Neural Network Research Articles

Sensor for Real-Time Glucose Measurement in Aqueous Media based on Nanomaterials Incorporating an Artificial Neural Network Algorithm on a System-On-Chip

The aim of this paper is to present the development of a real-time measurement system for glucose in aqueous media. The proposed system incorporates two lines of research: i) design, synthesis, and implementation of a non-enzymatic electrochemical sensor of Multi-Walled Carbon Nanotubes with Copper nanoparticles (MWCNT-Cu) and ii) design and implementation of a machine learning algorithm based on an Artificial Neural Network Multilayer Perceptron (ANN-MLP), which is embedded in an ESP32 SoC (System on Chip). From the current data that is extracted in real-time during the oxidation-reduction process to which an aqueous medium is subjected, it feeds the algorithm embedded in the ESP32 SoC to estimate the glucose value. The experimental results show that the nanostructured sensor improves the resolution in the amperometric response by identifying an ideal place for data collection. For its part, the incorporation of the algorithm based on an ANN embedded in a SoC provides a level of 97.8 % accuracy in the measurements. It is concluded that incorporating machine learning algorithms embedded in low-cost SoC in complex experimental processes improves data manipulation, increases the reliability of results, and adds portability.

Predicting power and solar energy using neural networks and PCA with meteorological parameters from Diass and Taïba Ndiaye

The excessive reliance on conventional fossil fuel-based resources poses a significant threat to our environment. To mitigate this impact, it has become increasingly crucial to increase the integration of intermittent and non-polluting energy sources into our electrical grids. However, while this higher penetration rate brings benefits such as improved producer satisfaction and reduced fossil fuel consumption, it also presents challenges for traditional non-smart electrical networks. To promote intermittent energy sources effectively and maintain a balance between consumption and production, accurate forecasting of these energy outputs plays a vital role. This research paper focuses on studying the application of artificial neural networks for predicting the power and energy output of the Diass solar power plant in the short and medium term. The proposed approach utilizes not only the meteorological data from the city where the power plant is located but also data from a nearby city with a data acquisition station. Principal component analysis (PCA) is employed to select the relevant variables for the prediction model. Furthermore, the results obtained from our approach are compared to existing literature that solely uses meteorological data from the power plant's location. The comparison shows that our method achieves more satisfactory results, with mean absolute errors and root mean square errors of 0.0223 KWh and 0.003 KWh, respectively, and a prediction accuracy of 94.57% in terms of energy and power. It is worth noting that the computational resource requirements for our approach are higher, with simulation times ranging between 1788 seconds and 2201 seconds. By utilizing a broader range of data sources and employing advanced techniques like artificial neural networks, this research contributes to improving the accuracy of solar power generation forecasts. The findings highlight the potential of incorporating additional data inputs and advanced modeling techniques to enhance the performance of renewable energy systems, paving the way for a more sustainable and efficient energy future.

A Lightweight Convolutional Neural Network for Salient Object Detection

A Lightweight Convolutional Neural Network for Salient Object Detection

Neighborhood Temperature and Contextual Factors Improves Prediction of Childhood Obesity: An Application of Novel Graph Neural Networks

Neighborhood Temperature and Contextual Factors Improves Prediction of Childhood Obesity: An Application of Novel Graph Neural Networks

INTEGRATED CONVOLUTIONAL NEURAL NETWORKS APPLICATION FOR ROAD TRAFFIC EFFICIENCY

Road traffic is a problem that every country has to face for the benefit of its population. Everything related to the transportation of people, goods, and services can be delayed due to a deplorable infrastructure. From minor tasks like individual shopping to important services like ambulances and fire trucks, all vehicles in Romania will eventually pass through a traffic light-controlled intersection that is not managed in real-time and that is also highly inefficient. At such a location, it is critical for traffic flow to be as smooth as possible to prevent time losses, and for traffic information to be stored into a cloud for process optimization. The purpose of the developed application is to reduce the real-time traffic jams to allow the optimal vehicles movement in safe conditions for all traffic participants. Using deep learning (DL) and artificial intelligence (AI) techniques, the application is based on two convolutional neural networks (CNN’s), respectively one for real-time detection of cars and people, and another one for real-time detection of cars license plates. The cars detections with CNN’s is important for road traffic optimization, for the efficient pedestrian detection necessary for crosswalks, and the car licence plates detection necessary for preventing traffic violations. The developed application contains several modules that operate based on deep learning and devices that communicate with each other, ensuring efficient data saving in cloud, data visualization, real-time traffic light control, and the information processing from traffic cameras.

