Optimised deep learning framework for glass defect detection using ResNet50V2 and Bayesian Optimisation

Brain tumor is one of the most aggressive diseases nowadays, resulting in a very short life span if it is diagnosed at an advanced stage. The treatment planning phase is thus essential for enhancing the quality of life for patients. The use of Magnetic Resonance Imaging (MRI) in the diagnosis of brain tumors is extremely widespread, but the manual interpretation of large amounts of images requires considerable effort and is prone to human errors. Hence, an automated method is necessary to identify the most common brain tumors. Convolutional Neural Network (CNN) architectures are successful in image classification due to their high layer count, which enables them to conceive the features effectively on their own. The tuning of CNN hyperparameters is critical in every dataset since it has a significant impact on the efficiency of the training model. Given the high dimensionality and complexity of the data, manual hyperparameter tuning would take an inordinate amount of time, with the possibility of failing to identify the optimal hyperparameters. In this paper, we proposed a Bayesian Optimization-based efficient hyperparameter optimization technique for CNN. This method was evaluated by classifying 3064 T-1-weighted CE-MRI images into three types of brain tumors (Glioma, Meningioma, and Pituitary). Based on Transfer Learning, the performance of five well-recognized deep pre-trained models is compared with that of the optimized CNN. After using Bayesian Optimization, our CNN was able to attain 98.70% validation accuracy at best without data augmentation or cropping lesion techniques, while VGG16, VGG19, ResNet50, InceptionV3, and DenseNet201 achieved 97.08%, 96.43%, 89.29%, 92.86%, and 94.81% validation accuracy, respectively. Moreover, the proposed model outperforms state-of-the-art methods on the CE-MRI dataset, demonstrating the feasibility of automating hyperparameter optimization.

Corrosion in metallic materials is a critical challenge in maintenance and safety, and traditional visual inspection methods are often time-consuming, labor-intensive, and dependent on human expertise, highlighting the need for more efficient and reliable solutions. Deep learning, particularly convolutional neural networks (CNNs), provides a promising approach by enabling automated and accurate image-based classification. This study investigates binary image classification of corrosion using four pre-trained CNN architectures, namely ResNet50, MobileNetV2, NASNetMobile, and EfficientNetV2B0, and integrates explainable artificial intelligence (XAI) techniques to provide interpretability and insight into each model’s decision-making process. A curated dataset of 4012 images, divided between corroded and non-corroded surfaces, was pre-processed, and augmented images resulted in a total of 9636 images used to train and evaluate the models. Performance was assessed through accuracy, confusion matrices, computational timing, receiver operating characteristic curves, precision–recall curves, and Cohen’s Kappa. In this paper, Gradient-weighted Class Activation Mapping (Grad-CAM) visualizations are incorporated as an XAI technique to provide interpretable insight into the model’s reasoning process, enabling clear identification of corrosion regions and offering justification for each prediction produced by the system. A key contribution of this work is the integration of Grad-CAM to enhance explainability. The results showed that EfficientNetV2B0 demonstrates stable training with minimal sign overfitting compared to other models. MobileNetV2 achieved the lowest time to train with the datasets given, and ResNet50 achieved the highest classification performance in terms of confusion matrix, with an accuracy of 96.58%. Through Grad-CAM reasoning, EfficientNetV2B0 shows a specific high activation towards corroded regions compared to the other three models that were evaluated.

