Neural Approach Research Articles

A neural modeling approach to study mechanisms underlying the heterogeneity of visual spatial frequency sensitivity in schizophrenia

Patients with schizophrenia exhibit abnormalities in spatial frequency sensitivity, and it is believed that these abnormalities indicate more widespread dysfunction and dysregulation of bottom-up processing. The early visual system, including the first-order Lateral Geniculate Nucleus of the thalamus (LGN) and the primary visual cortex (V1), are key contributors to spatial frequency sensitivity. Medicated and unmedicated patients with schizophrenia exhibit contrasting changes in spatial frequency sensitivity, thus making it a useful probe for examining potential effects of the disorder and antipsychotic medications in neural processing. We constructed a parameterized, rate-based neural model of on-center/off-surround neurons in the early visual system to investigate the impacts of changes to the excitatory and inhibitory receptive field subfields. By incorporating changes in both the excitatory and inhibitory subfields that are associated with pathophysiological findings in schizophrenia, the model successfully replicated perceptual data from behavioral/functional studies involving medicated and unmedicated patients. Among several plausible mechanisms, our results highlight the dampening of excitation and/or increase in the spread and strength of the inhibitory subfield in medicated patients and the contrasting decreased spread and strength of inhibition in unmedicated patients. Given that the model was successful at replicating results from perceptual data under a variety of conditions, these elements of the receptive field may be useful markers for the imbalances seen in patients with schizophrenia.

Estimating the soil water retention curve by the HYPROP-WP4C system, HYPROP-based PCNN-PTF and inverse modeling using HYDRUS-1D

Due to the difficulty of direct field measurement, soil hydraulic properties are often obtained in laboratory settings using small undisturbed soil samples or estimated indirectly through pedotransfer functions (PTFs). The pseudo-continuous pedotransfer function (PCNN-PTF) is a neural network-based approach for estimating soil hydraulic properties. The main objective of this study was to use field soil moisture and tension data to assess soil water retention curves obtained from the HYPROP-WP4C system and the HYPROP-based PCNN-PTF. The in-situ soil water retention data were simultaneously acquired using Acclima TDT and MeterGroup MPS-6 sensors (every 30 min, May to September 2020) from 11 hybrid bermudagrass plots under different irrigation treatments in Riverside, California. We utilized extended evaporation and dewpoint methods using HYPROP-WP4C (Meter Group Inc., USA) devices to obtain lab-measured SWRCs. In addition, SWRCs were estimated from Rosetta (Schaap et al., 2001) and by inverse modeling in HYDRUS-1D utilizing in-situ moisture retention data. Although the hysteresis impacted field data, overall, there was a good agreement between the in-situ and lab water retention data for most samples, especially within the pF range of 2–3.5. The PCNN-PTF outperformed Rosetta in estimating laboratory (RMSE=0.034 cm3 cm−3 vs 0.063 cm3 cm−3) and in-situ soil moisture data (RMSE=0.048 cm3 cm−3 vs 0.082 cm3 cm−3). Inverse modeling of in-situ data also performed well in estimating the SWRC (RMSE=0.043 cm3 cm−3); however, further attention is required in dry and saturated soil conditions. We developed a simple, free, and easy-to-access tool called PC-PTF for estimating the SWRC using the PCNN-PTF model evaluated in this study. The PC-PTF can be accessed from the Verdi Water Management Group website: http://www.ucrwater.com/software-and-tools.html.

Unlocking Security: The Significant Role of Machine Learning and Deep Learning Techniques in Fingerprint Recognition Systems