Research status and prospect of the application of artificial intelligence in the acupuncture and moxibustion field based on bibliometric

To explore the research hotspot, development trend and existing problem of artificial intelligence (AI) application in the field of acupuncture and moxibustion by using bibliometric method. Relevant articles of AI application in the field of acupuncture and moxibustion published in CNKI, Wanfang, VIP, PubMed and Web of Science from the database establishment to September 17, 2023 were searched. Excel 2019, CiteSpace 6.2.R4 and VOSviewer 1.6.19 were used to draw visual map of the number of publication, authors, research institutions and keywords, and further analyzed the research hotspot and trend. A total of 443 Chinese articles and 68 English articles were included. The number of annual publication of Chinese articles showed an overall increasing trend, while the number of publication of English articles was less, with a growing trend from 2020. Keywords analysis showed that "Machine learning", "neural network", "deep learning", "data mining", "robot" and other AI technologies were developing around acupuncture and moxibustion diagnosis and treatment of disease, efficacy prediction, teaching and intelligent equipment development. However, the research on the application of AI in the field of acupuncture and moxibustion is in the preliminary stage of development, and in the future, it is necessary to strengthen the communication and cooperation among the teams, to further explore the AI system in line with the characteristics of acupuncture and moxibustion diagnosis and treatment, and to promote the development of the digitalisation, intellectualisation and industrialisation of acupuncture and moxibustion.

Artificial intelligence algorithm for neoplastic cell percentage estimation and its application to copy number variation in urinary tract cancer.

Bladder cancer is characterized by frequent mutations, which provide potential therapeutic targets for most patients. The effectiveness of emerging personalized therapies depends on an accurate molecular diagnosis, for which the accurate estimation of the neoplastic cell percentage (NCP) is a crucial initial step. However, the established method for determining the NCP, manual counting by a pathologist, is time-consuming and not easily executable. To address this, artificial intelligence (AI) models were developed to estimate the NCP using nine convolutional neural networks and the scanned images of 39 cases of urinary tract cancer. The performance of the AI models was compared to that of six pathologists for 119 cases in the validation cohort. The ground truth value was obtained through multiplexed immunofluorescence. The AI model was then applied to 41 cases in the application cohort that underwent next-generation sequencing testing, and its impact on the copy number variation (CNV) was analyzed. Each AI model demonstrated high reliability, with intraclass correlation coefficients (ICCs) ranging from 0.82 to 0.88. These values were comparable or better to those of pathologists, whose ICCs ranged from 0.78 to 0.91 in urothelial carcinoma cases, both with and without divergent differentiation/ subtypes. After applying AI-driven NCP, 190 CNV (24.2%) were reclassified with 66 (8.4%) and 78 (9.9%) moved to amplification and loss, respectively, from neutral/minor CNV. The neutral/minor CNV proportion decreased by 6%. These results suggest that AI models could assist human pathologists in repetitive and cumbersome NCP calculations.

Standoff Identification of Plastic Waste Using a Low-Cost Compact Laser-Induced Breakdown Spectroscopy (LIBS) Detection System.

We report the standoff/remote identification of post-consumer plastic waste by utilizing a low-cost and compact standoff laser-induced breakdown spectroscopy (ST-LIBS) detection system. A single plano-convex lens is used for collecting the optical emissions from the plasma at a standoff distance of 6.5 m. A compact non-gated Czerny-Turner charge-coupled device (CCD) spectrometer (CT-CCD) is utilized to analyze the optical response. The single lens and CT-CCD combination not only reduces the cost of the detection system by tenfold, but also decreases the collection system size and weight compared to heavy telescopic-based intensified CCD systems. All the samples investigated in this study were collected from a local recycling plant. All the measurements were performed with only a single laser shot which enables rapid identification while probing a large number of samples in real time. Furthermore, principal component analysis has shown excellent separation among the samples and an artificial neural network analysis has revealed that plastic waste can be identified within ∼10 ms only (testing time) with accuracies up to ∼99%. Finally, these results have the potential to build a compact and low-cost ST-LIBS detection system for the rapid identification of plastic waste for real-time waste management applications.

Detection and tracking of barchan dunes using artificial intelligence

Barchans are crescent-shape dunes ubiquitous on Earth and other celestial bodies, which are organized in barchan fields where they interact with each other. Over the last decades, satellite images have been largely employed to detect barchans on Earth and on the surface of Mars, with AI (Artificial Intelligence) becoming an important tool for monitoring those bedforms. However, automatic detection reported in previous works is limited to isolated dunes and does not identify successfully groups of interacting barchans. In this paper, we inquire into the automatic detection and tracking of barchans by carrying out experiments and exploring the acquired images using AI. After training a neural network with images from controlled experiments where complex interactions took place between dunes, we did the same for satellite images from Earth and Mars. We show, for the first time, that a neural network trained properly can identify and track barchans interacting with each other in different environments, using different image types (contrasts, colors, points of view, resolutions, etc.), with confidence scores (accuracy) above 70%. Our results represent a step further for automatically monitoring barchans, with important applications for human activities on Earth, Mars and other celestial bodies.