Accurately determining sex using features like facial bone profiles and teeth is crucial for identifying unknown victims. Lateral cephalometric radiographs effectively depict the lateral cranial structure, aiding the development of computational identification models. This study develops and evaluates a sex prediction model using cephalometric radiographs with several convolutional neural network (CNN) architectures. The primary goal is to evaluate the model's performance on standardized radiographic data and real-world cranial photographs to simulate forensic applications. Six CNN architectures-VGG16, VGG19, MobileNetV2, ResNet50V2, InceptionV3, and InceptionResNetV2-were employed to train and validate 340 cephalometric images of Indonesian individuals aged 18 to 40 years. The data were divided into training (70%), validation (15%), and testing (15%) subsets. Data augmentation was implemented to mitigate class imbalance. Additionally, a set of 40 cranial images from anatomical specimens was employed to evaluate the model's generalizability. Model performance metrics included accuracy, precision, recall, and F1-score. CNN models were trained and evaluated on 340 cephalometric images (255 females and 85 males). VGG19 and ResNet50V2 achieved high F1-scores of 95% (females) and 83% (males), respectively, using cephalometric data, highlighting their strong class-specific performance. Although the overall accuracy exceeded 90%, the F1-score better reflected model performance in this imbalanced dataset. In contrast, performance notably decreased with cranial photographs, particularly when classifying female samples. That is, while InceptionResNetV2 achieved the highest F1-score for cranial photographs (62%), misclassification of females remained significant. Confusion matrices and per-class metrics further revealed persistent issues related to data imbalance and generalization across imaging modalities. Basic CNN models perform well on standardized cephalometric images but less effectively on photographic cranial images, indicating a domain shift between image types that limits generalizability. Improving real-world forensic performance will require further optimization and more diverse training data. Not applicable.

Rotated target recognition is a challenge for Convolutional Neural Networks (CNN), and the current solution is to make CNN rotational invariant through data augmentation. However, data augmentation makes CNN easy to overfit small scale sonar image datasets, and increases its numbers of parameters and training time. This paper proposes to recognize rotated targets of sonar images using a novel CNN with Rotated Inputs (RICNN), which doesn’t need data augmentation. During training, RICNN was trained with sonar images of targets only at one orientation, which avoid it to learn multiple rotated versions of the same targets, and reduces both number of parameters and training time of CNN. During testing, RICNN calculated classification scores for each test image and its all-possible rotated versions. The max of these classification scores were used to simultaneously estimate the category and orientation of each target. Besides, to improve the generalization of RICNN on imbalanced sonar datasets, this paper also designs an imbalanced data sampler. Experiments on a self-made small, imbalanced sonar image rotated target recognition dataset show that the improved RICNN achieves 4.25% higher classification accuracy than data augmentation, and reduces the number of parameters and training time to 2.25% and 19.2% of that of data augmentation method. Moreover, RICNN achieves comparable orientation estimation accuracy with a CNN orientation regressor trained with data augmentation. Codes, dataset are publicly available.

Exploring the interrelationships between microbes and disease can help microbiologists make decisions and plan treatments. Predicting new microbe-disease associations currently relies on biological experiments and domain knowledge, which is time-consuming and inefficient. Automated algorithms are used to uncover the intrinsic link between microbes and disease. However, due to data noise and inadequate understanding of relevant biology, the efficient prediction of microbe-disease associations is still crucial. This study develops a multi-view graph augmentation convolutional network (MVGCNMDA) to predict potential disease-associated microbes. First, we use two data augmentation methods, edge perturbation and node dropping, to remove the data noise in the preprocessing stage. Second, we calculate Gaussian interaction profile kernel similarity and cosine similarity. Therefore, the Graph Convolutional Network(GCN) can fully use multi-view features. Then, the multi-view features are fed into the multi-attention block to learn the weights of different features adaptively. Finally, the embedding results are obtained using a Convolutional Neural Network (CNN) combiner, and the matrix completion is used to predict the relationship between potential microbes and diseases. We test our model on the Human microbe-disease Association Database (HMDAD), Disbiome, and the Combined Dataset (Peryton and MicroPhenoDB). The area under PR curve (AUPR), area under ROC curve (AUC), F1 score, and RECALL value are calculated to evaluate the performance of the developed MVGCNMDA. The AUPR is 0.9440, AUC is 0.9428, F1 score is 0.9383, and RECALL value is 0.8858. The experiments show that our model can accurately predict potential microbe-disease associations compared with the state-of-the-art works on the global Leave-One-Out-Cross-Validation (LOOCV) and the fivefold Cross-Validation (fivefold CV). To further verify the effectiveness of the proposed graph data augmentation, we designed five different settings in the ablation study. Furthermore, we present two case studies that validate the prediction of the potential association between microbes and diseases by MVGCNMDA.