One of the more widely utilized biometrics for in-person identification is a fingerprint. It has been discovered that no two people have the same fingerprints and that each one is distinct. A person's fingerprint traits remain constant as they age. Compared to DNA, fingerprints are more distinctive. Identical twins cannot have the same fingerprints even though they have the same DNA. Artificial Intelligence encompasses the broader goal of creating intelligent systems, while Machine Learning focuses on developing algorithms that enable machines to learn from data. Deep Learning, as a subset of Machine Learning, leverages neural networks with multiple layers to learn complex representations of data. While Deep Learning has become increasingly prominent and successful in recent years, it is just one of the many approaches within the broader field of AI and Machine Learning. Fingerprint recognition stands as one of the oldest and most widely used biometric authentication methods, finding applications across diverse domains such as law enforcement, access control, and mobile device security. With the advent of machine learning and deep learning techniques, significant strides have been made in enhancing the accuracy, efficiency, and scalability of fingerprint recognition systems. This comprehensive review aims to provide a thorough examination of the pivotal role played by machine learning and deep learning methodologies in advancing fingerprint recognition technology. The paper commences with an introduction to the importance and ubiquity of fingerprint recognition, elucidating its significance in various real-world applications. Subsequently, it delves into the foundational concepts of machine learning and deep learning, elucidating their relevance to fingerprint recognition tasks. Key machine learning algorithms such as Support Vector Machines (SVM), k-Nearest Neighbors (k-NN), Random Forests, and neural network-based approaches are discussed, highlighting their strengths and limitations in the context of fingerprint recognition.

Neural discovery of balance-aware polarized communities

Signed graphs are a model to depict friendly (positive) or antagonistic (negative) interactions (edges) among users (nodes). 2-Polarized-Communities (2pc) is a well-established combinatorial-optimization problem whose goal is to find two polarized communities from a signed graph, i.e., two subsets of nodes (disjoint, but not necessarily covering the entire node set) which exhibit a high number of both intra-community positive edges and negative inter-community edges. The state of the art in 2pc suffers from the limitations that (i) existing methods rely on a single (optimal) solution to a continuous relaxation of the problem in order to produce the ultimate discrete solution via rounding, and (ii) 2pc objective function comes with no control on size balance among communities. In this paper, we provide advances to the 2pc problem by addressing both these limitations, with a twofold contribution. First, we devise a novel neural approach that allows for soundly and elegantly explore a variety of suboptimal solutions to the relaxed 2pc problem, so as to pick the one that leads to the best discrete solution after rounding. Second, we introduce a generalization of 2pc objective function – termed γ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\gamma $$\\end{document}-polarity – which fosters size balance among communities, and we incorporate it into the proposed machine-learning framework. Extensive experiments attest high accuracy of our approach, its superiority over the state of the art, and capability of function γ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\gamma $$\\end{document}-polarity to discover high-quality size-balanced communities.

Efficient and lightweight convolutional neural network architecture search methods for object classification

Determining the architecture of deep learning models is a complex task. Several automated search techniques have been proposed, but these methods typically require high-performance graphics processing units (GPUs), manual parameter adjustments, and specific training approaches. This study introduces an efficient, lightweight convolutional neural network architecture search approach tailored for object classification. It features an optimized search space design and a novel controller design.This study introduces a refined search space design incorporating optimizations in both spatial and operational aspects. The focus is on the synergistic integration of convolutional units, dimension reduction units, and the stacking of Convolutional Neural Network (CNN) architectures. To enhance the search space, ShuffleNet modules are integrated, reducing the number of parameters and training time. Additionally, BlurPool is implemented in the dimension reduction unit operation to achieve translational invariance, alleviate the gradient vanishing problem, and optimize unit compositions. Moreover, an innovative controller model, Stage LSTM, is proposed based on Long Short-Term Memory (LSTM) to generate lightweight architectural sequences. In conclusion, the refined search space design and the Stage LSTM controller model are synergistically combined to establish an efficient and lightweight architecture search technique termed Stage and Lightweight Network Architecture Search (SLNAS).The experimental results highlight the superior performance of the optimized search space design, primarily when implemented with the Stage LSTM controller model. This approach shows significantly improved accuracy and stability compared to random, traditional LSTM, and Genetic Algorithm (GA) controller models, with statistically significant differences. Notably, the Stage LSTM controller excels in accuracy while producing models with fewer parameters within the expanded architecture search space.The study adopts the Stage LSTM controller model due to its ability to approximate optimal sequence structures, particularly when combined with the optimized search space design, referred to as SLNAS. SLNAS's performance is evaluated through experiments and comparisons with other Neural Architecture Search (NAS) and object classification methods from different researchers. These experiments consider model parameters, hardware resources, model stability, and multiple datasets. The results show that SLNAS achieves a low error rate of 2.86 % on the CIFAR-10 dataset after just 0.2 days of architecture search, matching the performance of manually designed models but using only 2 % of the parameters. SLNAS consistently demonstrates robust performance across various image classification domains, with an approximate parameter count 700,000.To summarize, SLNAS emerges as a highly effective automated network architecture search method tailored for image classification. It streamlines the model design process, making it accessible to researchers without specialized knowledge in deep learning. Optimizing this method unlocks the full potential of deep learning across diverse research areas. Interested parties can publicly access the source code and pre-trained models through the following link: https://github.com/huanyu-chen/LNASG-and-SLNAS-model.

