Convolutional Filters Research Articles

Multimodal Biometric Authentication System for Military Weapon Access: Face and ECG Authentication

Unimodal or Single factor biometric systems refer to biometric systems that employ only one form of biometric data to authenticate an individual’s identity. These kinds of biometrics are susceptible to higher error rates and security vulnerabilities because it relays on a single trait for authentication. To overcome this, multimodal biometrics method is proposed. Multi-modal biometric system can authenticate more than once and some advantages include; highaccuracy, low error rate, and large population coverage. These biometrics systems increase integrity and privacy since it will contain several biometric features of every customer. So, here designed a multimodal biometrics project utilizing deep learning to enhance authentication security by combining face and Electrocardiogram(ECG) signals. VGG-16 model, a deep learning architecture used to capture complex patterns in accurate individual identification with both ECG and Facial data. The high-resolution convolutional filters capture the intricate details of the face and ECG waveform, ensuring high accuracy in distinguishing different individuals.

Artificial intelligence assisted common maternal fetal planes prediction from ultrasound images based on information fusion of customized convolutional neural networks.

Ultrasound imaging is frequently employed to aid with fetal development. It benefits from being real-time, inexpensive, non-intrusive, and simple. Artificial intelligence is becoming increasingly significant in medical imaging and can assist in resolving many problems related to the classification of fetal organs. Processing fetal ultrasound (US) images increasingly uses deep learning (DL) techniques. This paper aims to assess the development of existing DL classification systems for use in a real maternal-fetal healthcare setting. This experimental process has employed two publicly available datasets, such as FPSU23 Dataset and Fetal Imaging. Two novel deep learning architectures have been designed in the proposed architecture based on 3-residual and 4-residual blocks with different convolutional filter sizes. The hyperparameters of the proposed architectures were initialized through Bayesian Optimization. Following the training process, deep features were extracted from the average pooling layers of both models. In a subsequent step, the features from both models were optimized using an improved version of the Generalized Normal Distribution Optimizer (GNDO). Finally, neural networks are used to classify the fused optimized features of both models, which were first combined using a new fusion technique. The best classification scores, 98.5 and 88.6% accuracy, were obtained after multiple steps of analysis. Additionally, a comparison with existing state-of-the-art methods revealed a notable improvement in the suggested architecture's accuracy.

Lightweight concrete crack detection based on spiking neural networks

PurposeMost existing methods for concrete crack detection are based on deep learning techniques such as convolutional neural networks. However, these models, due to their large memory footprint, high power consumption and insufficient feature extraction capabilities, face challenges in mobile applications. To address these issues, this paper proposes a lightweight spiking neural network detection model.Design/methodology/approachThis model achieves fast and accurate crack detection. Firstly, the Gabor-Spiking (GS) module preprocesses input images, extracting texture features and edge features of crack images through Gabor filter convolution modules and spiking convolution modules, respectively. Next, the multiscale residual (MR) module is designed, composed of convolutional layers and residual modules of various scales, to process the fused features and perform crack detection.FindingsExperimental results demonstrate that the model’s size can be reduced to 4.6 MB, achieving accuracy improvements to 87.3 and 96.4% on the SDNET and OCD datasets, respectively.Originality/valueThis paper proposes a lightweight spiking neural network detection model based on the GS module for edge texture feature fusion and the MR module for crack detection.

Suppression of interference fringes in wavelength modulation spectroscopy based on convolutional neural networks

To mitigate interference fringes in wavelength modulated spectroscopy (WMS), we introduce a new filter based on convolutional neural networks (New CNNF). This filter effectively suppresses both high and low-frequency variations of interference fringes, including those with envelopes. In striving for comprehensive interference fringe suppression across the entire frequency spectrum, while accommodating diverse cavity length conditions, we confront the challenge of data sparsity. To address this, we introduce a pioneering methodology: finely segmenting the dataset through concentration column density normalization, thereby achieving notable noise reduction. Through the construction and training of the New CNNF model, it was found to exhibit superior performance compared to traditional filtering algorithms, particularly under low signal-to-noise ratio (SNR) conditions.Following processing with the novel convolutional neural network filters (New CNNF), the sample's signal-to-noise ratio (SNR) improved by 26.70 dB, increasing from the original −10.11 dB. When the etalon lengths were 2 cm and 100 cm, the goodness of fit between the predicted second harmonic signal amplitude by New CNNF and the corresponding label concentration reached 0.9995 and 0.9999, respectively. Experimental results demonstrate that New CNNF effectively suppresses high and low-frequency variations and enveloped interference fringes in the TDLAS-WMS system, thereby enhancing the accuracy and stability of methane concentration measurements. CH4 transitions at λ = 1.654 μm were selected to validate this approach. Our method shows promising application prospects and can be extended to the sensing of other gas molecules.