Imbalanced data classification is a challenging task in real applications. In this work. A method is proposed for image classification using imbalanced distribution of classes. The proposed method extracts an appropriate feature vector which can be fed into a simple classifier such as support vector machine (SVM) for classification. At first, the clustering-based feature extraction (CBFE) is applied to initially reduce the feature space. Then, the relationships among each considered sample with the minority classes and majority classes are explored through constructing two individual graphs. The Laplacian matrices of these graphs are combined to provide a dedicated projection matrix for each sample under test while preserve the manifold structure of minority and majority classes. The initial feature reduction just needs the first order data statistics and the second graph-based projection is unsupervised. Moreover, SVM is not sensitive to the training set size. So, the proposed method is not only efficient for imbalanced data classification but also it is efficient for data classification in small sample size situations. Different subsets of SVHN and CIFAR-10 datasets with various imbalance ratios are used for experiments. The results show superior performance of the proposed method compared to SVM and convolutional neural network (CNN). No data augmentation or oversampling is done for balancing the data distribution in the proposed method. But, due to weak performance of CNN with imbalanced datasets, the image augmentation is applied for CNN as the competitor. However, the proposed method without data augmentation provides better results than CNN with data augmentation.

Improved classification of oral cancer through a personalized transfer learning CNN architecture

This presentation shows the advantages of re-melting post-consumer glass, but also the potential risks of using contaminated cullet in the raw material batch of glass furnaces (e.g. container glass furnaces). As an example of potential advantages: increasing the cullet % in the batch of an efficient end-port fired regenerative container glass furnace from 65 up to 75% decreased the specific energy consumption from 3.95 MJ/kg molten glass to 3.8 MJ/kg and reduced the direct CO2 emissions with 31 grams per kg glass. Additional, lower indirect CO2 emissions can be taken into account, since less primary raw materials have to be applied, saving fossil energy in raw material synthesis (e.g. synthetic soda production). Waste glass has to be sorted and prepared in dedicated cullet recycling (treatment) plants (CTP) to meet the strict quality standards often expressed in maximum mass fraction of ceramics, stones, china, metal (ferro & non-ferro) and color mismatches that can still be accepted. But, also the presence of organic components (fats, oils, sugar, food residues,..) has to be controlled to avoid production or glass color problems. Contamination of cullet may lead to: 1. Glass quality problems: inclusions or color changes; 2. Glass melting disturbances by foaming or limited heat transfer into melt; 3. Glass furnace lifetime, by drilling of melts of metals, present in the cullet. The most important problems today are related to the presence of glass ceramics in the post-consumer waste glass and the variable amount of different colors in the glass cullet or fluctuating contamination by organics. Even small pieces of china or glass ceramics in the cullet, with sizes less than 5 mm, may end as glass defects in glass products. Larger pieces of glass ceramics may even lead to severe interruptions in the gob formation process, due to problems with cutting of the gobs with high viscous inclusions. In modern cullet treatment plants most ferro- and non-ferro metals are rather effectively removed. Glass-ceramics are very difficult to distinguish from normal soda-lime-silica glass, because color and transparency can be almost the same. Specific techniques have to be applied to detect glass-ceramic pieces in the cullet based on X-Ray absorption, X-Ray fluorescence, Hyper-spectral Imaging or UV techniques. Such systems are recently applied in modern waste glass treatment plants in Europe, delivering recycling cullet to the glass industry. This paper will show the glass defects related to the presence of glass-ceramics and the typical compositions of these inclusions (often present as cord or big knots). Organic materials can pyrolize within the batch blanket to form carbon rich residues. Carbon or cokes can react with sulfates in the batch and will cause formation of sulfides. A high level of sulfides in the batch (instead of sulfate) will jeopardize the fining process, may cause changes in fining onset temperature, may cause foaming around the batch blanket, or may lead to chemically reduced glass or even amber cords. The process of radiant heat transmission in the melt will change with a variation of the oxidation state of the melt (redox state of batch & cullet). Examples of glass defects caused by metal contamination in the batch will be shown. The composition of the metal inclusions (glass defects) in the glass product may be very different from the composition of the original contamination, causing the defect. An example is metallic aluminum that leads to silicon inclusions in the glass. Nickel sulfide can be formed by pollution of the glass melt by stainless steel flakes. Liquid metals are very aggressive towards the refractory bottom materials of the tank. A droplet of molten lead for instance will drill a hole in the refractory layers: downward drilling. The presentation will show the relation between glass defects and their origin in contaminated cullet and the mechanism of defect formation or defect conversion.

