Conditioning LSTM Decoder and Bi-directional Attention Based Question Answering System
This is an incremental improvement for question answering systems, enhancing accuracy through novel decoder conditioning and span prediction techniques.
The paper tackled question answering by implementing a model with bi-directional attention flow, multi-layer LSTM encoder, and a conditioning end-index decoder, which increased performance by 15.16% and achieved an F1 score of 73.97% and EM score of 64.95% on the test set.
Applying neural-networks on Question Answering has gained increasing popularity in recent years. In this paper, I implemented a model with Bi-directional attention flow layer, connected with a Multi-layer LSTM encoder, connected with one start-index decoder and one conditioning end-index decoder. I introduce a new end-index decoder layer, conditioning on start-index output. The Experiment shows this has increased model performance by 15.16%. For prediction, I proposed a new smart-span equation, rewarding both short answer length and high probability in start-index and end-index, which further improved the prediction accuracy. The best single model achieves an F1 score of 73.97% and EM score of 64.95% on test set.