Interpretability of the Intent Detection Problem: A New Approach
This work addresses the interpretability gap in deep learning for intent detection, offering mechanistic insights into how dataset properties affect RNN performance, which is incremental but provides a novel geometric framework.
The authors tackled the problem of understanding how Recurrent Neural Networks (RNNs) perform intent detection by applying dynamical systems theory to analyze hidden state trajectories, revealing that on balanced datasets, RNNs partition the state space into distinct intent clusters, but this ideal solution degrades for low-frequency intents in imbalanced datasets.
Intent detection, a fundamental text classification task, aims to identify and label the semantics of user queries, playing a vital role in numerous business applications. Despite the dominance of deep learning techniques in this field, the internal mechanisms enabling Recurrent Neural Networks (RNNs) to solve intent detection tasks are poorly understood. In this work, we apply dynamical systems theory to analyze how RNN architectures address this problem, using both the balanced SNIPS and the imbalanced ATIS datasets. By interpreting sentences as trajectories in the hidden state space, we first show that on the balanced SNIPS dataset, the network learns an ideal solution: the state space, constrained to a low-dimensional manifold, is partitioned into distinct clusters corresponding to each intent. The application of this framework to the imbalanced ATIS dataset then reveals how this ideal geometric solution is distorted by class imbalance, causing the clusters for low-frequency intents to degrade. Our framework decouples geometric separation from readout alignment, providing a novel, mechanistic explanation for real world performance disparities. These findings provide new insights into RNN dynamics, offering a geometric interpretation of how dataset properties directly shape a network's computational solution.