Hand gestures are a form of non-verbal communication. Apart from traditional input devices, hand gestures are used for interaction with computers too. Communication via hand gestures finds many applications in the real world. Different people have hands with different shapes and orientations, which is termed as nonlinearity. The nonlinearity affects the performance of hand gesture models. A convolutional neural network (CNN) is an approach of neural networks, specifically known as deep learning. CNN is used to recognize and classify images. Sometimes, CNN could not correctly understand the hand gesture due to nonlinearity. Data augmentation helps CNN to understand the nonlinearity and complexity of images better. Data augmentation generates enormous data from lesser data, thus increasing the data adversity. Data augmentation uses various operations like zooming, rotating, shifting, shearing, and scaling to generate more data from the existing data. This article executes a CNN model using augmented data for recognition of static hand gestures. The dataset consists of 10 different hand gestures. The experimented CNN model has been trained using 10000 images and tested using 1000 images. The changes in the output of CNN with and without data augmentation have been highlighted. The CNN model employing data augmentation achieved an accuracy of 98.10%, whereas the CNN model excluding the data augmentation process attained an accuracy of 94.90% only.

This chapter examines skin cancer, particularly melanoma, which has a high mortality rate, making early diagnosis essential. It explores how convolutional neural networks (CNNs) can improve melanoma detection, providing a detailed technical analysis of hyperparameters and their impact on model performance. Strategies for tuning hyperparameters, including random search and Bayesian optimization, are demonstrated. Using the HAM10000 dataset, the chapter assesses the impact of different hyperparameter settings on accuracy, sensitivity, and specificity. Issues like class imbalance are addressed with data augmentation and resampling. The optimization methods improve DenseNet121 and MobileNetV2 accuracies to 85.65% and 84.08%, respectively.

Objective. P300 can be analyzed in autism spectrum disorder (ASD) to derive biomarkers and can be decoded in brain–computer interfaces to reinforce ASD impaired skills. Convolutional neural networks (CNNs) have been proposed for P300 decoding, outperforming traditional algorithms but they (a) do not investigate optimal designs in different training conditions; (b) lack in interpretability. To overcome these limitations, an interpretable CNN (ICNN), that we recently proposed for motor decoding, has been modified and adopted here, with its optimal design searched via Bayesian optimization. Approach. The ICNN provides a straightforward interpretation of spectral and spatial features learned to decode P300. The Bayesian-optimized (BO) ICNN design was investigated separately for different training strategies (within-subject, within-session, and cross-subject) and BO models were used for the subsequent analyses. Specifically, transfer learning (TL) potentialities were investigated by assessing how pretrained cross-subject BO models performed on a new subject vs. random-initialized models. Furthermore, within-subject BO-derived models were combined with an explanation technique (ICNN + ET) to analyze P300 spectral and spatial features. Main results. The ICNN resulted comparable or even outperformed existing CNNs, at the same time being lighter. BO ICNN designs differed depending on the training strategy, needing more capacity as the training set variability increased. Furthermore, TL provided higher performance than networks trained from scratch. The ICNN + ET analysis suggested the frequency range [2, 5.8] Hz as the most relevant, and spatial features showed a right-hemispheric parietal asymmetry. The ICNN + ET-derived features, but not ERP-derived features, resulted significantly and highly correlated to autism diagnostic observation schedule clinical scores. Significance. This study substantiates the idea that a CNN can be designed both accurate and interpretable for P300 decoding, with an optimized design depending on the training condition. The novel ICNN-based analysis tool was able to better capture ASD neural signatures than traditional event-related potential analysis, possibly paving the way for identifying novel biomarkers.

