Confidence Calibration for Intent Detection via Hyperspherical Space and Rebalanced Accuracy-Uncertainty Loss

TL;DR

This paper introduces a novel intent detection model that improves confidence calibration by using hyperspherical space and a rebalanced loss, leading to better alignment between confidence and accuracy.

Contribution

It proposes a hyperspherical label projection and a rebalanced accuracy-uncertainty loss to enhance confidence calibration in intent detection models.

Findings

Outperforms existing calibration methods on open datasets.

Achieves significant improvement in calibration metrics.

Reduces over-confidence in neural network predictions.

Abstract

Data-driven methods have achieved notable performance on intent detection, which is a task to comprehend user queries. Nonetheless, they are controversial for over-confident predictions. In some scenarios, users do not only care about the accuracy but also the confidence of model. Unfortunately, mainstream neural networks are poorly calibrated, with a large gap between accuracy and confidence. To handle this problem defined as confidence calibration, we propose a model using the hyperspherical space and rebalanced accuracy-uncertainty loss. Specifically, we project the label vector onto hyperspherical space uniformly to generate a dense label representation matrix, which mitigates over-confident predictions due to overfitting sparce one-hot label matrix. Besides, we rebalance samples of different accuracy and uncertainty to better guide model training. Experiments on the open datasets verify that our model outperforms the existing calibration methods and achieves a significant improvement on the calibration metric.