Science topic

Neural Networks - Science topic

Everything about neural networks
Filters
All publications are displayed by default. Use this filter to view only publications with full-texts.
Article
Full-text available
The palm oil industry is a significant component of Indonesia's economy, driven by increasing global demand across various industries. Manual identification of palm oil fruit ripeness is often subjective and labor-intensive, creating a need for a faster and more accurate solution. This study proposes the use of deep learning models based on transfe...
Article
Full-text available
A novel strategy for generating datasets has been developed within the context of drag prediction for automotive geometries using neural networks. A primary challenge in this field is constructing a training database of sufficient size and diversity. Our method relies on a small number of initial data points and provides a recipe to systematically...
Article
Full-text available
Despite being effective in many application areas, Deep Neural Networks (DNNs) are vulnerable to being attacked. In object recognition, the attack takes the form of a small perturbation added to an image, that causes the DNN to misclassify, but to a human appears no different. Adversarial attacks lead to defences that are themselves subject to atta...
Article
Full-text available
Recent advancements in neural network design have given rise to the development of Kolmogorov–Arnold Networks (KANs), which enhance interpretability and precision of these systems. This paper presents the Fractional Kolmogorov–Arnold Network (fKAN), a novel neural network architecture that incorporates the distinctive attributes of KANs with a trai...
Article
Full-text available
Acoustic measurements of batteries are known to be correlated to their state-of-charge and can be practicably leveraged for state estimation using parametric machine learning models. Such models can be easily designed to have millions of tuneable parameters, which endows them with tremendous but often misinterpreted fitting ability. The real perfor...
Preprint
Full-text available
Graph Neural Networks (GNNs) have become an influential framework for learning and inference on graph-structured data [97, 185, 192, 215]. In the realm of uncertainty modeling, Fuzzy Graphs and Neutrosophic Graphs are increasingly recognized as essential tools [20]. Plithogenic Graphs, in particular, offer a broader generalization, encompassing bot...
Conference Paper
Full-text available
This paper introduces a novel deep learning framework for 2D shape classification that emphasizes equivariance and invariance through Generalized Finite Fourier-based Descriptors (GFID). Instead of relying on raw images, we extract contours from 2D shapes and compute equivariant, invariant, and stable descriptors, which represent shapes as column v...
Conference Paper
Full-text available
Image fusion is a process that combines data from multiple images of a scene to create a more comprehensive and higher-quality image. This technique has numerous applications in fields such as remote sensing, medical imaging, and computer vision. Initially, traditional methods like weighted averaging and Laplace filters were used for image fusion,...
Article
Full-text available
Skin cancer is a prevalent health concern, and accurate segmentation of skin lesions is crucial for early diagnosis. Existing methods for skin lesion segmentation often face trade-offs between efficiency and feature extraction capabilities. This paper proposes Dual Skin Segmentation (DuaSkinSeg), a deep-learning model, to address this gap by utiliz...
Article
Full-text available
Mammography is the recommended imaging modality for breast cancer screening. Expressions of human epidermal growth factor receptor 2 (HER2), estrogen receptor (ER), and progesterone receptor (PR) are critical to the development of therapeutic strategies for breast cancer. In this study, a deep learning model (CBAM ResNet-18) was developed to predic...
Article
Full-text available
In the research of non-intrusive load monitoring (NILM), the temporal characteristics of V–I trajectories are often overlooked, and using a single feature for identification may lead to insignificant differences between similar loads. Based on this, this paper proposes a non-intrusive load monitoring method based on time-enhanced multidimensional f...
Article
Full-text available
In this manuscript, a PID controller based on adaptive neural networks for manipulating robots with n degrees of freedom is presented. The neural network is given by a two-layer perceptron that compensates for the unknown dynamics of the system. The weights of the output layer are estimated online by the proposed adaptation law, while the weights a...
Article
Full-text available
Gasket inspection is a critical step in the quality control of a product. The proposed method automates the detection of misaligned or incorrectly fitting gaskets, ensuring timely repair action. The suggested method uses deep learning approaches to recognize and evaluate radiator images, with a focus on identifying misaligned or incorrectly install...
Article
Full-text available
Graph Neural Networks (GNNs) serve as a powerful framework for representation learning on graph-structured data, capturing the information of nodes by recursively aggregating and transforming the neighboring nodes’ representations. Topology in graph plays an important role in learning graph representations and impacts the performance of GNNs. Howev...
