ABSTRACT Domain adaptation can effectively enhance a model’s performance on target domain data with limited data. However, when some target domain labels are obtainable, training the model with both source and target domain data simultaneously can lead to worse performance due to the lower density of target domain data. If a large amount of target domain data is labeled without discrimination, it will necessitate a considerable expenditure of human resources. To address this issue, this paper proposes a human-in-one-loop active domain adaptation framework based on Target Domain Feature Generation to solve the problems. The oracle participates in only one iteration of data labeling, and a target domain classifier will take over the subsequent rest iterations. An image generator based on multiple CycleGANs forms an iterative co-training mechanism, which can continuously generate more high-quality labeled fake target domain data in iterations to improve the performance of the target domain classifier. The Top-N labeled data selection method with high confidence is devised to select the most accurately predicted data for labeling, reducing manual labeling workload. This framework can achieve an average accuracy of 0.8869 on six domain pairs, doubling the classical domain adaptation method DSN, requiring only a small amount of manual labeling.

ABSTRACT With the technology development, industrial defect detection based on deep learning has attracted extensive attention in the academic community. Different from general visual objects, industrial defects have the characteristics of small sample, weak visibility and irregular shape, which hinder the application of related studies. According to these problems, a few-shot object detection (FSOD) method based on Decoupled Faster R-CNN (DeFRCN) is proposed in this paper. Firstly, it includes fine-tuning processing, because of the small sample characteristics. To adapt to the invisible characteristics of defects, we introduce the Feature Pyramid Network (FPN) and Residual Attention Module (RAM) into DeFRCN, which can enhance the capture ability of multi-scale features and feature association information. Furthermore, the feature representation ability is strengthened by parallel connecting of two channels, consisting of R-CNN head, box classifier and box regression models. Finally, it is completed that the pre-training, fine-tuning and testing of the proposed network, with DAGM 2007 and NEU-DET public industrial defect datasets as the base class and flange shaft defect data collected in the laboratory as the new class. To verify the effectiveness of the proposed one, we compare them with other classical FSOD methods. The superiority of the proposed method is obvious.

ABSTRACT Dentists judge that the quality of dental treatment for each patient is very time-consuming and inefficient, lacks quantitative evaluation criteria, and is easy to cause errors. At the same time, the traditional method of extracting tooth and root canal image features based on experience is difficult to accurately extract the tooth area and root canal filling area, resulting in low accuracy of tooth and root canal segmentation, which in turn affects the accuracy of tooth treatment quality evaluation. In this paper, a deep learning convolutional neural network is used to segment the root canal filling area, tooth boundary, and the boundary between tooth and soft tissue for the real patient ‘s root canal treatment and filling image. Finally, the segmented image is quantitatively evaluated according to the multi-evaluation index of professional doctors. The experimental results show that the intelligent evaluation method of dental treatment quality combined with deep learning and multi-index decomposition proposed in this paper not only unifies the evaluation criteria of dental treatment quality but also the therapeutic effect of quantitative scoring can effectively improve the work efficiency of doctors, which has reference significance for the application of artificial intelligence in the medical field.

ABSTRACT Radar echo extrapolation is an important approach in precipitation nowcasting which utilizes historical radar echo images to predict future echo images. In this paper, we introduce the self-attention mechanism into Trajectory Gated Recurrent Unit (TrajGRU) model. Under the sequence-to-sequence framework, we have developed a novel convolutional recurrent neural network called Self-attention Trajectory Gated Recurrent Unit (SA-TrajGRU), which incorporates the self-attention mechanism. The SA-TrajGRU model which combines spatiotemporal variant structure in TrajGRU and self-attention module is simple and effective. We evaluate our approach on the Moving MNIST-2 dataset and the CIKM AnalytiCup 2017 radar echo dataset. The experimental results show that the performance of the proposed SA-TrajGRU model is comparable to other convolutional recurrent neural network models. HSS and CSI scores of the SA-TrajGRU model are higher than scores of other models under the radar echo threshold of 25 dBZ, indicating that the SA-TrajGRU model has the most accurate prediction results under this threshold.

ABSTRACT Enhancing the precision of retention ratio predictions holds profound significance for insurance industry decision-makers and those vested in advancing insurance services. Precision helps insurance companies navigate inflationary pressures and evaluate underwriting profitability, enabling reliable prognoses of future underwriting gains. As far as we know, although there have been multiple attempts to construct a predictive model for retention ratio, none of these attempts have used combining models or studied the Egyptian market. Therefore, this study contributes significantly to this developing field by providing combining models, which combined statistical time series models such as Exponential Smoothing (ES), and Autoregressive Integrated Moving Average (ARIMA), with Adaptive Neuro-Fuzzy Inference System (ANFIS). Two different types of combinations are employed with these models. Furthermore, the study introduces three ensemble models designed for the purpose of predicting the retention ratio within the Egyptian insurance market. Dataset was carefully gathered from the EFSA’s annual reports, focused on the property-liability insurance sector within the Egyptian insurance market and covers the time period from 1989 to 2021. Next, the proposed models are assessed employing well-established statistical assessment metrics, namely, Mean Absolute Error (MAE), Mean Absolute Percent Error (MAPE), R Square (R2), and Root Mean Square Error (RMSE). The results show that combining and ensemble methods improve predicted accuracy. A multi-linear regression-based ensemble model that combines ARIMA, ES, and ANFIS models outperforms both single and combined models in robustness. The article concludes that the insurance industry can greatly benefit from modern predictive methods to make sound decisions.

