方法对比

	多模态多层感知器	微调多层感知机
领域	深度学习	深度学习
方法族	Machine learning	Machine learning
起源年份	2011 (multimodal extension); 1986 (MLP backpropagation)	1986 (MLP); fine-tuning practice formalised c. 2014
提出者	Ngiam et al. / Rumelhart, Hinton & Williams (MLP foundations)	Rumelhart, Hinton & Williams (MLP); Yosinski et al. (fine-tuning analysis)
类型	Feedforward neural network with multi-stream fusion	Supervised deep learning with pre-trained weight initialisation
开创性文献	Ngiam, J., Khosla, A., Kim, M., Nam, J., Lee, H., & Ng, A. Y. (2011). Multimodal deep learning. In Proceedings of the 28th International Conference on Machine Learning (ICML 2011), pp. 689–696.	Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986). Learning representations by back-propagating errors. Nature, 323, 533–536.
别名	MM-MLP, multimodal MLP, multi-input feedforward network, fusion multilayer perceptron	fine-tuned MLP, adapted MLP, domain-adapted multilayer perceptron, MLP fine-tuning
相关	5	4
摘要	A Multimodal Multilayer Perceptron (MM-MLP) is a feedforward neural network that ingests features from two or more heterogeneous input modalities — such as structured tabular data, text embeddings, and image feature vectors — by encoding each stream separately and fusing them into a shared representation before passing it through fully connected layers to produce a classification or regression output.	A Fine-Tuned Multilayer Perceptron starts from weights learned on a source task — or a large general-purpose dataset — and continues training on a smaller target dataset with a reduced learning rate. This reuse of pre-learned representations allows the MLP to converge faster and generalise better than training from scratch, especially when labelled target data is scarce.