Transfer learning may explain pigeons' ability to detect cancer in histopathology.

Pigeons' unexpected competence in learning to categorize unseen histopathological images has remained an unexplained discovery for almost a decade (Levensonet al2015PLoS One10e0141357). Could it be that knowledge transferred from their bird's-eye views of the earth's surface gleaned during flight contributes to this ability? Employing a simulation-based verification strategy, we recapitulate this biological phenomenon with a machine-learning analog. We model pigeons' visual experience during flight with the self-supervised pre-training of a deep neural network on BirdsEyeViewNet; our large-scale aerial imagery dataset. As an analog of the differential food reinforcement performed in Levensonet al's study 2015PLoS One10e0141357), we apply transfer learning from this pre-trained model to the same Hematoxylin and Eosin (H&E) histopathology and radiology images and tasks that the pigeons were trained and tested on. The study demonstrates that pre-training neural networks with bird's-eye view data results in close agreement with pigeons' performance. These results support transfer learning as a reasonable computational model of pigeon representation learning. This is further validated with six large-scale downstream classification tasks using H&E stained whole slide image datasets representing diverse cancer types.

Aircraft ice accretion prediction based on geometrical constraints enhancement neural networks

Purpose The purpose of this study is to establish a novel airfoil icing prediction model using deep learning with geometrical constraints, called geometrical constraints enhancement neural networks, to improve the prediction accuracy compared to the non-geometrical constraints model. Design/methodology/approach The model is developed with flight velocity, ambient temperature, liquid water content, median volumetric diameter and icing time taken as inputs and icing thickness given as outputs. To enhance the icing prediction accuracy, the model involves geometrical constraints into the loss function. Then the model is trained according to icing samples of 2D NACA0012 airfoil acquired by numerical simulation. Findings The results show that the involvement of geometrical constraints effectively enhances the prediction accuracy of ice shape, by weakening the appearance of fluctuation features. After training, the airfoil icing prediction model can be used for quickly predicting airfoil icing. Originality/value This work involves geometrical constraints in airfoil icing prediction model. The proposed model has reasonable capability in the fast assessment of aircraft icing.

Assessment of future drought sensitivity shift associated with crop yield loss risk in the Anthropocene

ABSTRACT This study proposed a framework using artificial neural networks (ANNs), the designed residual regression method, and probabilistic indexing, for drought sensitivity shift analysis, associated with future yield loss risk in Anthropocene. The temporal assessment of the drought index (standardized precipitation evapotranspiration index, SPEI) illustrated prevalent occurrence of mild to moderate drought with a fluctuating trend. Furthermore, the optimal ANN model, with R2 = 0.86 for training and 0.7 for testing, was used to project yield for future analysis. The projected spatial changes in drought sensitivity patterns suggest most of the area will experience an increased sensitivity to drought in the far future (2036–2050). The estimated probability loss under the different risk levels defined by the yield reduction rate was 6.2–55.5%. Furthermore, the spatial distribution of the designed risk index (IR) revealed the districts with the highest probability of vulnerability and showed a probable increase in future vulnerability.

The Calculating Brain.

The human brain possesses neural networks and mechanisms enabling the representation of numbers, basic arithmetic operations, and mathematical reasoning. Without the ability to represent numerical quantity and perform calculations, our scientifically and technically advanced culture would not exist. However, the origins of numerical abilities are grounded in an intuitive understanding of quantity deeply rooted in biology. Nevertheless, more advanced symbolic arithmetic skills necessitate a cultural background with formal mathematical education. In the past two decades, cognitive neuroscience has seen significant progress in understanding the workings of the calculating brain through various methods and model systems. This review begins by exploring the mental and neuronal representations of non-symbolic numerical quantity, then progresses to symbolic representations acquired in childhood. During arithmetic operations (addition, subtraction, multiplication, and division), these representations are processed and transformed according to arithmetic rules and principles, leveraging different mental strategies and types of arithmetic knowledge that can be dissociated in the brain. While it was once believed that number processing and calculation originated from the language faculty, it is now evident that mathematical and linguistic abilities are primarily processed independently in the brain. Understanding how the healthy brain processes numerical information is crucial for gaining insights into debilitating numerical disorders, including acquired conditions like acalculia and learning-related calculation disorders such as developmental dyscalculia.