Breast cancer remains one of the most common cancers amongst women globally. Early detection is crucial for improving survival rates. While mammography is widely used and an effective imaging technique, it can sometimes yield false positive or false negatives. Mammogram interpretation is highly operator-dependent, introducing variability and the potential for diagnostic errors. Additionally, mammographic images have limitations, such as low contrast in breast tissue and overlapping structures that can obscure lesions or mimic abnormalities. These limitations can lead to unnecessary biopsies or delayed diagnosis. These challenges highlight the needs for advanced and data driven diagnostic tools to support and enhance current screening practices.</p> <p>In recent years, machine learning has emerged as a transformative force in the health care industry offering innovative solution to complex medical challenges. One of the most promising applications of this technology is in the early and accurate detection of breast cancer particularly through the mammography image analysis. Machine learning algorithms, especially deep learning models, have shown significant potential in identifying patterns in mammography that may indicate abnormal cases. This study utilizes the power of deep learning particularly Convolutional Neural Networks (CNNs) to enhance the classification of breast ultrasound images.</p> <p>This research proposes a novel two-level classification approach. At the first level classification, the system distinguishes normal from abnormal cases. Next, the second level classification, it further classifies abnormal cases into benign or malignant tumors. To address challenges such as limited and imbalanced datasets, the study incorporates comprehensive preprocessing techniques including contrast enhancement via Contrast Limited Adaptive Histogram Equalization (CLAHE) and region of interest (ROI). The preprocessing step were evaluated separately and in combination to assess their individual contributions. Additionally, data augmentation is utilized to improve model generalization. To further enhance performance and robustness, multiple CNN architectures are combined using an ensemble mechanism. Soft voting and hard voting leverage the strengths of diverse CNNs reducing individual model biases and improving overall accuracy.</p> <p>The main contributions of this research are: (1) the design of a two-level classification system (2) the utilization of multiple preprocessing techniques to improve image quality, (3) the application of voting ensemble learning to improve predictive stability and accuracy, and (4) an extensive evaluation on the BUSI dataset. This breaks down the complex diagnostic task into smaller, manageable steps to allow for more accurate comparisons, analysis, and improvements at each stage.</p> <p>The first preprocessing step involved the application of CLAHE. In the first level classification (normal vs abnormal), performance of various models has been evaluated. The models achieved an 82.05% accuracy for AlexNet, 84.62% for DenseNet, 85.26% for Inception, 82.69% for MobileNet, 82.05% for ResNet and 82.69% for VGG16. The accuracy of soft voting is 96.20% and accuracy of hard voting 92.09%. In the second level classification (benign vs malignant), the models achieved an 75.00% accuracy for AlexNet, 80.47% for DenseNet, 35.94% for Inception, 41.41% for MobileNet, 35.94% for ResNet and 82.81% for VGG16. The accuracy of soft voting is 90.84% and hard voting is 85.76%.</p> <p>The second preprocessing step involved the ROI was implemented. This step helped isolate the regions containing potential abnormalities, such as tumors, while filtering out irrelevant areas. After the application of ROI (normal vs abnormal), the system evaluated the performance of various models. In the first level classification (normal vs abnormal), the models achieved an 82.69% accuracy for AlexNet, 82.69% for DenseNet, 83.97% for Inception, 87.18% for MobileNet, 82.69% for ResNet and 82.05% for VGG16. The accuracy of soft voting is 100.00% and hard voting is 98.08%. In the second level classification (benign vs malignant), the system evaluated the performance of various models. The models achieved an 75.00% accuracy for AlexNet, 80.47% for DenseNet, 35.94% for Inception, 41.41% for MobileNet, 35.94% for ResNet and 82.81% for VGG16. Soft voting achieved 89.15% and hard voting achieved 80.95%.</p> <p>The final step in the preprocessing stage combines both CLAHE and ROI. The combination led to an improvement in classification performance. In the first level classification (normal vs abnormal), the models achieved an 91.77% accuracy for AlexNet, 92.72% for DenseNet, 95.57% for Inception, 96.52% for MobileNet, 76.90% for ResNet and 91.14% for VGG16. The accuracy of soft voting is 90.88% and hard voting is 93.99%. In the second level classification (benign vs malignant), the system evaluated the performance of various models. The models achieved an 87.69% accuracy for AlexNet, 99.23% for DenseNet, 96.92% for Inception, 93.08% for MobileNet, 94.62% for ResNet and 97.69% for VGG16. Soft voting achieved 96.92% and hard voting achieved 97.69%.</p> <p>These results indicate that every preprocessing step contributed to an improvement in classification accuracy across all models. This study demonstrates that a two-level, ensemble-based deep learning system can significantly improve the diagnostic accuracy of breast ultrasound images, offering a promising tool to assist radiologists and support early detection efforts, particularly in resource-constrained healthcare environments.

