Exper. ID | Embedding | Filter size | Filters nb. | Dropout | Batch size | Hidden dims | Remove stop words | Pre-processing | Acc. (%) |
|---|
Ex1 | 300 | 3, 8 | 30 | 0.5, 0.8 | 64 | 50 | No | No | 47.59 |
Ex2 | 300 | 3, 8 | 30 | 0.5, 0.8 | 128 | 50 | No | Yes | 54.55 |
Ex3 | 300 | 3, 8 | 30 | 0.5, 0.8 | 128 | 100 | No | Yes | 52.05 |
Ex4 | 300 | 3, 8 | 40 | 0.5, 0.8 | 128 | 100 | No | Yes | 53.48 |
Ex5 | 300 | 3, 8 | 50 | 0.5, 0.8 | 128 | 100 | No | Yes | 51.69 |
Ex6 | 300 | 3, 4, 5 | 40 | 0.5, 0.8 | 128 | 100 | No | Yes | 55.26 |
Ex7 | 512 | 3, 4, 5 | 40 | 0.5, 0.8 | 128 | 100 | No | Yes | 56.51 |
Ex8 | 300 | 3, 4, 5 | 40 | 0.5, 0.8 | 128 | 100 | Yes | Yes | 99.82 |
- where, Embedding is the first layer in a model. It requires that the input data be integer encoded, so that each word is represented by a unique integer, Filter size is the size of the filter used in the experiment, Filter nb is an integer which represents the dimensionality of the output space, Dropout represents applying a technique where randomly selected neurons are ignored during training, Batch size is the number of training examples in one forward/backward pass, Hidden Dims is the number of neurons in this hidden layer, Remove stop words is the elimination of the stop words from the text in this experiment, and Pre-processing is the implementation of the preprocessing steps mentioned in “Data preprocessing” section excluding the step of the stop words elimination