Brief Study on Convolutional Neural Networks

Convolutional Neural Networks (CNNs) have revolutionized the field of computer vision and pattern recognition, offering state-of-the-art performance in tasks such as image classification, object detection, and segmentation. This study explores the architecture, training strategies, and applications of CNNs, focusing on their ability to automatically learn spatial hierarchies of features through layers of convolutional filters. The research evaluates various CNN architectures, including Lent, Alex Net, and ResNet, highlighting their improvements in accuracy and efficiency. Additionally, the study investigates advanced techniques such as data augmentation, transfer learning, and regularization methods (e. g, dropout and batch normalization), which enhance the model's generalization capabilities. Through empirical experiments, we demonstrate the effectiveness of CNNs in real-world scenarios, including medical image analysis, autonomous driving, and facial recognition. The study concludes with a discussion of the future trends of CNNs, particularly in the context of deep learning optimization, hardware acceleration, and integration with emerging technologies like edge computing and quantum machine learning.

Multi-objective topological design considering functionally graded materials and coated fiber reinforcement

This study presents a multi-objective topology optimization method tailored to structures fabricated from functionally graded materials (FGMs), coated FGMs, and coated fiber-reinforced composite materials (FRCMs) with fixed fiber thickness. The design objective is the simultaneous minimization of elastic and thermal compliance. The material properties of these composite materials were derived to generate datasets using the representative volume element method under periodic boundary conditions. Subsequently, machine learning modules were developed based on the datasets to combine with the design process. The multi-objective optimization problem was addressed using the weighted sum method ensuring the generation of the Pareto front. The adaptive weighting strategy is employed to avoid biased results toward a single objective function. To define the coated boundaries within the design domain, image post-processing techniques such as convolution filters, interpolation schemes, and erosion methods were employed on the material layout information of the optimized FGM structures. Through numerical examples, optimized material layouts for coated assemblies incorporating FGMs and FRCMs are presented, with the performance verified through objective function values.

Local and global mixture network for image inpainting

In general, CNN-based inpainting can recover local patterns effectively using convolutional filters, but it may not exploit global correlation fully. On the other hand, transformer-based inpainting can fill in large holes faithfully based on global correlation, rather than local one. In this paper, we propose a novel image inpainting algorithm, called local and global mixture (LGM), to take advantage of the strengths of both approaches and compensate for their weaknesses. The LGM network comprises the local inpainting network (LIN) and the global inpainting network (GIN) in parallel, which are based on convolutional layers and transformer blocks, respectively, and exchange their intermediate results with each other. Furthermore, we develop an error propagation model with a continuous error mask, updated in LIN but used in both LIN and GIN to provide more reliable inpainting results. Extensive experiments demonstrate that the proposed LGM algorithm provides excellent inpainting performance, which indicates the efficacy of the parallel combination of LIN and GIN and the effectiveness of the error propagation model.

Mesh Convolution With Continuous Filters for 3-D Surface Parsing.

