SNOiC: Soft Labeling and Noisy Mixup based Open Intent Classification Model

TL;DR

This paper introduces SNOiC, a novel open intent classification model that leverages soft labeling and noisy mixup techniques to improve open intent detection accuracy and robustness over existing threshold-based methods.

Contribution

SNOiC combines soft labeling and noisy mixup strategies to reduce bias and generate pseudo-data, enhancing open intent classification performance beyond prior models.

Findings

Achieves 68.72% to 94.71% accuracy on benchmark datasets.

Improves open intent detection by up to 12.76% over state-of-the-art.

Validated effectiveness through ablation studies and parameter analysis.

Abstract

This paper presents a Soft Labeling and Noisy Mixup-based open intent classification model (SNOiC). Most of the previous works have used threshold-based methods to identify open intents, which are prone to overfitting and may produce biased predictions. Additionally, the need for more available data for an open intent class presents another limitation for these existing models. SNOiC combines Soft Labeling and Noisy Mixup strategies to reduce the biasing and generate pseudo-data for open intent class. The experimental results on four benchmark datasets show that the SNOiC model achieves a minimum and maximum performance of 68.72\% and 94.71\%, respectively, in identifying open intents. Moreover, compared to state-of-the-art models, the SNOiC model improves the performance of identifying open intents by 0.93\% (minimum) and 12.76\% (maximum). The model's efficacy is further established by analyzing various parameters used in the proposed model. An ablation study is also conducted, which involves creating three model variants to validate the effectiveness of the SNOiC model.