Update Frequently, Update Fast: Retraining Semantic Parsing Systems in a Fraction of Time
This work addresses the need for efficient updates in deployed voice assistants, though it is incremental as it builds on existing fine-tuning techniques.
The authors tackled the problem of slow retraining for semantic parsing systems in voice assistants, which typically take weeks, by proposing a method that reduces training time to less than 10% while matching the performance of models trained from scratch.
Currently used semantic parsing systems deployed in voice assistants can require weeks to train. Datasets for these models often receive small and frequent updates, data patches. Each patch requires training a new model. To reduce training time, one can fine-tune the previously trained model on each patch, but naive fine-tuning exhibits catastrophic forgetting - degradation of the model performance on the data not represented in the data patch. In this work, we propose a simple method that alleviates catastrophic forgetting and show that it is possible to match the performance of a model trained from scratch in less than 10% of a time via fine-tuning. The key to achieving this is supersampling and EWC regularization. We demonstrate the effectiveness of our method on multiple splits of the Facebook TOP and SNIPS datasets.