Geometric feature learning for 3-D surfaces is critical for many applications in computer graphics and 3-D vision. However, deep learning currently lags in hierarchical modeling of 3-D surfaces due to the lack of required operations and/or their efficient implementations. In this article, we propose a series of modular operations for effective geometric feature learning from 3-D triangle meshes. These operations include novel mesh convolutions, efficient mesh decimation, and associated mesh (un)poolings. Our mesh convolutions exploit spherical harmonics as orthonormal bases to create continuous convolutional filters. The mesh decimation module is graphics processing unit (GPU)-accelerated and able to process batched meshes on-the-fly, while the (un)pooling operations compute features for upsampled/downsampled meshes. We provide an open-source implementation of these operations, collectively termed Picasso. Picasso supports heterogeneous mesh batching and processing. Leveraging its modular operations, we further contribute a novel hierarchical neural network for perceptual parsing of 3-D surfaces, named PicassoNet++. It achieves highly competitive performance for shape analysis and scene segmentation on prominent 3-D benchmarks. The code, data, and trained models are available at https://github.com/EnyaHermite/Picasso.

Enhancing Breast Cancer Detection through Progressive Convolutional Neural Networks with Adaptive Kernel Pooling

Breast cancer is one of the leading causes of mortality among women worldwide, making early detection a crucial component in reducing fatality rates. Traditional diagnostic methods often rely on manual interpretation of mammograms or histopathological images, which can be time-consuming and prone to human error. Recent advances in deep learning, particularly Convolutional Neural Networks (CNNs), have demonstrated significant potential in automating and enhancing breast cancer detection. Breast cancer detection remains a critical challenge in medical diagnostics, where early identification can significantly improve patient outcomes. Leveraging advanced deep learning techniques, this research proposes a novel approach by integrating Progressive Convolutional Neural Networks (PCNNs) with Adaptive Kernel Pooling (AKP) for enhanced image classification. The model architecture dynamically adjusts convolutional filter sizes across layers to capture both coarse and fine-grained features, while AKP adapts pooling operations based on spatial dimensions, optimizing feature retention. Through rigorous experimentation on a large breast cancer histopathological dataset, the proposed method achieved an impressive accuracy of 98%, demonstrating its potential in real-world clinical applications. The high accuracy indicates the model's robustness in differentiating between malignant and benign tissues. This approach not only enhances detection precision but also addresses the challenge of feature loss during pooling, a common issue in traditional Convolutional Neural Networks (CNNs). Additionally, this study contributes to the growing field of medical imaging by offering an adaptable and scalable architecture that could be employed across various medical diagnostics beyond breast cancer detection. The proposed PCNN-AKP framework sets a new benchmark for automated cancer detection models, providing a pathway for future innovations in the use of deep learning for healthcare diagnostics.

Reproducible and Interpretable Machine Learning-Based Radiomic Analysis for Overall Survival Prediction in Glioblastoma Multiforme

Purpose: To develop and validate an MRI-based radiomic model for predicting overall survival (OS) in patients diagnosed with glioblastoma multiforme (GBM), utilizing a retrospective dataset from multiple institutions. Materials and Methods: Pre-treatment MRI images of 289 GBM patients were collected. From each patient’s tumor volume, 660 radiomic features (RFs) were extracted and subjected to robustness analysis. The initial prognostic model with minimum RFs was subsequently enhanced by including clinical variables. The final clinical–radiomic model was derived through repeated three-fold cross-validation on the training dataset. Performance evaluation included assessment of concordance index (C-Index), integrated area under curve (iAUC) alongside patient stratification into low and high-risk groups for overall survival (OS). Results: The final prognostic model, which has the highest level of interpretability, utilized primary gross tumor volume (GTV) and one MRI modality (T2-FLAIR) as a predictor and integrated the age variable with two independent, robust RFs, achieving moderately good discriminatory performance (C-Index [95% confidence interval]: 0.69 [0.62–0.75]) with significant patient stratification (p = 7 × 10−5) on the validation cohort. Furthermore, the trained model exhibited the highest iAUC at 11 months (0.81) in the literature. Conclusion: We identified and validated a clinical–radiomic model for stratification of patients into low and high-risk groups based on OS in patients with GBM using a multicenter retrospective dataset. Future work will focus on the use of deep learning-based features, with recently standardized convolutional filters on OS tasks.

Inferring ECG Waveforms from PPG Signals with a Modified U-Net Neural Network.

