DEER: Disentangled Mixture of Experts with Instance-Adaptive Routing for Generalizable Machine-Generated Text Detection

TL;DR

DEER is a novel framework that improves machine-generated text detection by disentangling domain-specific and domain-general features and using adaptive routing to handle domain shifts, outperforming existing methods.

Contribution

The paper introduces DEER, a two-stage architecture with disentangled experts and reinforcement learning-based routing, addressing domain shift issues in MGT detection.

Findings

Outperforms state-of-the-art methods on multiple benchmarks.

Achieves significant improvements in F1-score and accuracy.

Validates the importance of expert disentanglement and adaptive routing.

Abstract

Detecting machine-generated text (MGT) has emerged as a critical challenge, driven by the rapid advancement of large language models (LLMs) capable of producing highly realistic, human-like content. However, the performance of current approaches often degrades significantly under domain shift. To address this challenge, we propose a novel framework designed to capture both domain-specific and domain-general MGT patterns through a two-stage Disentangled mixturE-of-ExpeRts (DEER) architecture. First, we introduce a disentangled mixture-of-experts module, in which domain-specific experts learn fine-grained, domain-local distinctions between human and machine-generated text, while shared experts extract transferable, cross-domain features. Second, to mitigate the practical limitation of unavailable domain labels during inference, we design a reinforcement learning-based routing mechanism that dynamically selects the appropriate experts for each input instance, effectively bridging the train-inference gap caused by domain uncertainty. Extensive experiments on five in-domain and five out-of-domain benchmark datasets demonstrate that DEER consistently outperforms state-of-the-art methods, achieving average F1-score improvements of 1.39% and 5.32% on in-domain and out-of-domain datasets respectively, along with accuracy gains of 1.35% and 3.61% respectively. Ablation studies confirm the critical contributions of both disentangled expert specialization and adaptive routing to model performance.