Railway-Fastener Point Cloud Segmentation and Damage Quantification Based on Deep Learning and Synthetic Data Augmentation

As failures in rotating machines can have serious implications, the timely detection and diagnosis of faults in these machines is imperative for their smooth and safe operation. Although deep learning offers the advantage of autonomously learning the fault characteristics from the data, the data scarcity from different health states often limits its applicability to only binary classification (healthy or faulty). This work proposes synthetic data augmentation through virtual sensors for the deep learning-based fault diagnosis of a rotating machine with 42 different classes. The original and augmented data were processed in a transfer learning framework and through a deep learning model from scratch. The two-dimensional visualization of the feature space from the original and augmented data showed that the latter’s data clusters are more distinct than the former’s. The proposed data augmentation showed a 6–15% improvement in training accuracy, a 44–49% improvement in validation accuracy, an 86–98% decline in training loss, and a 91–98% decline in validation loss. The improved generalization through data augmentation was verified by a 39–58% improvement in the test accuracy.

Overhead depth map measurements capture sufficient amount of information to enable human experts to track pedestrians accurately. However, fully automating this process using image analysis algorithms can be challenging. Even though hand-crafted image analysis algorithms are successful in many common cases, they fail frequently when there are complex interactions of multiple objects in the image. Many of the assumptions underpinning the hand-crafted solutions do not hold in these cases and the multitude of exceptions are hard to model precisely. Deep Learning (DL) algorithms, on the other hand, do not require hand crafted solutions and are the current state-of-the-art in object localization in images. However, they require exceeding amount of annotations to produce successful models. In the case of object localization these annotations are difficult and time consuming to produce. In this work we present an approach for developing pedestrian localization models using DL algorithms with efficient weak supervision from an expert. We circumvent the need for annotation of large corpus of data by annotating only small amount of patches and relying on synthetic data augmentation as a vehicle for injecting expert knowledge in the model training. This approach of weak supervision through expert selection of representative patches, suitable transformations and synthetic data augmentations enables us to successfully develop DL models for pedestrian localization efficiently.

Accurate diagnosis of melanoma using hematoxylin and eosin (H&E)-stained histological images is often challenged by the scarcity and imbalance of biomedical datasets, limiting the performance of deep learning models. This study investigates the impact of synthetic image generation, via generative adversarial networks (GAN), on training automatic classifiers based on the ResNet-18 architecture. Two experimental setups were designed: one using only real images and another combining real images with synthetic ones of the melanocytic nevus class to balance the dataset. Models were trained and evaluated at resolutions up to 1024 × 1024 pixels, employing standard classification metrics and the Fréchet Inception Distance (FID) to assess the quality of the generated images. The results suggest that although mixed models do not consistently outperform those trained exclusively on real data, they achieve competitive performance, particularly in terms of specificity and reduction in false negatives. This study supports the use of synthetic data as a complementary tool in scenarios where the acquisition of new samples is limited and lays the groundwork for future research in conditional generation and synthesis of malignant samples. In our best-performing model (1024 × 1024 px, 50 epochs, mixed dataset), we achieved an accuracy of 96.00%, a specificity of 97.00%, and a reduction in false negatives from 80 to 75 cases compared with real-only training. These results highlight the potential of synthetic augmentation to improve clinically relevant outcomes, particularly in reducing missed melanoma diagnoses.

RNAs are a vast reservoir of untapped drug targets. Structure-based virtual screening (VS) identifies candidate molecules by leveraging binding site information, traditionally using molecular docking simulations. However, docking struggles to scale with large compound libraries and RNA targets. Machine learning offers a solution but remains underdeveloped for RNA due to limited data and practical evaluations. We introduce a data-driven VS pipeline tailored for RNA, utilizing coarse-grained 3D modeling, synthetic data augmentation, and RNA-specific self-supervision. Our model achieves a 10,000x speedup over docking while ranking active compounds in the top 2.8% on structurally distinct test sets. It is robust to binding site variations and successfully screens unseen RNA riboswitches in a 20,000-compound in-vitro microarray, with a mean enrichment factor of 2.93 at 1%. This marks the first experimentally validated success of structure-based deep learning for RNA VS.

