERT); Radford et al. (CLIP)
Тип	Recurrent neural network (multimodal variant)	Cross-modal attention-based deep learning model
Основополагающий источник	Cho, K., van Merriënboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., & Bengio, Y. (2014). Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation. Proceedings of EMNLP 2014, 1724–1734.	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.
Другие названия	MM-GRU, Multimodal Gated Recurrent Unit, Cross-modal GRU, Multi-input GRU	multimodal attention model, cross-modal transformer, vision-language transformer, multi-modal fusion transformer
Связанные	6	5
Сводка	Multimodal GRU extends the Gated Recurrent Unit architecture to jointly process sequential data from multiple input modalities — such as text, audio, and video frames — within a single recurrent framework. By fusing modality-specific encodings at the input or hidden-state level, it captures temporal dependencies across heterogeneous data streams and is widely used in multimodal sentiment analysis, video understanding, and audio-visual speech recognition.	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.