ISBN: 978-981-11-3671-9 DOI: 10.18178/wcse.2017.06.076
Multimodal Learning with Deep Associative Model
Abstract— An associative-generated model based on Deep Belief Network (DBN) and Bi-directional
Associative Memory (BAM) neural network is presented. The model is capable of extracting features from
multiple input modalities and forming an associative memory. By such associative memorization, the model
can regenerate one channel from the other, and perform classification with missing inputs.
Index Terms— associative memory, generation, multimodal learning, deep belief network
Dongwei Guo, Zhihua Zeng
College of Computer Science and Technology, Jilin University, CHINA
Cite: Dongwei Guo, Zhihua Zeng, "Multimodal Learning with Deep Associative Model," Proceedings of 2017 the 7th International Workshop on Computer Science and Engineering, pp. 444-447, Beijing, 25-27 June, 2017.