Caiping Hu

With the advancement of deep forgery techniques, particularly propelled by generative adversarial networks (GANs), identifying deepfake faces has become increasingly challenging. Although existing forgery detection methods can identify tampering details within manipulated images, their effectiveness significantly diminishes in complex scenes, especially in low-quality images subjected to compression. To address this issue, we proposed a novel deep face forgery video detection model named Two-Stream Feature Domain Fusion Network (TSFF-Net). This model comprises spatial and frequency domain feature extraction branches, a feature extraction layer, and a Transformer layer. In the feature extraction module, we utilize the Scharr operator to extract edge features from facial images, while also integrating frequency domain information from these images. This combination enhances the model’s ability to detect low-quality deepfake videos. Experimental results demonstrate the superiority of our method, achieving detection accuracies of 97.7%, 91.0%, 98.9%, and 90.0% on the FaceForensics++ dataset for Deepfake, Face2Face, FaceSwap, and NeuralTextures forgeries, respectively. Additionally, our model exhibits promising results in cross-dataset experiments.. The code used in this study is available at: https://github.com/hwZHc/TSFF-Net.git.

Compressed sensing (CS) is an innovative signal acquisition and reconstruction technology that has broken through the limit of the Nyquist sampling theory. It is widely employed to optimize the measurement processes in various applications. One of the core challenges of CS is the construction of a measurement matrix. However, traditional random measurement matrices are often impractical. Additionally, many existing deterministic binary measurement matrices fail to provide the required flexibility for practical applications. In this study, inspired by the observation that pseudo-random sequences share similar properties with random sequences, we constructed a deterministic sparse measurement matrix with a flexible measurement number based on an pseudo-random sequence—the Legendre sequence. Empirical analysis of the phase transition and an assessment of the practical features of the proposed measurement matrix were conducted. We validated the effectiveness of the proposed measurement matrix on randomly synthesized signals and images. The results of our simulations reveal that our proposed measurement matrix performs better than several other measurement matrices, particularly in terms of accuracy and efficiency.

The recognition of food images is of great significance for nutrition monitoring, food retrieval and food recommendation. However, the accuracy of recognition had not been high enough due to the complex background of food images and the characteristics of small inter-class differences and large intra-class differences. To solve these problems, this paper proposed a food image recognition method based on transfer learning and ensemble learning. Firstly, generic image features were extracted by using the convolutional neural network models (VGG19, ResNet50, MobileNet V2, AlexNet) pre-trained on the ImageNet dataset. Secondly, the 4 pre-trained models were transferred to the food image dataset for model fine-tuning. Finally, different basic learner combination strategies were adopted to establish the ensemble model and classify feature information. In this paper, several kinds of experiments were performed to compare the results of food image recognition between single models and ensemble models on food-11 dataset. The experimental results demonstrated that the accuracy of the ensemble model was the highest, reaching 96.88%, which was superior to any base learner. Therefore, the convolutional neural network model based on transfer learning and ensemble learning has strong learning ability and generalization ability, and it is feasible and practical to apply the method to food image recognition.

In the process of inspecting the self-exploding defects of power line insulators, traditional algorithms suffer from various issues such as long detection time, insufficient accuracy, and difficulties in effective detection under complex environments. To address these problems, we introduce an advanced one-stage object detection algorithm called YOLOv5s, which offers fast training and excellent detection performance. In this paper, we applied the YOLOv5s algorithm to improve the detection precision and classification accuracy of insulator self-explosions. To further enhance the YOLOv5s algorithm, we introduced a BiFPN (Bidirectional Feature Pyramid Network) module for feature fusion. This module improved the feature fusion process by learning the importance weights of different input features, considering their contributions. To tackle the challenge of detecting small objects in the self-exploding insulator dataset, we incorporated an SPD (spatial-to-depth convolution) module that focuses on capturing features in small regions and utilizes one-step convolution layers to avoid losing fine-grained information. To address the issue of high similarity between self-exploding insulator regions and intact insulator regions, we introduced an attention mechanism that concentrates attention on the defective insulator regions to gather more information about insulator defects. Experimental results validate that all three improvement methods significantly enhance detection precision. The final model achieves improvements of 2.0% in precision, 0.9% in recall, and 1.5% in average detection accuracy. Through target detection of the test dataset, insulators with self-explosion cases can be effectively detected.

Log anomaly detection is crucial for computer systems. By analyzing and processing the logs generated by a system, abnormal events or potential problems in the system can be identified, which is helpful for its stability and reliability. At present, due to the expansion of the scale and complexity of software systems, the amount of log data grows enormously, and traditional detection methods have been unable to detect system anomalies in time. Therefore, it is important to design log anomaly detection methods with high accuracy and strong generalization. In this paper, we propose the log anomaly detection method LogADSBERT, which is based on Sentence-BERT. This method adopts the Sentence-BERT model to extract the semantic behavior characteristics of log events and implements anomaly detection through the bidirectional recurrent neural network, Bi-LSTM. Experiments on the open log data set show that the accuracy of LogADSBERT is better than that of the existing log anomaly detection methods. Moreover, LogADSBERT is robust even under the scenario of new log event injections.

In the medical field, bone abnormality detection is a very important issue. Bone abnormalities include various diseases such as fractures, osteoporosis, bone tumors, and joint diseases. If these diseases are not diagnosed and treated in a timely manner, they can seriously affect the health and quality of life of patients. Artificial intelligence has made remarkable advances in Cluster analysis of medical big data, effectively mining its hidden associations to provide effective information for clinical diagnosis and medical research. However, the effectiveness of deep learning in domains with limited or no labeled data is often limited. To address this issue, we propose a novel and reliable two-stage unsupervised deep clustering framework for skeletal anomaly detection. This framework combines neural network parameters with feature clustering for collaborative learning to detect anomalies. We trained eight separate models, one for classification and seven for anomaly detection, using the MURA dataset, the largest publicly available skeletal imaging dataset. In the first stage, our approach achieved an average sensitivity and specificity of 99.76% and 99.53%, respectively. The second stage performed optimally with an average sensitivity and specificity of 83.28% and 97.56%, respectively. Our method can be easily implemented as software modules and used as a visualization tool for skeletal physicians, making it a promising approach for future development.

Unknown tag identification plays a pivotal role in radio frequency identification (RFID) systems, but it has not been fully investigated. This paper proposes a filter-based and parallel unknown tag identification protocol (FPUI) for open RFID systems. The FPUI adopts an RSQF-based fingerprint filter to reconcile the collision slots and discriminate the known tags from unknown tags. Meanwhile, it collects the IDs of unknown tags in parallel. FPUI achieves high performance through the following three steps: (1) adopting the RSQF-based filter to build an indicator vector, thus improving the space efficiency; (2) building a fingerprint filter to discriminate known tags from unknown tags, thus reducing the false positive rate; (3) employing a parallel identification scheme to collect the IDs of unknown tags, thus improving identification efficiency. The identification time of our protocol was minimized by conducting a theoretical analysis of the relevant parameters. Furthermore, the performance of our protocol was evaluated by conducting a wide range of simulation experiments. The theoretical analysis and simulation results indicated that our protocol significantly outperformed the current advanced protocols.