There are two widely used methods to measure the cardiac cycle and obtain heart rate measurements: the electrocardiogram (ECG) and the photoplethysmogram (PPG). The sensors used in these methods have gained great popularity in wearable devices, which have extended cardiac monitoring beyond the hospital environment. However, the continuous monitoring of ECG signals via mobile devices is challenging, as it requires users to keep their fingers pressed on the device during data collection, making it unfeasible in the long term. On the other hand, the PPG does not contain this limitation. However, the medical knowledge to diagnose these anomalies from this sign is limited by the need for familiarity, since the ECG is studied and used in the literature as the gold standard. To minimize this problem, this work proposes a method, PPG2ECG, that uses the correlation between the domains of PPG and ECG signals to infer from the PPG signal the waveform of the ECG signal. PPG2ECG consists of mapping between domains by applying a set of convolution filters, learning to transform a PPG input signal into an ECG output signal using a U-net inception neural network architecture. We assessed our proposed method using two evaluation strategies based on personalized and generalized models and achieved mean error values of 0.015 and 0.026, respectively. Our method overcomes the limitations of previous approaches by providing an accurate and feasible method for continuous monitoring of ECG signals through PPG signals. The short distances between the infer-red ECG and the original ECG demonstrate the feasibility and potential of our method to assist in the early identification of heart diseases.

A visual measurement method for grinding surface roughness combining filter and branch convolution network

ABSTRACT The visual measurement method of grinding surface roughness is mainly predicted by designing image feature indicators. In contrast, the self-extraction method of grinding surface features based on deep learning has problems such as noise, low resolution and poor perception of details. This paper proposes a visual measurement method for grinding surface roughness to address these problems by combining filters and branching convolutional networks. The method adopts a two-branch convolutional neural network, inputs the images using filter denoising and the original grinding workpiece surface images simultaneously, self-extracts roughness correlation features and fuses them to achieve feature enhancement, effectively improves the recognition accuracy and generalisation ability of the model, and verifies the robustness of the model in the case of rusting on the workpiece surface. The experimental results show that the method can effectively solve the problem of the grinding process’s weak and difficult recognition of surface roughness feature information and provide a basis for automated visual online measurement of grinding surface roughness.

Intrusion detection system for cloud environment based on convolutional neural networks and PSO algorithm

Authentication of clients and their applications to cloud services is a major concern. Network security and the identification of hostile activities are greatly aided by intrusion detection systems (IDS). In general, optimisation strategies can be applied to improve IDS model performance. Convolutional neural networks (CNN) and other deep learning (DL) algorithms is utilised to enhance IDS’s capability to identify and categories intrusions. IDSs can identify prior attacks, adapt to changing threats, and minimise false positives by utilising these strategies. In this work, a lightweight CNN is proposed for intrusion detection in cloud environment. The main contribution of this research is to use particle swarm optimization (PSO), ametaheuristic algorithm to find the CNNs optimal parameters that comprise the number of convolutional layers, the size of the filter utilized in the convolutional procedure, the number of convolutional filters, and the batch size. Heuristic based searches are useful for solving these kinds of problems. The experimental outcomes demonstrate that the proposed method reaches 91.70% of accuracy, 91.82% of precision, 91.99% of recall and 91.90% of F1-score. Cloud providers can gain from improved security measures by incorporating the proposed IDS paradigm into cloud settings, thereby minimizing unauthorized access and any data breaches.

HardVD: High-capacity cross-modal adversarial reprogramming for data-efficient vulnerability detection

The substantial proliferation of software vulnerabilities poses a persistent threat to system security, driving increased interest in applying deep learning (DL) for vulnerability detection. However, DL-based detectors often operate with a fixed number of input tokens, leading to semantic loss over large code snippets. Additionally, developing these detectors demands substantial labeled data and training time. To address these limitations, this paper proposes HardVD, which explores High-capacity cross-modal adversarial reprogramming for data-efficient Vulnerability Detection. HardVD devises a high-capacity semantic extractor to capture salient features per line of code, which are then arranged as patches to form an image representing the target function. These images are processed using convolutional filters as universal perturbations and non-parametric label remapping to adapt a pretrained Vision Transformer (ViT) for vulnerability detection, updating only the limited parameters of the perturbation filters during training. Extensive experiments demonstrate that HardVD outperforms DL-based baselines in terms of detection effectiveness, data-limited performance, and computational overhead. The ablation study also confirms the essential role of our high-capacity semantic extractor, without which an averaged relative decrease of 5.87% and 7.98% in accuracy and F1 score is observed, respectively.