Article
Full-text available
Deep neural networks (DNNs) are widely studied in various applications. A DNN consists of layers of neurons that compute affine combinations, apply nonlinear operations, and produce corresponding activations. The rectified linear unit (ReLU) is a typical nonlinear operator, outputting the max of its input and zero. In scenarios like max pooling, wh...
Article
Full-text available
Purpose Commonly used MR image quality (IQ) metrics have poor concordance with radiologist‐perceived diagnostic IQ. Here, we develop and explore deep feature distances (DFDs)—distances computed in a lower‐dimensional feature space encoded by a convolutional neural network (CNN)—as improved perceptual IQ metrics for MR image reconstruction. We furth...
Preprint
Full-text available
We propose compleX-PINN, a novel physics-informed neural network (PINN) architecture that incorporates a learnable activation function inspired by Cauchy integral theorem. By learning the parameters of the activation function, compleX-PINN achieves high accuracy with just a single hidden layer. Empirical results show that compleX-PINN effectively s...
Article
Full-text available
Light Field Angular Super-Resolution (LFASR) addresses the issue where Light Field (LF) images can not simultaneously achieve both high spatial and angular resolution due to the limited resolution of optical sensors. Since Spatial-Angular Correlation (SAC) features are closely related to the structure of LF images, its accurate and complete extract...
Article
Full-text available
Background In the power system, the identification of the health status of the transmission tower is a daily task that must be performed. In addition, bolt loosening is a common damage mode affecting the main materials of transmission towers. When bolt loosening occurs, it weakens the bearing capacity of the transmission tower. If not detected and...
Article
Full-text available
Improved turbulence models are necessary for achieving more accurate solutions in Reynolds-averaged Navier–Stokes (RANS) simulations. RANS is widely used in various engineering applications, and enhancing its accuracy is crucial for geometry design and control applications. With the increasing availability of high-fidelity datasets, machine learnin...
Preprint
Full-text available
Graph Neural Networks (GNNs) have demonstrated remarkable success across various graph-related tasks; however, their performance often suffers when dealing with heterophilic graphs, where connected nodes tend to have dissimilar characteristics. This paper introduces a novel approach, Adaptive Neighborhood Feature Mixing (ANFM), that addresses the l...
Article
Full-text available
A persistent challenge in machine learning is the computational inefficiency of neural architecture search (NAS), particularly in resource-constrained domains like pre-dictive maintenance. This work introduces a novel learning-curve estimation framework that reduces NAS computational costs by over 50% while maintaining model performance, addressing...
Article
Full-text available
Sequential press forming is a method used to form a workpiece while changing its location and to manufacture large structures, such as storage tanks. To optimize the sequential press forming conditions, accurate and efficient predictions of product shapes using numerical simulations are required. However, such predictions are currently difficult be...
Article
Full-text available
In this paper, a hybrid controller with a sampled data control is investigated to achieve finite-time master–slave synchronization of delayed fractional-order neural networks (DFONNs). A Lyapunov-Krasovskii functional is constructed to obtain the sufficient conditions that incorporate delay information. For the first time, the asymptotic stability...
Article
Full-text available
The use of neural networks in the architectural design pre-phase is becoming increasingly prevalent among designers. This article presents a method of textual and graphical analysis of construction sites using neural networks. On the example of two projects, which won the architectural competition, the following are considered: qualitative characte...
Article
Full-text available
The nonstationary fuzzy neural network (NFNN) has proven to be an effective and interpretable tool in machine learning, capable of addressing uncertainty problems similarly to type-2 fuzzy neural networks, while offering reduced computational complexity. However, the update of disturbance parameters in an NFNN is restricted due to the necessity of...
Article
Full-text available
The topic under consideration, text classification using neural network technologies has significant potential in various industries, including universities of agriculture. Monitoring of key indicators is extremely important, but performing it manually with the involvement of experts can be costly and time-consuming compared to using neural network...
Article
Full-text available
This paper explores an approach to analyzing the quality of meat raw materials using convolutional neural networks. The study focuses on the development and application of a comprehensive system that integrates deep learning capabilities with evolutionary algorithms to enhance the accuracy and efficiency of estimating parameters such as the hydroge...