ABSTRACT The Singing Muscle Ability Training System (SMATS) represents a cutting-edge technological solution dedicated to honing an individual’s vocal prowess. By employing a sophisticated combination of techniques, exercises, and methodologies, this system strategically targets and cultivates the muscle groups integral to singing. In its innovative approach, SMATS integrates Artificial Intelligence (AI) to elevate vocal capabilities further. The proposed SMATS-AI-VONN-GSOA stands out by incorporating the Variational Onsager Neural Network (VONN) alongside the Golden Search Optimization Algorithm (GSOA). This amalgamation tailors training routines based on the user’s progress and preferences, emphasizing the development of muscle memory and control for enhanced vocal performance. Noteworthy is the system’s capacity to analyze data through AI, enabling the creation of personalized training plans. In comparative evaluations, the SMATS-AI-VONN-GSOA method demonstrates a significant performance boost, surpassing existing methods like SMATS-AI-CNN, SMATS-AI-DNN, and SMATS-AI-DNN-HSV by 26.78%, 29.55%, and 21.41% in accuracy, respectively.

ABSTRACT This research addresses critical gaps in Mobile Ad hoc Networks (MANETs) by proposing a hybrid secure cluster-based routing algorithm, focusing on enhancing network security, robustness, and reliability through multipath routing. Methodologically, the approach integrates Convolutional Neural Networks (CNN) for optimal path routing and Emperor Penguin Optimization (EPO) for clustering, introducing a novel combination for efficient cluster head selection. A novel contribution lies in the development of a prediction technique utilizing a trust assessment algorithm to calculate direct trust ratings at each node, incorporating fuzzy values between zero and one. Trust values are further influenced by node performance, adding a dynamic dimension to the trust evaluation process. Key novelties include the emphasis on energy efficiency, network longevity, remaining energy, security level, bandwidth, and packet delivery ratio as evaluation criteria. The proposed CNN-EPO model demonstrates superior results compared to traditional routing protocols, achieving a remarkable 95% energy efficiency, a heightened security level of 99%, and a throughput reaching up to 8 Mbps. Additionally, the Packet Delivery Ratio (PDR) attains close to 99% and routing overhead remains below 0.5, ensuring efficiency in challenging network scenarios with 50 adversaries. In summary, this research contributes a comprehensive solution to MANET challenges, introducing a novel hybrid routing algorithm, incorporating advanced methodologies for path optimization and clustering. These outcomes highlight how important the suggested strategy is to improve the existing state of the art in MANETs.

ABSTRACT A novel Electromagnetic Band Gap (EBG) structure is proposed to minimize common connector between closely spaced rectangular patch antennas. The 2D EBG architecture is placed on the horizontal plane of a 1/2 rectangle patch antenna array, reducing mutual coupling by over 20 dB without compromising far-field performance, gain, or bandwidth. By optimizing the EBG structure using a genetic algorithm, noise is reduced by −60 dB compared to −25 dB without it, improving the isolation of the antenna array. Inserting 3 EBG modules between the two patches decreases the 2.45 GHz mutual coupling to −60 dB, showcasing the significant enhancement in isolation. This MIMO antenna design with an EBG structure is suitable for wireless LAN communication systems, offering improved signal separation and performance. The proposed method integrates square patch antennas operating at 24 GHz with a metamaterial EBG structure on a Rogers R04350B substrate, using a genetic algorithm for optimization. Experimental results demonstrate over 22 dB improvement in isolation between antenna elements, maintaining consistent radiation patterns across frequencies.

ABSTRACT Money laundering (ML) is a critical source of extracting the money illegally from the financial system. It is linked to various types of crimes, including corruption, exploitation of a specific community, drug use, and many others. Detection of ML operations is a difficult task on a global scale due to the large volume of financial transactions. However, it also allows criminals to use financial systems to carry out fraudulent transactions. It mainly concern minimizing the potentially risks associated with money laundering. Anti-money laundering-(AML) tools based on AI-driven applications are now tracking transactions to overcome this challenge. A total of 112 research papers are assessed to identify the literature’s gaps and suggest new directions for the research area accordingly. The findings of this systemic literature review work will not only open new paths for the research community, but will also assist the state agencies in developing an optimal AML system to counter these major issues and provide a healthy environment for their residents. This article seeks to assess the existing situation from various angles and open up new pathways for future research directions to investigate and build high levels of authenticity and security in the financial industry using artificial intelligence (AI).

ABSTRACT Water pollutions can severely affect water environment, causing water quality degradation and threatening aquatic wildlife. Deemed as guideline for maximum environmental impact assessment, water pollution peak concentration (WPPC) has been intensively studied to organize effective countermeasures. In this study, a back propagation artificial neural network (BPANN) coupled with genetic algorithm (GA) was constructed to predict peak concentrations. Compared with BPANN, multiple linear regressions model (MLRM) and step-wise multiple linear regressions model (SMLRM), GA-BPANN model showed superior accuracy in both simulating and predicting peak concentrations (R 2  = 0.93 and 0.67 0.69 respectively). In 12 peak concentration cases, GA-BPANN model’s mean absolute relative error (MARE) ranges from 0.0 to 0.58, averaged at 0.09, significantly lower than BPANN, MLRM and SMLRM (MARE = 0.29, 0.45 and 0.48). Further analysis revealed that GA-BPANN model can be used as an effective and efficient tool for water quality simulation and early warning prediction.