Singular-Value-Decomposition-Based Matrix Surgery.

This paper is motivated by the need to stabilise the impact of deep learning (DL) training for medical image analysis on the conditioning of convolution filters in relation to model overfitting and robustness. We present a simple strategy to reduce square matrix condition numbers and investigate its effect on the spatial distributions of point clouds of well- and ill-conditioned matrices. For a square matrix, the SVD surgery strategy works by: (1) computing its singular value decomposition (SVD), (2) changing a few of the smaller singular values relative to the largest one, and (3) reconstructing the matrix by reverse SVD. Applying SVD surgery on CNN convolution filters during training acts as spectral regularisation of the DL model without requiring the learning of extra parameters. The fact that the further away a matrix is from the non-invertible matrices, the higher its condition number is suggests that the spatial distributions of square matrices and those of their inverses are correlated to their condition number distributions. We shall examine this assertion empirically by showing that applying various versions of SVD surgery on point clouds of matrices leads to bringing their persistent diagrams (PDs) closer to the matrices of the point clouds of their inverses.

Temperature field in the crack-free ductile dry grinding of fused silica based on wheel wear topographies

Fused silica is an excellent window material widely used in ultraviolet transmission optical system. Crack-free ductile dry grinding is a novel method for the efficient fabrication of fused silica. The grinding temperature field has an important influence on the grinding process. However, most previous studies assumed that the grinding temperature was independent of the wheel’s wear. In this paper, a temperature field model of the ductile dry grinding of fused silica is developed based on wheel wear topographies. Simulated wheel topographies with the same statistical parameters as the realistic wheel wear topographies are reconstructed based on the convolution filtering and Johnson transformation algorithm. The theoretical temperature field is the superposition of the thermal effects induced by effective cutting grain point heat sources extracted from the simulated wheel topographies. The theoretical prediction accuracy of the wheel-workpiece contact zone is validated by an infrared radiation transmission method. This model not only provides opportunity to explore the material removal mechanisms and improve the surface generation quality of fused silica during the wear process of the wheel, but also could be extended to provide the basis for the utilization of grinding heat or prevention of grinding thermal damage for other isotropic materials.

Unsupervised physics-informed deep learning for assessing pulmonary artery hemodynamics

Deep learning advancements have significantly benefited medical applications. One such helpful application is noninvasive fractional flow reserve (FFR) evaluation along the pulmonary artery tree, which aids in planning optimal balloon pulmonary angioplasty. This study proposes a surrogate model that employs an unsupervised physics-informed neural network (UPNN) to predict FFR based on pulmonary CT angiography. To ensure the UPNN strictly follows the unique solution of the governing equations, we implemented a hard boundary conditions enforcement approach. Subsequently, a finite difference convolutional filter was developed to enhance connectivity among neighboring points. This allows the neural networks to propagate boundary conditions into the unseen vessel interior and perceive the geometric structure of the computational domain. We also introduced regularization constraints on the three components contributing to pressure drop within the artery tree. A total of 4500 synthetic pulmonary artery trees were used to train the UPNN with impedance outlet boundary conditions. We found that the limits of agreement of FFR trained by UPNN ranged from −0.05 to 0.04 with computational fluid dynamics (CFD) simulation results. The testing results showed that the correlation coefficient between FFR predicted by UPNN and CFD were 96.1% and 94.7% for synthetic and patient-specific data, respectively. Using invasive FFR as reference, the accuracy of FFR predicted by UPNN and CFD was 81.4% and 84.8%, respectively. These results demonstrate that our UPNN-based surrogate model’s performance in evaluating FFR aligns closely with CFD simulations.

