Depression impacts millions worldwide, with increasing efforts to use social media data for early detection and intervention. Traditional Risk Detection (TRD) uses a user’s complete posting history for predictions, while Early Risk Detection (ERD) seeks early detection in a user’s posting history, emphasizing the importance of prediction earliness. However, ERD remains relatively underexplored due to challenges in balancing accuracy and earliness, especially with evolving partial data. To address this, we introduce the Early Sensing Depression Model (ESDM), which comprises two modules classification with partial information module (CPI) and decision for classification moment module (DMC), alongside an early detection loss function. Experiments show ESDM outperforms benchmarks in both earliness and accuracy.

The objective of suicide risk detection on social media is to identify individuals who may attempt suicide and determine their suicide risk level based on their online behavior. Although data-driven learning models have been used to predict suicide risk levels, these models often lack theoretical support and explanation from psychiatric research. To address this issue, we propose the incorporation of professional psychiatric scales into research to provide theoretical support and explanations for our model. Our proposed Scale-based Neural Network (SNN) architecture aims to extract content associated with scales from the posting history of social media users to predict their suicide risk level. Additionally, our approach provides scale-based explanations for the model's predictions. Experimental results demonstrate that our proposed method outperforms several strong baseline methods and highlights the potential of combining psychiatric scales and computational techniques to improve suicide risk detection.

Specific emitter identification (SEI) is a non-cryptographic authentication technique to provide an extra security layer for wireless devices, which has promising applications. However, the traditional methods of SEI are only available in limited equipments. In actual application scenarios, new devices (as new classes) are constantly appearing. In this paper, an effective class incremental learning (CIL) method is proposed for SEI, named class-incremental with self-training and prototype augmentation (CISP). It is a teacher-student network. Firstly, the teacher network trained by the old-class data is utilized to instruct the student network to adapt the new classes while retaining the old-class knowledge through the knowledge distillation (KD) techniques. Secondly, in order to mitigate the problem of favoring the new classes, weight aligning (WA) method is introduced to balance the weights of the new-class and old-class classification layers in the student network. Lastly, the old-class samples are recalled from the unlabeled dataset by the student network and input into the teacher network. Then the feature prototypes of the old classes are constructed and augmented. This would further ease the imbalance between the old and new classes and alleviate the problem of noisy pseudo-labels. Experiment results on the real AIS-100 dataset and ADS-B-100 dataset with the number of the initial classes being 20 and 20 classes per incremental step demonstrate that the proposed method can achieve an average accuracy of 95.29% and 95.84%, respectively. It effectively mitigates the catastrophic forgetting of the model and is superior to the state-of-the-art incremental learning approaches of not saving the old-class samples.

The advanced graphics rendering techniques and image generation algorithms significantly improve the visual quality of computer-generated (CG) images, and this makes it more challenging to distinguish between CG images and natural images (NIs) for a forensic detector. For the identification of CG images, human beings often need to inspect and evaluate the entire image and its local region as well. In addition, we observe that the distributions of both near and far patch-wise correlation have differences between CG images and NIs. Current mainstream methods adopt the CNN-based architecture with the classical cross entropy loss, however, there are several limitations: 1) the weakness of long-distance relationship modeling of image content due to the local receptive field of CNN; 2) the pixel sensitivity due to the convolutional computation; 3) the insufficient supervision due to the training loss on the whole image. In this paper, we propose a novel vision transformer (ViT)-based network with token labeling for CG image identification. Our network, called CGFormer, consists of patch embedding, feature modeling, and token prediction. We apply patch embedding to sequence the input image and weaken the pixel sensitivity. Stacked multi-head attention-based transformer blocks are utilized to model the patch-wise relationship and introduce a certain level of adaptability. Besides the conventional classification loss on class token of the whole image, we additionally introduce a soft cross entropy loss on patch tokens to comprehensively exploit the supervision information from local patches. Extensive experiments demonstrate that our method achieves the state-of-the-art forensic performance on six publicly available datasets in terms of classification accuracy, generalization, and robustness. Code is available athttps://github.com/feipiefei/CGFormer.

The automatic identification system (AIS) is the automatic tracking system for automatic traffic control and collision avoidance services, which plays an important role in maritime traffic safety. However, it faces a possible security threat when the maritime mobile service identity (MMSI) that specifies the vessels’ identity in AIS is illegally counterfeited. To guarantee the communication security of AIS for preventing fraudulent devices, we design a novel lightweight Transformer-based network GLFormer for specific emitter identification (SEI) to provide an extra security layer for AIS terminal emitters. Concretely, the gated local attention unit (GLAU) and the gated sliding local attention unit (GSLAU) modules that combine a simplified gated attention unit (GAU) and a sliding local self-attention (SLA) are developed in GLFormer to extract the radio frequency fingerprint (RFF) features automatically from the raw in-phase signals. Especially, the simplified GAU focuses on more critical RFF features and filters out the irrelevant information from the raw signal to improve performance, which is also a single-head self-attention module with fewer parameters for lightweight. Meanwhile, the SLA limits self-attention operation to a window, introducing the inductive bias of local information to enhance performance further and reducing the quadratic computational complexity to linearity for efficiency. Experimental results demonstrate that the GLFormer achieves 96.31% and 89.38% identification accuracy in the constructed AIS transient and AIS steady-state datasets with 50 vessels, respectively. The 99.90% identification accuracy is achieved in the universal software radio peripheral (USRP) dataset with ten devices. It is not only better than the existing methods but requires much fewer parameters and lower computational complexity; besides, it is also suitable for working with long signal sequences.

To obtain an insight about the effect of channel parameters on buffer‐aided relay selection systems, the max‐link selection (MLS) schemes are investigated over independent and non‐identically distributed (i.ni.d) Rayleigh fading channels. Specially, by employing Markov chain, the authors first model the state transition matrix and outage probability. Secondly, they obtain the closed‐form expressions of the corresponding statistic properties. The presented results show: (i) when the relaying channels are asymmetric (but the source–relay links are independent and identically distributed fading, so does the relay–destination links), the MLS scheme outperforms the traditional best relay selection (T‐BRS) and max–max best relay selection (MM‐BRS) schemes. However, when the relaying links are unbalanced severely, the MLS scheme does not provide diversity gain over the T‐BRS and MM‐BRS schemes; and (ii) For the more general i.ni.d fading case, it is observed that the MLS schemes can be inferior to the MM‐BRS schemes. The derivations of this work have values for reference. For example, by using the achieved statistic properties they can further perform the investigation on the delay for the buffer‐aided MLS relaying systems over i.ni.d fading channels, which is a critical issue of buffer‐aided relay selection systems.