Article
Full-text available
Accurately estimating house values is a critical challenge for real-estate stakeholders, including homeowners, buyers, sellers, agents, and policymakers. This study introduces a novel approach to this problem using Kolmogorov–Arnold networks (KANs), a type of neural network based on the Kolmogorov–Arnold theorem. The proposed KAN model was tested o...
Preprint
Full-text available
Graph Neural Networks (GNNs) have demonstrated significant potential for learning from graph-structured data. However, their performance often degrades on heterophilous graphs, where linked nodes exhibit dissimilar features. In this paper, we introduce the Adaptive Relational Graph Network (ARGN), a novel GNN architecture designed specifically to h...
Article
Full-text available
This paper addresses the challenge of establishing rigorous error bounds for zero-trace Rectified Linear Unit (ReLU) Neural Networks (NNs). We derive theoretical results to provide insights into the accuracy of these networks in approximating continuous functions, focusing on the influence of network architecture, such as the number of layers and n...
Article
Full-text available
Recognizing violent behaviors in videos is a challenging task due to the complexity of human actions and background information. In this paper, we propose STIG-Net, a spatial–temporal interactive graph framework, to address this issue. By extracting keypoints from video frames and constructing special edges based on their relationships, STIG-Net le...
Article
Full-text available
This paper addresses the three-dimensional trajectory tracking problem of underactuated autonomous underwater vehicles (AUVs) operating in the presence of external disturbances and unmodeled dynamics by proposing a predefined-time adaptive control scheme. Firstly, the underactuated AUV system was decoupled into drive and non-drive subsystems to fac...
Article
Full-text available
The article presents and discusses the results of the research of forecasting power demands in Polish Power System with time horizon of one hour ahead in conditions of limited availability of forecasting model input data, covering only three months. The prediction was carried out using deep neural networks - LSTM (Long Short-Term Memory) connected...
Article
Full-text available
Writer identification based on deep learning has shown great potential in fields such as forensic analysis and financial security due to its high efficiency and accuracy. However, the specificity of deep neural networks limits the acceptance and adoption of their identification results in these fields.This is due to the ‘opacity’ of deep neural net...
Article
Full-text available
Nowadays, neural networks are increasingly being used to generate sketchy architectural solutions. In this article, two projects that won the architectural competition are used as examples to demonstrate how to use the MidJorney neural network. Examples are given for generating text from the source image, generating images based on text queries and...
Preprint
Full-text available
Early Exit Neural Networks (EENNs) endow a standard Deep Neural Network (DNN) with Early Exit Clas-sifiers (EECs), to provide predictions at intermediate points of the processing when enough confidence in classification is achieved. This leads to many benefits in terms of effectiveness and efficiency. Currently, the design of EENNs is carried out m...
Article
Full-text available
In ultra-precision diamond turning (UPDT), the effective monitoring of cutting force is promising to guarantee the desired surface quality. However, most of the monitoring methods require pre-processing of the original signal and it would induce data loss. This issue is especially serious during in-situ monitoring of cutting force signals with dyna...
Article
Full-text available
A generalized neural network with four arbitrary delays is considered in this paper. A criterion of both Devaney chaos and Li-Yorke chaos is given under some weak conditions. Furthermore, lower bounds of the parameters of making the network chaotic are effectively determined. For different values of the delays, the proving processes of chaos are di...
Article
Full-text available
Artificial intelligence based on neural network technology has provided innovative methods for predicting unsteady flow fields. However, both purely data-driven and single physics-driven methods can only perform short-term predictions for unsteady flow fields and are unable to achieve medium- to long-term predictions. A composite neural network CNN...
Article
Full-text available
Yogasana, often simply referred to as “asana”, is a Sanskrit term that encompasses the physical postures or poses practiced in yoga. Traditionally, there are almost 90 classical Yogasanas, each with its unique benefits and variations. It's very important to identify the postures and benefits of each asana. Classifications of Yogasana can have a sig...
Article
Full-text available
As the observation frequency of large-aperture antennas increases, the requirements for measuring main reflector deformation have become more stringent. Recently, the rapid development of deep learning has led to its application in antenna deformation prediction. However, achieving high accuracy requires a large number of high-fidelity deformation...
Preprint
Full-text available
Anomaly detection based on 3D point cloud data is an important research problem and receives more and more attention recently. Untrained anomaly detection based on only one sample is an emerging research problem motivated by real manufacturing industries such as personalized manufacturing that only one sample can be collected without any additional...
