Selected work

Representative Papers

Out-of-Distribution (OOD) Detectors for Open-Set RF Fingerprinting

Jun 10, 2026

This work addresses the challenge of out-of-distribution (OOD) signal detection in open-world radio frequency fingerprinting, where unknown transmitters and time-varying drifts induce distribution shifts—a problem particularly acute in realistic scenarios lacking genuine OOD tuning data. For the first time, this study systematically introduces OOD detection methods that require no real OOD tuning data into this domain. By establishing a unified information-theoretic mathematical framework, the authors propose an adaptive detection algorithm that integrates and extends multiple existing detection strategies. Experiments on the POWDER dataset demonstrate that the proposed approach achieves performance comparable to baseline methods that rely on real OOD data, while significantly outperforming current alternatives that operate without such tuning data.

0 citationsRead paper

A Variational Information Theoretic Approach to Out-of-Distribution Detection

Jun 17, 2025

This paper addresses the limited discriminability and interpretability of features in neural-network-based out-of-distribution (OOD) detection. We propose a variational information-theoretic framework for dual-objective feature learning. Methodologically, we introduce the first unified modeling of the information bottleneck and KL-divergence-based distribution separation: KL divergence explicitly enlarges the distance between in-distribution (ID) and OOD representations in latent space, while the information bottleneck compresses redundancy and preserves OOD-discriminative information. A novel shaping function is theoretically derived to enhance feature robustness and generalization. Our contributions are threefold: (1) the first OOD feature learning principle jointly driven by information bottleneck and distribution separation; (2) an interpretable and scalable feature construction paradigm; and (3) state-of-the-art performance—achieving a 12.3% reduction in FPR95 and a 3.8% improvement in AUROC on benchmarks including CIFAR/SVHN and ImageNet-O—significantly surpassing existing methods.

0 citationsRead paper

Recent publications

Latest Papers

Out-of-Distribution (OOD) Detectors for Open-Set RF Fingerprinting

Jun 10, 2026

This work addresses the challenge of out-of-distribution (OOD) signal detection in open-world radio frequency fingerprinting, where unknown transmitters and time-varying drifts induce distribution shifts—a problem particularly acute in realistic scenarios lacking genuine OOD tuning data. For the first time, this study systematically introduces OOD detection methods that require no real OOD tuning data into this domain. By establishing a unified information-theoretic mathematical framework, the authors propose an adaptive detection algorithm that integrates and extends multiple existing detection strategies. Experiments on the POWDER dataset demonstrate that the proposed approach achieves performance comparable to baseline methods that rely on real OOD data, while significantly outperforming current alternatives that operate without such tuning data.

0 citationsRead paper

A Variational Information Theoretic Approach to Out-of-Distribution Detection

Jun 17, 2025

This paper addresses the limited discriminability and interpretability of features in neural-network-based out-of-distribution (OOD) detection. We propose a variational information-theoretic framework for dual-objective feature learning. Methodologically, we introduce the first unified modeling of the information bottleneck and KL-divergence-based distribution separation: KL divergence explicitly enlarges the distance between in-distribution (ID) and OOD representations in latent space, while the information bottleneck compresses redundancy and preserves OOD-discriminative information. A novel shaping function is theoretically derived to enhance feature robustness and generalization. Our contributions are threefold: (1) the first OOD feature learning principle jointly driven by information bottleneck and distribution separation; (2) an interpretable and scalable feature construction paradigm; and (3) state-of-the-art performance—achieving a 12.3% reduction in FPR95 and a 3.8% improvement in AUROC on benchmarks including CIFAR/SVHN and ImageNet-O—significantly surpassing existing methods.

0 citationsRead paper