MGNC-CNN: A Simple Approach to Exploiting Multiple Word Embeddings for Sentence Classification

摘要: We introduce a novel, simple convolution neural network (CNN) architecture - multi-group norm constraint CNN (MGNC-CNN) that capitalizes on multiple sets of word embeddings for sentence classification. MGNC-CNN extracts features from input embedding independently and then joins these at the penultimate layer in to form final feature vector. adopt group regularization strategy differentially penalizes weights associated with subcomponents generated respective sets. This model is much simpler than comparable alternative architectures requires substantially less training time. Furthermore, it flexible does not require be same dimensionality. show consistently outperforms baseline models.