Fine-tuning Handwriting Recognition systems with Temporal Dropout
This addresses overfitting in handwriting recognition systems, but it is incremental as it builds on existing RNN/LSTM methods.
The paper tackles overfitting in LSTM-based handwriting recognition by proposing Temporal Dropout, which drops information at random positions in sequences, and shows improved results, outperforming previous state-of-the-art on the Rodrigo dataset.
This paper introduces a novel method to fine-tune handwriting recognition systems based on Recurrent Neural Networks (RNN). Long Short-Term Memory (LSTM) networks are good at modeling long sequences but they tend to overfit over time. To improve the system's ability to model sequences, we propose to drop information at random positions in the sequence. We call our approach Temporal Dropout (TD). We apply TD at the image level as well to internal network representation. We show that TD improves the results on two different datasets. Our method outperforms previous state-of-the-art on Rodrigo dataset.