An enhanced AlexNet-Based model for femoral bone tumor classification and diagnosis using magnetic resonance imaging

ObjectiveBone tumors, known for their infrequent occurrence and diverse imaging characteristics, require precise differentiation into benign and malignant categories. Existing diagnostic approaches heavily depend on the laborious and variable manual delineation of tumor regions. Deep learning methods, particularly convolutional neural networks (CNNs), have emerged as a promising solution to tackle these issues. This paper introduces an enhanced deep-learning model based on AlexNet to classify femoral bone tumors accurately. MethodsThis study involved 500 femoral tumor patients from July 2020 to January 2023, with 500 imaging cases (335 benign and 165 malignant). A CNN was employed for automated classification. The model framework encompassed training and testing stages, with 8 layers (5 Conv and 3 FC) and ReLU activation. Essential architectural modifications included Batch Normalization (BN) after the first and second convolutional filters. Comparative experiments with various existing methods were conducted to assess algorithm performance in tumor staging. Evaluation metrics encompassed accuracy, precision, sensitivity, specificity, F-measure, ROC curves, and AUC values. ResultsThe analysis of precision, sensitivity, specificity, and F1 score from the results demonstrates that the method introduced in this paper offers several advantages, including a low feature dimension and robust generalization (with an accuracy of 98.34 %, sensitivity of 97.26 %, specificity of 95.74 %, and an F1 score of 96.37). These findings underscore its exceptional overall detection capabilities. Notably, when comparing various algorithms, they generally exhibit similar classification performance. However, the algorithm presented in this paper stands out with a higher AUC value (AUC=0.848), signifying enhanced sensitivity and more robust specificity. ConclusionThis study presents an optimized AlexNet model for classifying femoral bone tumor images based on convolutional neural networks. This algorithm demonstrates higher accuracy, precision, sensitivity, specificity, and F1-score than other methods. Furthermore, the AUC value further confirms the outstanding performance of this algorithm in terms of sensitivity and specificity. This research makes a significant contribution to the field of medical image classification, offering an efficient automated classification solution, and holds the potential to advance the application of artificial intelligence in bone tumor classification.

DL-based multi-artifact EEG denoising exploiting spectral information

The artifacts affecting electroencephalographic (EEG) signals may undermine the correct interpretation of neural data that are used in a variety of applications spanning from diagnosis support systems to recreational brain-computer interfaces. Therefore, removing or - at least - reducing the noise content in respect to the actual brain activity data becomes of fundamental importance. However, manual removal of artifacts is not always applicable and appropriate, and sometimes the standard denoising techniques may encounter problems when dealing with noise frequency components overlapping with neural responses. In recent years, deep learning (DL) based denoising strategies have been developed to overcome these challenges and learn noise-related patterns to better discriminate actual EEG signals from artifact-related data. This study presents a novel DL-based EEG denoising model that leverages the prior knowledge on noise spectral features to adaptively compute optimal convolutional filters for multi-artifact noise removal. The proposed strategy is evaluated on a state-of-the-art benchmark dataset, namely EEGdenoiseNet, and achieves comparable to better performances in respect to other literature works considering both temporal and spectral metrics, providing a unique solution to remove muscle or ocular artifacts without needing a specific training on a particular artifact type.

Mid-infrared methane standoff sensor using a frequency channel attention based convolutional neural network filter

Highly sensitive standoff methane detection is vital for atmospheric science, environmental protection, and production safety. We develop a mid-infrared methane standoff sensor using a cooperative target based on tunable diode laser absorption spectroscopy (TDLAS). To enhance sensor sensitivity, we propose a one-dimensional frequency channel attention-based convolutional neural network (1D-FCACNN) filter, which can effectively denoise the methane absorbance signals and its second harmonic signals. The filter model is adequately trained on a simulated spectrum dataset, which we constructed based on data augmentation. In comparison with reported filtering algorithms, the proposed filter shows the best performance in both measuring modes and evaluation metrics. Real-time measurements show that the measuring accuracy and limit of detection (LOD) of the proposed sensor reach a minimum of 30.40 ppb and 5.04 ppb over a 10-meter optical range, a significant improvement compared to previous reports of methane standoff sensors. The proposed methane standoff sensor proves the feasibility of enhancing the performance of TDLAS gas sensors with the attention mechanism, bringing a new option for high-sensitivity measurements of methane and other atmospheric trace gases.