The outbreak of novel coronavirus disease 2019, also called COVID-19, in Wuhan, China, began in December 2019. Since its outbreak, infectious disease has rapidly spread across the globe. The testing methods adopted by the medical practitioners gave false negatives, which is a big challenge. Medical imaging using deep learning can be adopted to speed up the testing process and avoid false negatives. This work proposes a novel approach, COVID-19 GAN, to perform coronavirus disease classification using medical image synthesis by a generative adversarial network. Detecting coronavirus infections from the chest X-ray images is very crucial for its early diagnosis and effective treatment. To boost the performance of the deep learning model and improve the accuracy of classification, synthetic data augmentation is performed using generative adversarial networks. Here, the available COVID-19 positive chest X-ray images are fed into the styleGAN2 model. The styleGAN model is trained, and the data necessary for training the deep learning model for coronavirus classification is generated. The generated COVID-19 positive chest X-ray images and the normal chest X-ray images are fed into the deep learning model for training. An accuracy of 99.78% is achieved in classifying chest X-ray images using CNN binary classifier model.

Customer churn is the loss of business clients to a competitor. Since keeping current clients is more economical than finding new ones, customer retention measures such as churn detection are now essential aspects of modern banking strategy. However, many existing studies rely heavily on conventional machine learning approaches such as Support Vector Machines, Logistic Regression , Random Forest, etc., often neglecting the deeper learning capabilities of neural networks. Also, the repeated use of the same small dataset by the banking studies may limit the improvement of the models’ generalisation. To address these gaps, this study presents a method that integrates deep learning for customer churn detection and a soft and hard voting classifier ensemble embedded with the best performing models over the years for results comparison, supported by a synthetic data augmentation method for model improvement. The study utilised a secondary banking churn dataset from Kaggle, which contained 10,000 unique customer records. To address the dataset limitations, a Conditional Tabular Generative Adversarial Network (CTGAN) model was used to generate an additional 10,000 records, expanding the dataset used for the study to 20,000 rows. Data preprocessing steps were done before training, including oversampling using Synthetic Minority Oversampling Technique (SMOTE). Model development and analysis processes were implemented using Python programming language with prominent libraries and frameworks on Google Colab. In this study, a Feedforward neural network and a soft and hard voting classifier were developed. The voting classifier ensembles integrated three prominent classifiers: Random Forest, XGBoost, and Logistic Regression. The performances were evaluated using Accuracy, F1 Score, and Area Under ROC Curve as metrics. Results show that while the Feedforward Neural Network achieved strong predictive performance with an accuracy of 88.23%, an F1 Score of 87.83% and an AUC of 94.73%, the ensemble approaches performed slightly better as the soft voting classifier delivered the best results, obtaining an accuracy of 89.46%, F1 Score of 88.92% and AUC of 95.40% showing the advantage of combining multiple models to leverage complementary strengths. After comparison with past studies, the proposed models did not surpass the very best outcomes. However, they remain highly competitive, achieving performance levels that are on par with or exceed many earlier works. The contribution of this work is to show how synthetic data augmentation, enhanced preprocessing, deep learning techniques, and machine learning ensembles can improve churn detection in banking studies. Banking institutions can utilise the results from this study to accurately detect churn, supporting proactive customer retention strategies, targeted marketing, and personalised financial services, thereby reducing revenue losses.