Deep Learning-Based Real-Time Ureter Identification in Laparoscopic Colorectal Surgery.

Iatrogenic ureteral injury is a serious complication of abdominopelvic surgery. Identifying the ureters intraoperatively is essential to avoid iatrogenic ureteral injury. We developed a model that may minimize this complication. We applied a deep learning-based semantic segmentation algorithm to the ureter recognition task and developed a deep learning model called UreterNet. This study aimed to verify whether the ureters could be identified in videos of laparoscopic colorectal surgery. Semantic segmentation of the ureter area was performed using a convolutional neural network-based approach. Feature Pyramid Networks were used as the convolutional neural network architecture for semantic segmentation. Precision, recall, and the Dice coefficient were used as the evaluation metrics in this study. We created 14,069 annotated images from 304 videos, with 9537, 2266, and 2266 images in the training, validation, and test data sets, respectively. Concerning ureter recognition performance, the precision, recall, and Dice coefficient for the test data were 0.712, 0.722, and 0.716, respectively. Regarding the real-time performance on recorded videos, it took 71 milliseconds for UreterNet to infer all pixels corresponding to the ureter from a single still image and 143 milliseconds to output and display the inferred results as a segmentation mask on the laparoscopic monitor. UreterNet is a noninvasive method for identifying the ureter in videos of laparoscopic colorectal surgery and can potentially improve surgical safety. Although this deep learning model could lead to the development of an image-navigated surgical system, it is necessary to verify whether UreterNet reduces the occurrence of iatrogenic ureteral injury.

Second-Order Sensitivity Neural Network Modeling Approach With Applications to Microwave Devices

Second-Order Sensitivity Neural Network Modeling Approach With Applications to Microwave Devices

Koopman neural operator approach to fast flow prediction of airfoil transonic buffet

Transonic buffet on airfoil is of great importance in the aerodynamic characteristics of aircraft. In the present work, a modified Koopman neural operator (KNO) is applied to predict flow fields during the transonic buffet process of the OAT15A [ONERA (National Office for Aerospace Studies and Research) Aerospatiale Transport aircraft 15 Airfoil] airfoil. Transonic buffet flow with different angles of attack is simulated by Reynolds averaged numerical simulation with the Menter's k−ω shear stress transport (SST) model at Reynolds number Re=3×106. A prediction model is directly constructed between the flow fields at several previous time nodes and that at the future time node by KNO. The predictions of flow fields with single sample and multi samples are performed to demonstrate the prediction accuracy and efficiency of KNO. The prediction of sequence flow fields based on the iterative prediction strategy is achieved for the transonic buffet process. The results indicate that KNO can achieve a fast and accurate prediction of flow physical quantities for the transonic buffet. Compared with other deep learning models including Unet and Fourier neural operator, KNO has a more advanced capability of predicting airfoil transonic buffet flow fields with higher accuracy and efficiency and less hardware requirements.

A Neural Approach for Skin Spectral Reconstruction

A Neural Approach for Skin Spectral Reconstruction

High-Precision Photoacoustic Neural Modulation Uses a Non-Thermal Mechanism.

