说明：  我们探索了使用细胞神经网络和LSTM来解决新闻标题分为四类的问题的潜力...
(We explore the potential of using CNNs and LSTMs to solve the problem of classification of news headlines into four categories - business, entertainment, health, science and technology. We show that pre-trained word2vec embeddings are able to improve pure deep architectures. We demonstrate that LSTMs are able to outperform CNNs on this particula...)