Infrared (IR) images are essential to improve the visibility of dark or camouflaged objects. Object recognition and segmentation based on a neural network using IR images provide more accuracy and insight than color visible images. But the bottleneck is the amount of relevant IR images for training. It is difficult to collect real-world IR images for special purposes, including space exploration, military and fire-fighting applications. To solve this problem, we created color visible and IR images using a Unity-based 3D game editor. These synthetically generated color visible and IR images were used to train cycle consistent adversarial networks (CycleGAN) to convert visible images to IR images. CycleGAN has the advantage that it does not require precisely matching visible and IR pairs for transformation training. In this study, we discovered that additional synthetic data can help improve CycleGAN performance. Neural network training using real data (N = 20) performed more accurate transformations than training using real (N = 10) and synthetic (N = 10) data combinations. The result indicates that the synthetic data cannot exceed the quality of the real data. Neural network training using real (N = 10) and synthetic (N = 100) data combinations showed almost the same performance as training using real data (N = 20). At least 10 times more synthetic data than real data is required to achieve the same performance. In summary, CycleGAN is used with synthetic data to improve the IR image conversion performance of visible images.

Mathematical Problem Solving in Arabic: Assessing Large Language Models

Animal-vehicle collisions (AVCs) pose a significant threat to road safety, wildlife conservation, and transportation systems worldwide. Advances in artificial intelligence (AI) and computer vision have enabled intelligent detection and mitigation systems aimed at reducing such collisions. This review synthesizes the current state of AI-powered AVC prevention systems, examining deep learning architectures, multimodal sensor technologies, real-time processing frameworks, and system-level integration strategies. We analyze the transition from traditional computer vision methods to modern deep neural networks, evaluate sensor fusion approaches, and assess existing wildlife detection datasets and benchmarking practices. Key technical challenges are identified, including environmental variability, long-range detection constraints, dataset scarcity, cross-species generalization limitations, and real-time safety requirements. Rather than framing AVC prevention solely as an object detection task, this review conceptualizes it as a safety-critical perception and risk assessment pipeline operating under strict latency and deployment constraints. Persistent gaps in wildlife-specific detection, standardized evaluation protocols, and scalable edge deployment are discussed. To organize these insights, we present WildSafe-Edge as a conceptual reference architecture derived from the literature, synthesizing system-level design considerations and highlighting open research directions. Future research directions include transfer learning, synthetic data augmentation, vehicle-to-everything (V2X) integration, and edge-centric architectures to enable robust, real-world collision mitigation systems.

Pneumonia remains one of the leading causes of death, particularly in children, and it affects a significant portion of the global population. Chest X-rays are the primary diagnostic tool for detecting pneumonia; however, accurate interpretation of these images is challenging, even for trained radiologists. This study presents an efficient deep learning-based model for pneumonia detection designed to assist radiologists in making more accurate diagnoses. The model utilizes transfer learning techniques with pre-trained architectures, specifically ResNet50 and MobileNetV2, to leverage the learned features for optimal performance. Generative Adversarial Networks (GANs) were employed to augment the training dataset, producing over 1 lakh synthetic chest X-ray images to improve model generalization. The final model was trained and evaluated using a combined dataset of real and synthetic data, which demonstrated high performance. The model achieved 98% test accuracy, significantly reducing computational complexity compared with other models, making it suitable for deployment in resource-constrained environments. Our findings suggest that this model can be used for rapid and reliable pneumonia diagnosis, offering a valuable tool to support healthcare professionals in clinical decision making.

Accurate ecological prediction is critical for sustainable environmental management and carbon cycle assessment, yet model development is often constrained by limited datasets and inconsistent preprocessing practices. Reliable litterfall prediction plays a key role in understanding nutrient cycling and supporting sustainable forest ecosystem management. Although gradient boosting models have shown promising performance in ecological applications, structured evaluations integrating preprocessing strategies with synthetic data augmentation remain limited under data-scarce conditions. This study proposes the Hybrid Preprocessing and Augmented Boosting Framework (HPABF), which combines multi-stage preprocessing—including MICE imputation, log transformation, and feature engineering—with synthetic data augmentation to enhance predictive robustness. The framework was evaluated across eight machine learning models using a 968-sample forest ecological dataset. To mitigate data scarcity, 5000 synthetic samples were generated while preserving the statistical distribution and multivariate structure of the original data (91% fidelity). Fractal dimension analysis was further introduced as a geometric validation metric to assess prediction structure and stability beyond conventional performance measures. Within the HPABF, gradient boosting models achieved a 7% improvement over baseline performance (R2 = 0.96, MAE = 0.06) under cross-validation strategies designed to reduce overfitting. Training with synthetic data further improved predictive accuracy (R2 = 0.98), demonstrating the framework’s effectiveness for data-scarce ecological applications. By improving prediction reliability under limited data conditions, the proposed framework supports more accurate environmental monitoring, informed decision-making, and sustainable management of forest ecosystems.

