Jun 2026· Zanco Journal of Pure and Applied Sciences· Vol 38, pp. 203-213· 0 citations
TL;DR
It was concluded that the most accurate method of handling OoD is the Mahalanobis distance, which handled the OoD in model on real data from 85.29% to achieve a high prediction accuracy of this model.
Abstract
Detecting out-of-distribution (OoD) data is a critical process in artificial intelligence models that specialize in critical systems that require accurate results, such as medical systems, air navigation systems, and financial transaction systems. Therefore, researchers in the field of neural network programming place great importance on OoD handling. OoD detection approaches aim to identify inputs that deviate from a model’s training distribution, preventing the overconfident and false predictions that neural networks suffer from. In this paper using Post-hoc scoring approaches which as selective classification gates to handle OoD in a classification convolutional neural networks CNN model. That model is supposed to implemented in a vending machine, it’s trained on a label dataset of the Iraqi currency, that model suffers from OoD because of the convergent probabilities of the mathematical model of the neural network. With this approach, test four types; four methods were chosen to prevent out-of-distribution for the CNN model, which are: Confidence, Entropy, Cosine similarity, and Mahalanobis distance. Mathematical calculations were performed for each method with the tested data for the model and found threshold value to make as gate to pass currency classes and reject non-currency classes. Then the results of these gates comparing and analysis, it was concluded that the most accurate method of handling OoD is the Mahalanobis distance, which handled the OoD in model on real data from 85.29% to achieve a high prediction accuracy of this model.
Transfer learning has become a key technique for improving the accuracy of neural networks in low-resource, low-data environments. The quantitative comparative analysis of the pre-trained models includes ResNet50, VGG16, BERT, GPT, and the baseline CNN and LSTM models. They are compared across three different application areas: computer vision, natural language processing (NLP), and medical imaging. The five benchmark datasets used were ImageNet, CIFAR-10, SST-2, IMDB, and Chest X-Ray. All experiments used the same preprocessing pipeline and evaluation metrics (accuracy, F1 score, precision, recall, and ROC-AUC). Results showed that models trained on the pre-trained data achieved consistently greater accuracy than the baselines in all domains (9-20%) and F1-score (0.09-0.16) gains. ResNet50 achieved 92% accuracy on CIFAR-10, compared to 72% for the CNN baseline, whereas BERT hit 92% on SST-2, with 80% accuracy for LSTM. VGG16 improved the accuracy of Chest X-Ray classification from 78% to 87% and reduced training time by up to 60%. There were a few instances of minor overfitting and domain mismatch, emphasizing the need for adaptive fine-tuning strategies. The results demonstrate that transfer learning significantly improves convergence speed, generalization, and computational efficiency, making it a promising approach for AI applications across domains such as healthcare, NLP, and autonomous systems.
Abdul Sttar Ismail Wdaa, Iraq Ali Hussein, A. Ahmed· Future Technology· 0 citations
Convolutional Neural Networks (CNNs) have significantly changed image classification over the years by allowing computers to learn features directly from raw pixel data. However, deciding between building a customised model and using a pre-trained one can be a difficult task, especially when working with small datasets. In this study, we compare a custom CNN with three pre-trained models—VGG16, ResNet50, and MobileNetV2—on the CIFAR-10 dataset, which comprises 60,000 colour images (32×32 pixels) across 10 categories. We measured model performance using accuracy, precision, recall, F1-score, and training time. The results show that pre-trained models performed much better than the customised model. ResNet50 had the highest accuracy at 92.4%. However, MobileNetV2 gave the best mix of speed (1,800 seconds to train) and accuracy (90.2%). The custom CNN reached 82.3% accuracy, used less memory, and did not need image resizing. These results offer clear benchmarks for choosing models in the face of limited resources. They also demonstrate that transfer learning can achieve strong performance, while showing that custom CNNs remain useful for learning and simple tasks.
