This paper proposes a dual-encoder framework that learns syntax- and semantics-aligned sentence embeddings using contrastive learning. The proposed Syntax-Semantic Contrastive Pretraining (SSCP) model employs two transformer encoders to separately model syntactic structure and semantic content, which are aligned in a shared embedding space via a symmetric contrastive objective. Across standard benchmarks, SSCP consistently outperforms strong baselines such as BERT, SimCSE, and SyntaxBERT. In particular, SSCP improves Spearman correlation on STS-B by +1.4 points over SimCSE, increases PAWS paraphrase accuracy by +2.5 points, and achieves 96.1% accuracy on TREC question classification, exceeding existing syntax-aware models. Probing experiments further show gains of up to +4–5 points on syntactic structure prediction tasks, confirming that SSCP preserves grammatical information while maintaining semantic robustness. These results demonstrate that explicitly aligning syntactic and semantic views yields representations that are more discriminative, interpretable, and robust than single-view or syntax-augmented approaches, positioning SSCP as a principled multi-view pretraining strategy for structure-aware language understanding.

Microscopic examination of stained blood smears remains a reference method for malaria diagnosis, but is labour-intensive and operator-dependent. We evaluated image-based machinelearning models for parasite detection in single-cell images and report a deployment-oriented configuration. We used the NIH “Cell Images for Malaria” set (27,558 labelled cells; 13,780 parasitised; 13,778 uninfected). We compared a lightweight CNN against SVM and Naïve Bayes baselines. Variants with HOG features and dimensionality reduction (PCA/LDA) were explored for analysis only. The best CNN achieved 95.0% accuracy, 94.7% precision, 95.3% recall, and 95.0% F1 on the held-out test set, with a ROC-AUC of 0.91. We showed that a compact CNN trained on publicly available cell images yields high diagnostic performance and can be embedded in a simple desktop GUI for point-of-care assistance.

Recent advancements in artificial intelligence (AI) and machine learning (ML) have significantly transformed healthcare education by enhancing efficiency, accuracy, and standardization in patient data analysis, while also being applied to explore the impacts of test anxiety and self-efficacy on academic achievement. A study using a feedforward artificial neural network, specifically Multi-Layer Perceptrons (MLPs), identified four critical factors for academic success: having a positive mindset (AR1, importance rate 0.997), monitoring and evaluating achievements (AR5, 0.996), a well-thought-out plan (AR2, 0.981), accountability for progress (AR3), and acknowledging stress and negative emotions (AR4). Additionally, the study highlighted key test anxiety factors, such as visible signs of nervousness before a test (AT1, 0.146) and heightened nervousness during exams (AT7, 0.126), which impact academic performance. Using machine learning, distinct patterns in academic achievement and test anxiety were identified across student groups, forming a “blueprint” for targeted interventions to improve academic outcomes. A predictive model was also developed to

This research investigates the development of predictive models for recognizing dual-hand gestures using surface electromyography (sEMG) signals. The main focus was put on dualhand gestures, an area that remains a gap in research and is essential to advance technologies in human-computer interaction. Most of the previous systems are designed for single-hand gestures, leading to low accuracy for more complex dual-hand gestures. The basic objective of this research is to develop a robust predictive model which enhances the recognition of dual-hand gestures significantly through analysis of the NinaPro DB1 database. We used two approaches to explore into the problem: one using pre-trained models and the other with

Dimensionality reduction techniques are fundamental for analyzing and visualizing highdimensional data. With established methods like t-SNE and PCA presenting a trade-off between representational power and interpretability. This paper introduces a novel approach that bridges this gap by combining the interpretability of linear methods with the expressiveness of non-linear transformations. The proposed algorithm constructs a non-linear mapping between high-dimensional and low-dimensional spaces through a combination of linear transformations, each weighted by Gaussian functions. This architecture enables complex non-linear transformations while preserving the interpretability advantages of linear methods, as each transformation can be analyzed independently. The resulting model provides both powerful dimensionality reduction and transparent insights into the transformed space. Techniques for interpreting the learned transformations are presented, including methods for identifying suppressed dimensions and how space is expanded and contracted. These tools enable practitioners to understand how the algorithm preserves and modifies geometric relationships during dimensionality reduction. To ensure the practical utility of this algorithm, the creation of user-friendly software packages is emphasized, facilitating its adoption in both academia and industry.

With the fast growth of financial transaction fraud, there is a need for advanced detection systems capable of real-time analysis. Rule-based and machine-learning approaches to fraud traditionally suffer from being unable to adapt to changing fraud patterns, returning very high back result rates and much inefficiency in the security of financial operations. However, convolutional neural networks (CNNs) and recurrent neural networks (RNNs) methods are suitable, but they lack adaptability and interpretability. This paper proposes an Adaptive Transactional Anomaly Detection Network (ATAD-Net), a new deep learning (DL) framework for improving fraud detection accuracy, minimizing false positives, and guaranteeing real-time adaptability. ATAD-Net dynamically adjusts to evolving fraud tactics by integrating CNNs for local pattern recognition and Long Short-Term Memory (LSTM) for sequential transaction analysis. After training and testing the model using the IEEE CIS Credit Card Fraud Detection Dataset, a large-scale benchmark for evaluating financial fraud detection models, the accuracies of the different models were assessed.