Attention improvement for data-driven analyzing fluorescence excitation-emission matrix spectra via interpretable attention mechanism

Analyzing three-dimensional excitation-emission matrix (3D-EEM) spectra through machine learning models has drawn increasing attention, whereas the reliability of these machine learning models remains unclear due to their “black box” nature. In this study, the convolutional neural network (CNN) for classifying numbers of fluorescent components in 3D-EEM spectra was interpreted by gradient-weighted class activation mapping (Grad-CAM), guided Grad-CAM, and structured attention graphs (SAGs). Results showed that the original CNN classifier with high classification accuracy may make a classification based on misleading attention to the non-fluorescence area in 3D-EEM spectra. By removing Rayleigh scatterings in 3D-EEM spectra and integrating convolutional block attention module (CBAM) in CNN classifiers, the correct attention of the trained CNN classifier with CBAM greatly increased from 17.6% to 57.2%. This work formulated strategies for improving CNN classifiers associated with environmental fields and would provide great help for water determination in both natural and artificial environments.

Assessment of the potential protective effectsof culture filtrate ofTrichoderma harzianum to ameliorate the damagedhistoarchitecture of brain in epileptic rats.

The simultaneous hyperexcitability of the neural network is the most well-known manifestation of epilepsy that causes recurrent seizures. The current study was aimed to examine any potential safety benefits of the culture filtrate of Trichoderma harzianum (ThCF) to ameliorate damaged histoarchitecture of the brain in epileptic rats by assessing seizure intensity scale and behavioral impairments and follow up the spontaneous motor seizures during status epilepticus phases in rats. Twenty-four rats were divided into four groups; control (C), epileptic (EP) valproic acid-treated epileptic (EP-VPA), and epileptic treated with T. harzianum cultured filtrate (ThCF). In addition to a seizure intensity score and behavioral tests, routine H&E and Golgi-Copsch histopathology, were used to examine the cell somas, dendrites, axons, and neural spines. ThCF treatment increased activity and recorded movements during grooming, rearing, and ambulation frequency. Brain tissues of epileptic rats exhibited detached meninges, hypercellularity, mild edema in the cortex and markedly degenerated neurons, degenerated glial cells, and microcyst formation in the hippocampus. Moreover, brains of EP-ThCF were noticed with average blood vessels, and increased dendritogenesis. The current data revealed some of negative effects of epileptogenesis brought on by seizure intensity score and retarded histopathological alterations in the hippocampus. Therefore, the study is forecasting to identify novel active components from the metabolites of T. harzianum with a crucial therapeutic role in various disorders.

Am I Hurt?: Evaluating Psychological Pain Detection in Hindi Text Using Transformer-based Models

The automated evaluation of pain is critical for developing effective pain management approaches that seek to alleviate pain while preserving patients’ functioning. Transformer-based models can aid in detecting pain from Hindi text data gathered from social media by leveraging their ability to capture complex language patterns and contextual information. By understanding the nuances and context of Hindi text, transformer models can effectively identify linguistic cues and sentiments and expressions associated with pain, enabling the detection and analysis of pain-related content present in social media posts. The purpose of this research is to analyze the feasibility of utilizing NLP techniques to automatically identify pain within Hindi textual data, providing a valuable tool for pain assessment in Hindi-speaking populations. The research showcases the HindiPainNet model, a deep neural network that employs the IndicBERT model, classifying the dataset into two class labels {pain, no_pain} for detecting pain in Hindi textual data. The model is trained and tested using a novel dataset, दर्द-ए-शायरी (pronounced as Dard-e-Shayari ), curated using posts from social media platforms. The results demonstrate the model's effectiveness, achieving an accuracy of 70.5%. This pioneer research highlights the potential of utilizing textual data from diverse sources to identify and understand pain experiences based on psychosocial factors. This research could pave the path for the development of automated pain assessment tools that help medical professionals comprehend and treat pain in Hindi-speaking populations. Additionally, it opens avenues to conduct further NLP-based multilingual pain detection research, addressing the needs of diverse language communities.

DCE-Qnet: deep network quantification of dynamic contrast enhanced (DCE) MRI.

