This is a brief summary of paper for me to study and organize it, Convolutional Neural Networks for Sentence Classification (Kim., EMNLP 2014) I read and studied.
They propose CNN architecture on top two sets of pre-training word2vecs, static and dynamic one.
Their architecture is the followings:
For two sets of pre-training word vectors, one of word2vecs is kept static throughout training training and another is fine-tuned via backpropagation.
The following is examples of top 4 neighboring words based on cosine similarity for static channel(left) and fine-tuned vectors in the non-static channel (right) for multichannel model.
Note(Abstract):
They report on a series of experiments with convolutional neural networks (CNN) trained on top of pre-trained word vectors for sentence-level classification tasks. They show that a simple CNN with little hyperparameter tuning and static vectors achieves excellent results on multiple benchmarks. Learning task-specific vectors through fine-tuning offers further gains in performance. They additionally propose a simple modification to the architecture to allow for the use of both task-specific and static vectors.
Download URL:
The paper: Convolutional Neural Networks for Sentence Classification (Kim., EMNLP 2014)
The paper: Convolutional Neural Networks for Sentence Classification (Kim., EMNLP 2014)
Reference
- Paper
- How to use html for alert
- For your information