Out-of-distribution (OOD) detection remains a fundamental challenge for deep neural networks, particularly due to overconfident predictions on unseen OOD samples during testing. We reveal a key insight: OOD samples predicted as the same class, or given high probabilities for it, are visually more similar to each other than to the true in-distribution (ID) samples. Motivated by this class-specific observation, we propose DCAC (Dynamic Class-Aware Cache), a training-free, test-time calibration module that maintains separate caches for each ID class to collect high-entropy samples and calibrate the raw predictions of input samples. DCAC leverages cached visual features and predicted probabilities through a lightweight two-layer module to mitigate overconfident predictions on OOD samples. This module can be seamlessly integrated with various existing OOD detection methods across both unimodal and vision-language models while introducing minimal computational overhead. Extensive experiments on multiple OOD benchmarks demonstrate that DCAC significantly enhances existing methods, achieving substantial improvements, i.e., reducing FPR95 by 6.55% when integrated with ASH-S on ImageNet OOD benchmark.

Recent developments in few-shot out-of-distribution (OOD) detection have yielded remarkable performance, benefiting from large pre-trained vision-language models (VLMs). Our prior work focuses on using in-distribution (ID) knowledge as references to learn richer knowledge beyond the textual semantics of class labels, which is prone to cause the model overconfidence and results in a limited score gap between ID and OOD data. In this paper, rather than treating ID knowledge as references, we propose Prompt Learning with Neutral Knowledge (PLNK) to better differentiate ID from OOD data. Our key insight lies in leveraging diverse neutral knowledge to improve ID discrimination while alleviating the inherent model overconfidence on OOD data induced by ID knowledge, thereby capturing the notable discrepancy between ID and OOD data. By introducing neutral knowledge with a balanced degree of similarity to both ID and OOD data, we amplify the discrepancy between the learnable prompt and the references (i.e., diverse neutral knowledge) for ID data, while reducing it for OOD data. In this way, the simple yet effective PLNK framework brings a notable score gap between ID and OOD data, thereby improving OOD detection. Moreover, we incorporate the visual neutral prompt with richer semantics alongside the original text-only reference. Comprehensive experiments show that our method consistently surpasses current state-of-the-art methods. The codes will be released publicly.

Pre-trained vision-language models (VLMs) have advanced out-of-distribution (OOD) detection recently. However, existing CLIP-based methods often focus on learning OOD-related knowledge to improve OOD detection, showing limited generalization or reliance on external large-scale auxiliary datasets. In this study, instead of delving into the intricate OOD-related knowledge, we propose an innovative CLIP-based framework based on Forced prompt leArning (FA), designed to make full use of the In-Distribution (ID) knowledge and ultimately boost the effectiveness of OOD detection. Our key insight is to learn a prompt (i.e., forced prompt) that contains more diversified and richer descriptions of the ID classes beyond the textual semantics of class labels. Specifically, it promotes better discernment for ID images, by forcing more notable semantic similarity between ID images and the learnable forced prompt. Moreover, we introduce a forced coefficient, encouraging the forced prompt to learn more comprehensive and nuanced descriptions of the ID classes. In this way, FA is capable of achieving notable improvements in OOD detection, even when trained without any external auxiliary datasets, while maintaining an identical number of trainable parameters as CoOp. Extensive empirical evaluations confirm our method consistently outperforms current state-of-the-art methods. Code is available at https://github.com/0xFAFA/FA.

Wireless signals are frequently disturbed by complex noise sources, presenting a challenge to traditional direction of arrival (DOA) estimation methods that rely on the assumption of Gaussian noise. To address this issue, our letter proposes an innovative Bayesian DOA estimation approach. This method utilizes Gaussian mixture model (GMM) and Dirichlet process prior for accurately modeling the density function of complex noise environments. Additionally, an efficient combined message passing algorithm is formulated on the factor graph through the use of generalized approximate message passing (GAMP) and mean field (MF) techniques. Simulation results validate the effectiveness of this algorithm.

The design of message passing (MP) algorithms on factor graphs is an effective manner to implement channel estimation (CE) in wireless communication systems, which performance can be further improved by exploiting prior probability models that accurately match the channel characteristics. In this work, we study the CE problem in a downlink massive multiple-input multiple-output (MIMO) orthogonal frequency division multi-plexing (OFDM) system. As the prior probability, we propose the Markov chain two-state Gaussian mixture with large variance differences (TSGM-LVD) model to exploit the structured sparsity in the angle-frequency domain of the channel. Existing single and combined MP rules cannot deal with the message computation of the proposed probability model. To overcome this issue, we present a general method to derive the hybrid message passing (HMP) rule, which allows the calculation of messages described by mixed linear and non-linear functions. Accordingly, we design the HMP-TSGM-LVD algorithm under the structured turbo framework (STF). Simulation results demonstrate that the proposed algorithm converges faster and obtains better and more stable performance than its counterparts. In particular, the gain of the proposed approach is maximum (3 dB) in the high signal-to-noise ratio regime, while benchmark approaches experience oscillating behavior due to the improper prior model characterization.

This letter deals with variational inference for Dirichlet process mixtures (DPM) models. We propose a combined message-passing algorithm introducing belief propagation (BP) into the original mean field (MF) rules, which leads to a more precise approximate posterior in DPM. To compute the BP message, we change an exponential distribution to a non-exponential utilizing a flexible expression of Dirac delta function. Therefore, BP rules can be used to handle such functions, resulting to a local exact expectation instead of approximate expectation from the original MF method. Simulation results show that the proposed combined BP-MF algorithm results in a significant performance improvement compared to the state-of-the-art inference methods.