Quantification of dynamic contrast-enhanced (DCE)-MRI has the potential to provide valuable clinical information, but robust pharmacokinetic modeling remains a challenge for clinical adoption. A 7-layer neural network called DCE-Qnet was trained on simulated DCE-MRI signals derived from the Extended Tofts model with the Parker arterial input function. Network training incorporated B1 inhomogeneities to estimate perfusion (Ktrans, vp, ve), tissue T1 relaxation, proton density and bolus arrival time (BAT). The accuracy was tested in a digital phantom in comparison to a conventional nonlinear least-squares fitting (NLSQ). In vivo testing was conducted in ten healthy subjects. Regions of interest in the cervix and uterine myometrium were used to calculate the inter-subject variability. The clinical utility was demonstrated on a cervical cancer patient. Test-retest experiments were used to assess reproducibility of the parameter maps in the tumor. The DCE-Qnet reconstruction outperformed NLSQ in the phantom. The coefficient of variation (CV) in the healthy cervix varied between 5 and 51% depending on the parameter. Parameter values in the tumor agreed with previous studies despite differences in methodology. The CV in the tumor varied between 1 and 47%. The proposed approach provides comprehensive DCE-MRI quantification from a single acquisition. DCE-Qnet eliminates the need for separate T1 scan or BAT processing, leading to a reduction of 10 min per scan and more accurate quantification.

Artificial intelligence methods available for cancer research.

Cancer is a heterogeneous and multifaceted disease with a significant global footprint. Despite substantial technological advancements for battling cancer, early diagnosis and selection of effective treatment remains a challenge. With the convenience of large-scale datasets including multiple levels of data, new bioinformatic tools are needed to transform this wealth of information into clinically useful decision-support tools. In this field, artificial intelligence (AI) technologies with their highly diverse applications are rapidly gaining ground. Machine learning methods, such as Bayesian networks, support vector machines, decision trees, random forests, gradient boosting, and K-nearest neighbors, including neural network models like deep learning, have proven valuable in predictive, prognostic, and diagnostic studies. Researchers have recently employed large language models to tackle new dimensions of problems. However, leveraging the opportunity to utilize AI in clinical settings will require surpassing significant obstacles-a major issue is the lack of use of the available reporting guidelines obstructing the reproducibility of published studies. In this review, we discuss the applications of AI methods and explore their benefits and limitations. We summarize the available guidelines for AI in healthcare and highlight the potential role and impact of AI models on future directions in cancer research.

Secure and Reliable Intrusion Detection Scheme for Software-Defined Networking Using LFTS-Rnn and PC-JTFOA

In recent times, Software Defined Networking (SDN) has gained immense popularity among communication networks due to its high network connectivity and security. But, owing to the lack of sufficient security schemes and dynamic access control, conventional techniques could be more efficient. Hence, this paper proposes a secure intrusion detection scheme for SDN utilizing LogishFTS-based Recurrent Neural Networks (LFTS-RNN) and Poisson Clumping Japanese Tree Frog Optimization Algorithm (PC-JTFOA). Initially, by using a username, password, and Signed Access Control (SAC), the registered users log in to the network. The Entropy Makwa-based Digital Signature Algorithm (EMDSA) is utilized to identify the attackers during the login process. Then, the Intrusion Detection System (IDS) is trained using the historical dataset. The critical features are extracted in IDS; subsequently, by using the PC-JTFOA, the optimal features are selected. Likewise, the selected features are inputted into the LFTS-RNN to recognize the intruders in the application plane. Moreover, the process of data security and load balancing is undergone by the data. Afterward, the balanced data is subjected to IDS to detect the attackers in the control plane. According to experimental outcomes, the proposed technique obtained a higher prominence with an accuracy of 98.35%.

Fully forward mode training for optical neural networks

Optical computing promises to improve the speed and energy efficiency of machine learning applications1–6. However, current approaches to efficiently train these models are limited by in silico emulation on digital computers. Here we develop a method called fully forward mode (FFM) learning, which implements the compute-intensive training process on the physical system. The majority of the machine learning operations are thus efficiently conducted in parallel on site, alleviating numerical modelling constraints. In free-space and integrated photonics, we experimentally demonstrate optical systems with state-of-the-art performances for a given network size. FFM learning shows training the deepest optical neural networks with millions of parameters achieves accuracy equivalent to the ideal model. It supports all-optical focusing through scattering media with a resolution of the diffraction limit; it can also image in parallel the objects hidden outside the direct line of sight at over a kilohertz frame rate and can conduct all-optical processing with light intensity as weak as subphoton per pixel (5.40 × 1018- operations-per-second-per-watt energy efficiency) at room temperature. Furthermore, we prove that FFM learning can automatically search non-Hermitian exceptional points without an analytical model. FFM learning not only facilitates orders-of-magnitude-faster learning processes, but can also advance applied and theoretical fields such as deep neural networks, ultrasensitive perception and topological photonics.