Artificial intelligence algorithms like the convolutional neural network (CNN) allow automated medical image analysis. However, accurate classification of tumour tissues in histopathological images including bone tumour poses significant challenges due to the subtle variations of the different classes. In this paper the impact of CNN architecture depth on the classification performance is evaluated. Additionally, the effectiveness of preprocessing techniques such as contrast enhancement and data augmentation is also studied. One of the key challenges faced in this study is the need to balance model complexity with computational efficiency while preventing overfitting on small and imbalanced datasets. The model was trained using a bone or osteosarcoma tumour dataset consists of 536 non-tumour tiles, 263 necrotic tumour tiles, and 345 viable tumour tiles. Our results demonstrate that custom-designed, lightweight CNN models can achieve competitive performance, with the best model achieving a validation accuracy of 87.44% and a validation loss of 0.3594, often outperforming deeper, pretrained architectures. This research contributes valuable insights into effective CNN architecture choices for histopathological analysis, specifically offering a pathway toward more efficient and accurate bone tumour classification systems.

WiFi fingerprinting using Convolutional Neural Networks (CNN) is one of the most promising techniques for indoor localisation due to the extraordinary performance of CNN in image classification. However, the performance of CNN is architecture dependant, and thus an architecture that works well in one case may not work in another, especially for the WiFi-based localisation problems. Most of the solutions use an existing hand-crafted architecture or a semi-automated CNN design for fingerprinting, which requires significant CNN expertise and time. Therefore, a satisfactory solution may not be guaranteed as it is challenging to design numerous possible architectures. In this work, we address this challenge by developing a framework that completely automates the CNN architecture design process. Our automated architectures based on VGG blocks have shown superior performance compared to standard architectures such as VGG-16. We further explore three different heuristics for automation: Bayesian optimisation, Hyperband, and Random Search and demonstrate their importance towards the automated CNN architecture development for WiFi fingerprinting. Experiments are conducted on real-world datasets and, a comparative study between our automated architecture and other models is presented. This work would, therefore, facilitate the CNN design for indoor localisation.

In semiconductor foundries, wafer map defect analysis is crucial to prevent yield excursion. However, traditional manual inspection can hardly meet the high-throughput demand. Deep learning-based automatic defect detection shows promising efforts to achieve high accuracy and efficiency, yet the current approaches’ performance is limited by the imbalanced dataset and lack of interpretability. In this article, we propose a variational autoencoder-enhanced deep learning model (VAEDLM) for wafer defect imbalanced classification. It is light-weighted and effective in wafer defect pattern recognition on imbalanced dataset. It used variational autoencoders (VAEs) and decoders to generate similar wafer defect maps and a refined deep convolutional neural network (CNN) for feature learning. We demonstrate the method using an authentic wafer map dataset, WM-811K. The performance is not only significantly improved after data augmentation, but it also beats the state-of-the art methods, reaching 99.19% accuracy, 99.10% recall, 99.23% precision, 99.96% AUC, and 99.16% for F1-score. It clearly demonstrates the method’s efficacy to deal with the imbalanced defect pattern. Our study using saliency map and t-distributed stochastic neighbor embedding (t-SNE) further leads to enhanced interpretability.