Deep learning-based traffic prediction models require vast amounts of data to learn embedded spatial and temporal dependencies. The inherent privacy and commercial sensitivity of such data has encouraged a shift towards decentralised data-driven methods, such as Federated Learning (FL). Traditional machine learning models capture spatial and temporal relationships in centralised data. In reality, traffic data is likely distributed across separate data silos owned by multiple stakeholders. In this work, a cross-silo FL setting is motivated to facilitate stakeholder collaboration for optimal traffic flow prediction applications. This work introduces FedTPS, a novel framework which augments each client’s local dataset with synthetic data generated by a federated diffusion-based model. The proposed framework is evaluated on three large-scale real-world datasets and assessed against various generative and spatio-temporal prediction models, including a novel prediction model we introduce which leverages Temporal and Graph Attention mechanisms to learn embedded Spatio-Temporal dependencies. Experimental results show that FedTPS outperforms several FL baselines with respect to global model performance.

Intelligent Cotton Yield Prediction Using Hybrid Random Forest–LSTM Model with Synthetic Data Augmentation

The integration of Deep Belief Networks (DBNs) into forensic linguistics enhances AI-driven policing in smart cities by automating witness testimony analysis. DBNs address limitations of traditional methods, such as subjectivity and inefficiency, by detecting linguistic markers of deception and inconsistency. AI tools have accelerated investigations by analyzing digital evidence, though specific architectures like DBNs were not disclosed in these cases. Ethical considerations, including algorithmic bias and data privacy, are addressed through strategies like adversarial debiasing and synthetic data augmentation. Collaboration between law enforcement, AI experts, and policymakers ensures ethical deployment and inclusive dataset curation. Future directions include expanding DBN applications to real-time threat detection and multilingual forensic analysis. DBNs improve urban security by enhancing testimony credibility assessments, fostering community trust, and aligning AI-driven policing with ethical standards.

Medical doctors and biological researchers often rely on imaging techniques to analyze tissue structures and cellular morphology.For example, pathologists use H&E-stained images for disease diagnosis and grading.Since cells are fundamental components in many imaging modalities, an automatic and accurate cell segmentation pipeline can greatly benefit biological research and medical diagnostics.In recent years, deep learning has significantly advanced cell segmentation, with foundation models (FMs) emerging as a promising approach.One such model, CellViT, based on the Vision Transformer (ViT) architecture, has gained attention for its strong generalization capabilities across different cell imaging modalities and datasets.In this study, we evaluate the accuracy and robustness of CellViT on the publicly available MoNuSeg dataset for cell segmentation.While the model produces reasonable results, its performance can be further improved.To enhance CellViT, we develop a fine-tuning pipeline that integrates both manually labeled cell images (with precise cell annotations) and synthetic images generated by a diffusion model, DiffInfinite (with pseudo annotations).Our pipeline incorporates key components, including FM fine-tuning, human-in-the-loop feedback for using synthetic images, and FM training on a combination of real labeled and pseudo-labeled generated images.Experimental results demonstrate significant improvements in segmentation accuracy and generalization when applying our method to CellViT.This study highlights the potential of enhancing foundation models through fine-tuning and synthetic data augmentation, paving the way for more robust and accurate biomedical image analysis.