Neuromodulation is a powerful tool for fundamental studies in neuroscience and potential treatments of neurological disorders. Both photoacoustic (PA) and photothermal (PT) effects are harnessed for non-genetic high-precision neural stimulation. Using a fiber-based device excitable by a nanosecond pulsed laser and a continuous wave laser for PA and PT stimulation, respectively, PA and PT neuromodulation is systematically investigated at the single neuron level. These results show that to achieve the same level of neuron activation recorded by Ca2+ imaging, the laser energy needed for PA stimulation is 1/40 of that needed for PT stimulation. The threshold energy for PA stimulation is found to be further reduced in neurons overexpressing mechano-sensitive channels, indicating direct involvement of mechano-sensitive channels in PA stimulation. Electrophysiology study of single neurons upon PA and PT stimulation is performed by patch clamp recordings. Electrophysiological features induced by PA are distinct from those by PT, confirming that PA and PT stimulation operate through different mechanisms. These insights offer a foundation for the rational design of more efficient and safer non-genetic neural modulation approaches.

HpGraphNEI: A network entity identification model based on heterophilous graph learning

Network entities have important asset mapping, vulnerability, and service delivery applications. In cyberspace, where the network structure is complex and the number of entities is large, effectively obtaining the relevant attributes of entities is a difficult task. Graph neural network-based approaches focus on target IP node messaging from neighboring nodes; however, the graph learning task ignores the heterophilous relationship of network entity identification (NEI) tasks in the graph structure and fails to effectively message from non-neighboring nodes. To address the limitations of the existing task, we propose a NEI model based on heterophilous graph learning (HpGraphNEI); HpGraphNEI converts heterophilous graphs under the NEI task into homophilous graphs and uses the graph learning mechanism to carry out attribute completion task for incomplete entity attributes. First, the acquired dataset is feature-extracted by network measurement, and the clustering algorithm is employed to divide the target nodes into communities. Second, the network topology graph is constructed to embed the node attribute information and neighborhood structure information into the graph in the form of feature vectors. Then, the global attention in the community is calculated according to the attention results, the edges with strong correlation in the network are filtered, the adjacency matrix is reconstructed, and then the updated node information is aggregated to complete the incomplete attribute completion. Fourth, the updated nodes are categorized to output network entity categories and construct network entity portraits based on the attribute completion nodes. We conducted a 2-month data collection in three real regions and successfully identified 6 types of network entities. Compared with the optimal baseline, all the metrics have significantly improved, with NEI accuracy above 93.74% and up to 96.28%, improved 2.27% to 2.69%.

BiasBuster: a Neural Approach for Accurate Estimation of Population Statistics using Biased Location Data.

While extremely useful (e.g., for COVID-19 forecasting and policy-making, urban mobility analysis and marketing, and obtaining business insights), location data collected from mobile devices often contain data from a biased population subset, with some communities over or underrepresented in the collected datasets. As a result, aggregate statistics calculated from such datasets (as is done by various companies including Safegraph, Google, and Facebook), while ignoring the bias, leads to an inaccurate representation of population statistics. Such statistics will not only be generally inaccurate, but the error will disproportionately impact different population subgroups (e.g., because they ignore the underrepresented communities). This has dire consequences, as these datasets are used for sensitive decision-making such as COVID-19 policymaking. This paper tackles the problem of providing accurate population statistics using such biased datasets. We show that statistical debiasing, although in some cases useful, often fails to improve accuracy. We then propose BiasBuster, a neural network approach that utilizes the correlations between population statistics and location characteristics to provide accurate estimates of population statistics. Extensive experiments on real-world data show that BiasBuster improves accuracy by up to 2 times in general and up to 3 times for underrepresented populations.

A one-dimensional convolutional neural network-based deep learning approach for predicting cardiovascular diseases

