bugfix-time-cnn

Use a character level cnn to predict the time to fix bug based on the different states during the life cycle of a bug.

Construct Model

We use the char_cnn_zhang model from this paper:

• Xiang Zhang, Junbo Zhao, Yann LeCun. Character-level Convolutional Networks for Text Classification. NIPS 2015

The model structure:

This graph may look difficult to understand. Here is the model setup.

We choose the small frame, 256 filters in convolutional layer and 1024 output units in dense layer.

• Embedding Layer
• Six convolutional layers, and 3 convolutional layers followed by a max pooling layer
• Two fully connected layer(dense layer in keras), neuron units are 1024.
• Output layer(dense layer), neuron units depends on classes. In this task, we set it 2.