Сравнение методов

	Мультимодальное обучение с подкреплением	Мультимодальная графовая нейронная сеть
Область	Глубокое обучение	Глубокое обучение
Семейство	Machine learning	Machine learning
Год появления	2015–2022	2019–2020
Автор метода	Multiple contributors (DeepMind, OpenAI, Google Brain, 2010s–2020s)	Kipf & Welling (GNN foundation); extended to multimodal settings by multiple research groups c. 2019–2020
Тип	Multimodal deep RL agent	Graph-based deep learning with multimodal input fusion
Основополагающий источник	Reed, S., Zolna, K., Parisotto, E., Colmenarejo, S. G., Novikov, A., Barth-Maron, G., ... & de Freitas, N. (2022). A Generalist Agent. Transactions on Machine Learning Research.	Kipf, T. N., & Welling, M. (2017). Semi-Supervised Classification with Graph Convolutional Networks. International Conference on Learning Representations (ICLR).
Другие названия	Multimodal RL, Multi-Sensory Reinforcement Learning, Vision-Language RL, Multi-Input RL	MM-GNN, Multimodal GNN, Multi-modal Graph Network, Cross-modal Graph Neural Network
Связанные	6	6
Сводка	Multimodal Reinforcement Learning trains agents to make sequential decisions by perceiving and integrating multiple input modalities — such as raw pixels, language instructions, audio, and proprioceptive sensors — simultaneously. Rather than acting on a single data stream, the agent fuses heterogeneous signals into a unified state representation and learns a policy through environmental reward feedback.	A Multimodal Graph Neural Network (MM-GNN) combines data from multiple modalities — such as text, images, and structured features — into a unified graph structure and applies graph-based message passing to learn joint representations. It enables relational reasoning across heterogeneous data sources, going beyond what unimodal or simple concatenation approaches can capture.