Early detection of cardiovascular diseases (CVDs) is crucial for managing cardiovascular diseases and improving patient outcomes. Deep neural networks have the potential to reduce the reliance on costly and time-consuming clinical tests, leading to cost savings for patients and healthcare systems. This study proposes the development of specialized convolutional neural networks for the automated selection of essential variables, employing various preprocessing procedures. It evaluates the approach using the UCI repository heart disease dataset, focusing on early-stage heart disease identification to enhance early prediction and intervention for CVD. To address the challenge of achieving higher accuracy, we introduce an approach using one-dimensional convolutional neural networks, incorporating extensive testing to optimize the network architecture and enhance predictive performance. Additionally, recognizing the impact of features on accuracy, a comprehensive data analysis was performed. Through a meticulous selection process, we identified and utilized key features that significantly influenced the accuracy of our model, contributing to more reliable predictions. Finally, cross-validation techniques were implemented to precisely evaluate the efficacy of our work. Numerous experiments were conducted to demonstrate the relevance of our research. The prediction accuracy was found to be 99.95% when employing a train–test approach, while it was approximately 98.53% when employing K-Fold cross-validation. In comparison to existing literature, our approach outperforms a recent best study that proposed a Catboost model, achieving an F1-score of about 92.3% and an average accuracy of 90.94%. This signifies a substantial improvement in predictive performance, with a percentage improvement of approximately 9.90% compared to the Catboost model.

Intelligent fault detection of photovoltaic panel using neural networks

This research primarily aims to leverage artificial neural network technology for diagnosing power output issues in photovoltaic (PV) panels stemming from fluctuations in solar irradiance and temperature. The proposed diagnostic approach relies on constructing a reference model that captures the expected normal operating behavior of the PV panel under fault-free conditions. This reference model is then compared against the actual power output, and the difference, known as the residual, is analyzed to detect potential faults. The neural network is trained using real-world data inputs like solar irradiance and temperature measurements, with the sole output being the power produced by the PV panel. Through training, the neural network learns to map the complex non-linear relationships between environmental inputs and expected power output, effectively modeling the PV system's intrinsic behavior under healthy conditions. Results demonstrate the neural network-based approach's remarkable ability to diagnose faults with high accuracy while avoiding potential non-linear complications. This intelligent monitoring system provides a reliable protocol for early fault detection through training on actual measurements, eliminating the need for complex mathematical models. Consequently, it streamlines the maintenance process by negating intricate procedures to identify PV panel issues. The neural network effectively learns the mapping between environmental conditions and expected power output through exposure to real-world data during training. By analyzing deviations from this learned mapping, represented by the residual signal, the approach can reliably detect anomalies indicative of faults or performance degradation in the PV system. This data-driven nature allows the system to adapt to site-specific characteristics and capture non-linear effects without explicit modeling.

An Artificial Neural Network-Based Approach to Improve Non-Destructive Asphalt Pavement Density Measurement with an Electrical Density Gauge

Asphalt pavement density can be measured using either a destructive or a non-destructive method. The destructive method offers high measurement accuracy but causes damage to the pavement and is inefficient. In contrast, the non-destructive method is highly efficient without damaging the pavement, but its accuracy is not as good as that of the destructive method. Among the devices for non-destructive measurement, the nuclear density gauge (NDG) is the most accurate, but radiation in the device is a serious hazard. The electrical density gauge (EDG), while safer and more convenient to use, is affected by the factors other than density, such as temperature and moisture of the environment. To enhance its accuracy by minimizing or eliminating those non-density factors, an original approach based on artificial neural networks (ANNs) is proposed. Density readings, temperature, and moisture obtained by the EDG are the inputs, and the corresponding densities obtained by the NDG are the outputs to train the ANN models through Levenberg-Marquardt, Bayesian regularization, and Scaled Conjugate Gradient algorithms. Results indicate that the ANN models trained greatly improve the measurement accuracy of the electrical density gauge.

A discretization-free deep neural network-based approach for advection-dispersion-reaction mechanisms

This study aims to provide insights into new areas of artificial intelligence approaches by examining how these techniques can be applied to predict behaviours for difficult physical processes represented by partial differential equations, particularly equations involving nonlinear dispersive behaviours. The current advection-dispersion-reaction equation is one of the key formulas used to depict natural processes with distinct characteristics. It is composed of a first-order advection component, a third-order dispersion term, and a nonlinear response term. Using the deep neural network approach and accounting for physics-informed neural network awareness, the problem has been elaborately discussed. Initial and boundary conditions are added as constraints when the neural networks are trained by minimizing the loss function. In comparison to the existing results, the approach has produced qualitatively correct kink and anti-kink solutions, with losses often remaining around 0.01%. It has also outperformed several traditional discretization-based methods.