Article
Full-text available
Deep reinforcement learning has achieved significant success in complex decision-making tasks. However, the high computational cost of policies based on deep neural networks restricts their practical application. Specifically, each decision made by an agent requires a complete neural network computation, leading to a linear increase in computationa...
Preprint
Full-text available
The sign problem that arises in Hybrid Monte Carlo calculations can be mitigated by deforming the integration manifold. While simple transformations are highly efficient for simulation, their efficacy systematically decreases with decreasing temperature and increasing interaction. Machine learning models have demonstrated the ability to push furthe...
Article
Full-text available
In this paper, we considered the problem of the simultaneous approximation of a function and its derivatives by means of the well‐known neural network (NN) operators activated by the sigmoidal function. Other than a uniform convergence theorem for the derivatives of NN operators, we also provide a quantitative estimate for the order of approximatio...
Article
Full-text available
The demand for the customization of tactile sensations in rotary encoders is increasing to align with individual user preferences. In light of this objective, this study aimed to construct a tactile inference model that considers individuality with the goal of regulating targeted factor perceptions. An experimental evaluation was performed involvin...
Preprint
Full-text available
Neural architectures such as Recurrent Neural Networks (RNNs), Transformers, and State-Space Models have shown great success in handling sequential data by learning temporal dependencies. Decision Trees (DTs), on the other hand, remain a widely used class of models for structured tabular data but are typically not designed to capture sequential pat...
Preprint
Full-text available
This work presents a data-driven approach to estimating the sound absorption coefficient of an infinite porous slab using a neural network and a two-microphone measurement on a finite porous sample. A 1D-convolutional network predicts the sound absorption coefficient from the complex-valued transfer function between the sound pressure measured at t...
Article
Full-text available
Graph contrastive learning-based recommendations have attracted a lot of research attention due to their exceptional performance. However, these approaches, which hinge on the optimization of downstream recommendations, often deviate from the original purpose of graph contrastive learning (i.e., learning embeddings independently of downstream tasks...
Article
Full-text available
In recent years, deep learning-based watermarking methods have been developed to address the shortcomings of traditional watermarking algorithms. Some methods adopt an end-to-end framework to train the watermarking model, enabling excellent watermark embedding and extraction. However, the visual quality and robustness of these approaches remain ins...
Preprint
Full-text available
Explaining Graph Neural Network (XGNN) has gained growing attention to facilitate the trust of using GNNs, which is the mainstream method to learn graph data. Despite their growing attention, Existing XGNNs focus on improving the explanation performance, and its robustness under attacks is largely unexplored. We noticed that an adversary can slight...
Article
Full-text available
Abstract: In this work, one new approach for RSTB-invariant object representation is presented based on the modified Mellin–Fourier Transform (MFT). For this, in the well-known steps of MFT, the logarithm operation in the log-polar transform is re-placed by the operation “rising on a power”. As a result, the central part of the pro-cessed area is r...
Preprint
Full-text available
We present the first extraction of transverse-momentum-dependent distributions of unpolarised quarks from experimental Drell-Yan data using neural networks to parametrise their nonperturbative part. We show that neural networks outperform traditional parametrisations providing a more accurate description of data. This work establishes the feasibili...
Preprint
Full-text available
The multi-timestep command governor (MCG) is an add-on algorithm that enforces constraints by modifying, at each timestep, the reference command to a pre-stabilized control system. The MCG can be interpreted as a Model-Predictive Control scheme operating on the reference command. The implementation of MCG on nonlinear systems carries a heavy comput...
Preprint
Full-text available
The asymptotic properties of Bayesian Neural Networks (BNNs) have been extensively studied, particularly regarding their approximations by Gaussian processes in the infinite-width limit. We extend these results by showing that posterior BNNs can be approximated by Student-t processes, which offer greater flexibility in modeling uncertainty. Specifi...
Article
Full-text available
The existing sequential recommendation algorithms generally face the problems of not fully utilizing item relationships across sequences, being incapable of effectively capturing users’ global preferences, and being susceptible to data sparsity. To address above problems, we propose a dual-view data augmentation at subgraph level and graph contrast...
Article
Full-text available
No‐reference image quality assessment (NR‐IQA) has garnered significant attention due to its critical role in various image processing applications. This survey provides a comprehensive and systematic review of NR‐IQA methods, datasets, and challenges, offering new perspectives and insights for the field. Specifically, we propose a novel taxonomy f...