TasksSotADatasetsPapersMethodsSubmitAbout
Papers With Code 2

A community resource for machine learning research: papers, code, benchmarks, and state-of-the-art results.

Explore

Notable BenchmarksAll SotADatasetsPapersMethods

Community

Submit ResultsAbout

Data sourced from the PWC Archive (CC-BY-SA 4.0). Built by the community, for the community.

Papers/Metric Learning and Adaptive Boundary for Out-of-Domain De...

Metric Learning and Adaptive Boundary for Out-of-Domain Detection

Petr Lorenc, Tommaso Gargiani, Jan Pichl, Jakub Konrád, Petr Marek, Ondřej Kobza, Jan Šedivý

2022-04-22Open Intent DetectionOut of Distribution (OOD) DetectionMetric Learning
PaperPDFCode(official)

Abstract

Conversational agents are usually designed for closed-world environments. Unfortunately, users can behave unexpectedly. Based on the open-world environment, we often encounter the situation that the training and test data are sampled from different distributions. Then, data from different distributions are called out-of-domain (OOD). A robust conversational agent needs to react to these OOD utterances adequately. Thus, the importance of robust OOD detection is emphasized. Unfortunately, collecting OOD data is a challenging task. We have designed an OOD detection algorithm independent of OOD data that outperforms a wide range of current state-of-the-art algorithms on publicly available datasets. Our algorithm is based on a simple but efficient approach of combining metric learning with adaptive decision boundary. Furthermore, compared to other algorithms, we have found that our proposed algorithm has significantly improved OOD performance in a scenario with a lower number of classes while preserving the accuracy for in-domain (IND) classes.

Results

TaskDatasetMetricValueModel
Intent DetectionBANKING-77 (50% known)1:1 Accuracy83.78Metric learning + Adaptive Decision Boundary
Intent DetectionBANKING-77 (50% known)F1-score84.93Metric learning + Adaptive Decision Boundary
Intent DetectionOOS(25%known)1:1 Accuracy91.81Metric learning + Adaptive Decision Boundary
Intent DetectionOOS(25%known)F1-score85.9Metric learning + Adaptive Decision Boundary
Intent DetectionBANKING-77 (75% known)1:1 Accuracy84.4Metric learning + Adaptive Decision Boundary
Intent DetectionBANKING-77 (75% known)F1-score88.39Metric learning + Adaptive Decision Boundary
Intent DetectionBANKING77 (25%known)1:1 Accuracy85.71Metric learning + Adaptive Decision Boundary
Intent DetectionBANKING77 (25%known)F1-score78.86Metric learning + Adaptive Decision Boundary
Intent DetectionOOS(75%known)1:1 Accuracy88.54Metric learning + Adaptive Decision Boundary
Intent DetectionOOS(75%known)F1-score92.21Metric learning + Adaptive Decision Boundary
Intent DetectionOOS(50%known)1:1 Accuracy88.81Metric learning + Adaptive Decision Boundary
Intent DetectionOOS(50%known)F1-score89.19Metric learning + Adaptive Decision Boundary

Related Papers

Unsupervised Ground Metric Learning2025-07-17Are encoders able to learn landmarkers for warm-starting of Hyperparameter Optimization?2025-07-16ZClassifier: Temperature Tuning and Manifold Approximation via KL Divergence on Logit Space2025-07-14$\texttt{Droid}$: A Resource Suite for AI-Generated Code Detection2025-07-11Safe Domain Randomization via Uncertainty-Aware Out-of-Distribution Detection and Policy Adaptation2025-07-08FA: Forced Prompt Learning of Vision-Language Models for Out-of-Distribution Detection2025-07-06Grid-Reg: Grid-Based SAR and Optical Image Registration Across Platforms2025-07-06Dare to Plagiarize? Plagiarized Painting Recognition and Retrieval2025-06-29