An Optimized Artificial Neural Network Model of a Limaçon-to-Circular Gas Expander with an Inlet Valve

In this work, an artificial neural network (ANN)-based model is proposed to describe the input–output relationships in a Limaçon-To-Circular (L2C) gas expander with an inlet valve. The L2C gas expander is a type of energy converter that has great potential to be used in organic Rankine cycle (ORC)-based small-scale power plants. The proposed model predicts the different performance indices of a limaçon gas expander for different input pressures, rotor velocities, and valve cutoff angles. A network model is constructed and optimized for different model parameters to achieve the best prediction performance compared to the classic mathematical model of the system. An overall normalized mean square error of 0.0014, coefficient of determination (R2) of 0.98, and mean average error of 0.0114 are reported. This implies that the surrogate model can effectively mimic the actual model with high precision. The model performance is also compared to a linear interpolation (LI) method. It is found that the proposed ANN model predictions are about 96.53% accurate for a given error threshold, compared to about 91.46% accuracy of the LI method. Thus the proposed model can effectively predict different output parameters of a limaçon gas expander such as energy, filling factor, isentropic efficiency, and mass flow for different operating conditions. Of note, the model is only trained by a set of input and target values; thus, the performance of the model is not affected by the internal complex mathematical models of the overall valved-expander system. This neural network-based approach is highly suitable for optimization, as the alternative iterative analysis of the complex analytical model is time-consuming and requires higher computational resources. A similar modeling approach with some modifications could also be utilized to design controllers for these types of systems that are difficult to model mathematically.

Prediction of on-Base Percentage (OBP) of Baseball Players Using Neural Networks

This research paper presents a detailed study on the prediction of On-Base Percentage (OBP) of baseball players utilizing neural networks. OBP is a crucial statistic in baseball, measuring how frequently a batter reaches base through hits, walks, and hit-by-pitches, excluding errors, fielder's choices, and dropped catches. Traditional methods of predicting OBP have relied heavily on historical statistics and linear models. This study explores the application of neural networks to improve prediction accuracy. We discuss the existing systems, propose a novel neural network-based approach, and evaluate the performance of the model using real-world data. The results demonstrate significant improvements over traditional methods, highlighting the potential of neural networks in sports analytics. Keywords: Baseball, Neural Networks , Analytics, ML.

Metric multidimensional scaling for large single-cell datasets using neural networks

Metric multidimensional scaling is one of the classical methods for embedding data into low-dimensional Euclidean space. It creates the low-dimensional embedding by approximately preserving the pairwise distances between the input points. However, current state-of-the-art approaches only scale to a few thousand data points. For larger data sets such as those occurring in single-cell RNA sequencing experiments, the running time becomes prohibitively large and thus alternative methods such as PCA are widely used instead. Here, we propose a simple neural network-based approach for solving the metric multidimensional scaling problem that is orders of magnitude faster than previous state-of-the-art approaches, and hence scales to data sets with up to a few million cells. At the same time, it provides a non-linear mapping between high- and low-dimensional space that can place previously unseen cells in the same embedding.

ESC-NAS: Environment Sound Classification Using Hardware-Aware Neural Architecture Search for the Edge

The combination of deep-learning and IoT plays a significant role in modern smart solutions, providing the capability of handling task-specific real-time offline operations with improved accuracy and minimised resource consumption. This study provides a novel hardware-aware neural architecture search approach called ESC-NAS, to design and develop deep convolutional neural network architectures specifically tailored for handling raw audio inputs in environmental sound classification applications under limited computational resources. The ESC-NAS process consists of a novel cell-based neural architecture search space built with 2D convolution, batch normalization, and max pooling layers, and capable of extracting features from raw audio. A black-box Bayesian optimization search strategy explores the search space and the resulting model architectures are evaluated through hardware simulation. The models obtained from the ESC-NAS process achieved the optimal trade-off between model performance and resource consumption compared to the existing literature. The ESC-NAS models achieved accuracies of 85.78%, 81.25%, 96.25%, and 81.0% for the FSC22, UrbanSound8K, ESC-10, and ESC-50 datasets, respectively, with optimal model sizes and parameter counts for edge deployment.