Oluwadamilare (Asabia) Joseph Omoniyi, Omotosho Olawale Jacob, Ajaegbu Chigozirim et al.· International journal of res...· 0 citations
Despite the growing need for object-detection systems, companies, governments, industries, and organisations face significant challenges in implementing them to address real-world problems. Object detection supports applications ranging from quality control and inventory management to robotics, surveillance, autonomous systems, and defect detection. Advances in deep learning have enabled notable progress through architectures such as convolutional neural networks (CNNs). This study examined the performance of two deep-learning architectures for detecting household items under controlled experimental conditions. A dataset of 2,100 images, comprising three balanced classes—bottles, boxes, and cups - was annotated and used in the experiment. The results showed that the fine-tuned VGG16 model consistently outperformed the plain CNN across the principal performance measures. Intersection over Union (IoU) increased from 57.11% for the plain CNN to 98.00% for the fine-tuned model, while loss decreased from 6.40% to 4.21%. The fine-tuned model also produced a smoother and more stable learning curve, indicating more consistent detection performance. However, this improvement required a longer training time: approximately 14,400 seconds for the fine-tuned model compared with 12,240 seconds for the plain CNN. The comparison was conducted using the same dataset, training duration, and evaluation framework for both architectures. These findings indicate a trade-off between computational efficiency and predictive performance under the experimental conditions used.
Anas Tukur Balarabe, Najib Hassan Adamu, Mahmood Umar et al.· Asian Journal of Research in...· 0 citations
Control charts do more than raise an alarm: their shapes can give an early indication of what has changed in a process. This study considers the case in which one chart window contains more than one abnormal behavior. The observed sequence is then a mixture rather than a pure pattern. We formulate this problem directly as multi-label classification. A one-dimensional CNN receives a raw-scale window of 32 observations and predicts the active elementary labels. The controlled protocol contains twelve scenarios: normal behavior, six single abnormal patterns, and five selected concurrent patterns. Raw-scale input is retained because shift patterns depend partly on level information that may be weakened by window-wise normalization. The retained training setup gives additional exposure to difficult shift and trend cases, while validation and testing remain balanced. Across five repeated trainings, the model achieved 96.11% exact match accuracy, 96.41% precision, 96.46% recall, 96.44% F1-score, and 1.04% Hamming loss. The 95% confidence interval for exact match was 96.05–96.17%. Additional analyses show stable performance around a decision threshold of 0.5, strong cyclic and systematic recognition, and lower performance for short shift cases. The results support direct multi-label CNN recognition for the selected protocol. Broader shift-containing mixtures, more complex combinations, varying noise conditions, and real industrial validation remain outside the scope of the present controlled study.
Mohammed Modar, Abdelilah Ganmati, O. E. Farissi· International Journal of Adv...· 0 citations
Deep Neural Networks (DNNs) deployed in high-risk domains, such as healthcare and autonomous driving, must be not only accurate but also understandable to ensure user trust. In real-world computer vision tasks, these models often operate on complex images containing background noise and are heavily annotated. To make such models explainable, Concept-based Explainable AI (CXAI) methods need to be assessed for their applicability and problem-solving capacity. In this work, we explore CXAI use cases in multi-label classification by training two DNNs, VGG16 and ResNet50, on the 20 most annotated labels in the MS-COCO dataset (Microsoft Common Objects in Context). We apply two CXAI methods, CRP (Concept Relevance Propagation) and CRAFT (Concept Recursive Activation FacTorization), to generate concept-level explanations and investigate the overall evaluations. Our analysis reveals three key findings: (1) CXAI highlights learning weaknesses in DNNs, (2) higher concept distinctiveness reduces label and concept confusion, and (3) environmental concepts expose dataset-induced biases. Our results demonstrate the potential of CXAI to enhance the understanding of model generalizability and to diagnose bias instigated by the dataset.
Pneumonia is a respiratory condition, which is critical and needs an accurate diagnosis to minimize death. The following paper introduces a deep learning model, which can be explained, to identify pneumonia in the X-ray images of the chest, using the Convolutional Neural Networks (CNN) architecture. The image resizing, normalization and data augmentation are some of the preprocessing strategies included in the proposed approach to improve the quality of data and the generalization of the model. CNN model is created to automatically identify discriminative features and binomial classification between normal and pneumonia cases with the help of a binary cross-entropy loss and a sigmoid activation function. Gradient-weighted Class Activation Mapping (Grad-CAM) is added to enhance the interpretability of the model by producing visual explanations by highlighting the areas that do affect the model predictions. The results of the experiment prove that the proposed model can show high classification rates (up to 97 percent) and can compete with those of the pretrained models like VGG16, ResNet50, and DenseNet121, at the same time, having lower computational complexity. The suggested framework offers high precision, as well as interpretability, and is an effective instrument to use in the process of computer-aided diagnosis and assist in clinical decision-making in the field of medical imaging.
Rushmitha Sebas S, R. M, J. A· 2026 6th International Confe...· 0 citations