Occlusion removal is one of the challenging tasks in face recognition system. Facial occlusion is nothing but face object is hiding the other object which is one important factor that influences how well facial recognition works. There is insufficient research regarding the issue of occlusion. Researchers proposed several methods for removal of occlusion to recognize faces. But are unable to produce results that are realistic for the removal of significant objects, particularly in images of faces. The performance test of face recognizer recognition system degrades when dealing with masked faces. The objective of this research work to remove mask object from face images to recognizes faces easily. We divide the problem into two modules: occlusion detection module and occlusion removal module. We proposed deep learning-based technique called GAN-based network. Two discriminators are used in a GAN-based network; one discriminator helps in learning the face’s global structure, while the other learns the deep missing region. MaskedFace-CelebA dataset is used to evaluate our model that is utilised to compare masked images in pairs with ground truth in order to verify faces. The evaluation metric used in this research is SSIM. Using the proposed method can be useful task to generate new unmasked face samples and help criminal investigation agencies1to reveal the identity of criminals who committed the crimes behind the mask. Once the occlusion of face is removed, the system can easily recognize the face and the biometric authentication during COVID-19 pandemic no need to remove masks while taking biometric attendance, this will reduce the spread of COVID-19 disease.

Globally, Artificial Intelligence (AI) is changing higher education, and African institutions are becoming more and more involved in this digital revolution. AI applications such as predictive analytics, automated grading, and intelligent tutoring systems offer promising opportunities to enhance research, teaching, learning, and administration—particularly in settings with limited resources. However, the incorporation of AI into African higher education also raises significant socio-technical and ethical issues. Academic dishonesty, algorithmic bias, data privacy, and unequal access to digital infrastructure are some of the issues that risk exacerbating existing educational inequalities. Furthermore, concerns regarding digital sovereignty, fairness, and cultural relevance are raised by the dependence on AI tools created in the Global North. This article highlights the dual potential for both empowerment and marginalization as it critically analyzes the advantages and hazards of implementing AI in African higher institutions. It calls for a responsible, inclusive, and culturally grounded approach to AI in African higher education through a contextual and ethical lens, using concepts like digital justice, the ethics of technology, and decolonial theory.

The retail industry has evolved from Retail 1.0 to Retail 4.0 in response to the advancements brought about by the industrial revolution. After the COVID-19 epidemic, the customer had formed the habit of online shopping. The retail enterprise business model change from traditional retail to new retail and omni-channel retail. Artificial intelligence (AI), big data, Internet of Things (IoT), and other technologies make this change an inevitable trend in the era of Retail 4.0. The present study provides an overview of the current state of China’s retail industry in the post-epidemic era.It explores how retail enterprises are transitioning their business models from traditional retail to new retail and omni-channel retail within the framework of Retail 4.0. In conjunction with an analysis of the development and research status of big data, this paper explores the application of big data in China’s retail industry across five key areas, namely risk management, category management, customer relationship management, logistics management, and market analysis, then summaries its challenges, limitations and future trends. The purpose of this paper is to provide enterprises with a clear understanding of their current stage and offer guidance for decision-making in the application of big data.

Convolutional neural networks (CNNs), the most important deep learning networks for computer vision, have undergone a series of developments and improvements for image-related tasks such as object recognition, image classification, semantic segmentation, etc. However, in the field of natural language processing (NLP), the novel attention-based network Transformer had a profound impact on machine translation, which subsequently led to a boom in attention-based models for computer vision. State-of-the-art models with attention have already shown good performance for computer vision tasks due to the sophisticated design of network architectures and advanced computational efficiency techniques. For example, selfattention learns relationships between segments or words in different positions compared to the current performance of convolutional neural networks. Inspired by Vision Transformer (ViT), we propose a simple novel transformer architecture model, called Flexible Transformer, which inherits the properties of attention-based architectures and is flexible for inputs of arbitrary size. Besides self-attention, the inputs in ViT are not pre-processed, such as resizing or cropping, but kept intact without altering them, which could lead to distortion or loss of information. In this paper, we want to present a novel and simple architecture that meets these requirements. Compared to state-of-arts, our model processes inputs with arbitrary image sizes without any pre-processing and pretraining costs. Also, the results of the experiments show that the model can potentially provide good results with high accuracy despite limited resources. Even though the results of Flexible Transformer are not as accurate as those of Vision Transformer, they show the potential of a model with high performance in image classification tasks with variable size images. The significance of the research opens up possibilities for dealing with primitive images in deep learning tasks. Based on the original inputs, reliable results with good accuracy could be obtained if the proposed model is optimized and further trained on large datasets.