Search

High-throughput technologies and machine learning (ML), when applied to a huge pool of medical data such as omics data, result in efficient analysis. Recent research aims to apply and develop ML models to predict a disease well in time using available omics datasets. The present work proposed a framework, 'OmicPredict', deploying a hybrid feature selection method and deep neural network (DNN) model to predict multiple diseases using omics data. The hybrid feature selection method is developed using the Analysis of Variance (ANOVA) technique and firefly algorithm. The OmicPredict framework is applied to three case studies, Alzheimer's disease, Breast cancer, and Coronavirus disease 2019 (COVID-19). In the case study of Alzheimer's disease, the framework predicts patients using GSE33000 and GSE44770 dataset. In the case study of Breast cancer, the framework predicts human epidermal growth factor receptor 2 (HER2) subtype status using Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) dataset. In the case study of COVID-19, the framework performs patients' classification using GSE157103 dataset. The experimental results show that DNN model achieved an Area Under Curve (AUC) score of 0.949 for the Alzheimer's (GSE33000 and GSE44770) dataset. Furthermore, it achieved an AUC score of 0.987 and 0.989 for breast cancer (METABRIC) and COVID-19 (GSE157103) datasets, respectively, outperforming Random Forest, Naïve Bayes models, and the existing research. [ABSTRACT FROM AUTHOR]

The Internet of Things (IoT) has gained widespread importance in recent time. However, the related issues of security and privacy persist in such IoT networks. Owing to device limitations in terms of computational power and storage, standard protection approaches cannot be deployed. In this article, we propose a lightweight distributed intrusion detection system (IDS) framework, called FCAFE‐BNET (Fog based Context Aware Feature Extraction using BranchyNET). The proposed FCAFE‐BNET approach considers versatile network conditions, such as varying bandwidths and data loads, while allocating inference tasks to cloud/edge resources. FCAFE‐BNET is able to adjust to dynamic network conditions. This can be advantageous for applications with particular quality of service requirements, such as video streaming or real‐time communication, ensuring a steady and reliable performance. Early exit deep neural networks (DNNs) have been employed for faster inference generation at the edge. Often, the weights that the model learns in the initial layer may be sufficiently qualified to perform the required classification tasks. Instead of using subsequent layers of DNNs for generating the inference, we have employed the early‐exit mechanism in the DNNs. Such DNNs help to predict a wide range of testing samples through these early‐exit branches, upon crossing a threshold. This method maintains the confidence values corresponding to the inference. Employing this approach, we achieved a faster inference, with significantly high accuracy. Comparative studies exploit manual feature extraction techniques, that can potentially overlook certain valuable patterns, thus degrading classification performance. The proposed framework converts textual/tabular data into 2‐D images, allowing the DNN model to autonomously learns its own features. This conversion scheme facilitated the identification of various intrusion types, ranging from 5 to 14 different categories. FCAFE‐BNET works for both network‐based and host‐based IDS: NIDS and HIDS. Our experiments demonstrate that, in comparison with recent approaches, FCAFE‐BNET achieves a 39.12%–50.23% reduction in the total inference time on benchmark real‐world datasets, such as: NSL‐KDD, UNSW‐NB 15, ToN_IoT, and ADFA_LD. [ABSTRACT FROM AUTHOR]

Complex applications, as well as multimedia services, are increasing swiftly with the substantial growth of Mobile Users (MUs) and IoT devices; thus, demanding extra computations along with higher data communication. Nevertheless, with limited computation power along with energy, these devices are still resource-constrained. Numerous methodologies have been developed for efficient Resource Allocation (RA) in the cloud; however, owing to the loss of network resources, energy consumption, along lower battery lifetime, the prevailing methodologies are deemed to be ineffective. So, by utilizing an Entropy-centric Fish Swarm Optimization Algorithm (E-FSO) with Soft Swish Deep Dense Neural Network (SS-DDNN)-guided Crossover Mutation – Linear Scaling-centered Golden Eagle Optimization (CM-LSGEO) technique, a practical localization-centric safe RA has been suggested to help with the previously mentioned issues. Initially, the mobile's optimal location is detected by utilizing the E-FSO algorithm. Next, for the secure transmission of packets, the failure node is detected effectively by employing the SS-DDNN algorithm. By utilizing the CM-LSGEO algorithm, a model has been constructed to perform efficient RA. The performance of the suggested model is compared to that of the current models in the experimental assessment, which takes into account some metrics. The outcomes displayed that in comparison with the prevailing methodologies, the proposed framework is more efficient. [ABSTRACT FROM AUTHOR]

Medium Resolution Spectral Imager-II (MERSI-II) on FengYun 3D (FY-3D) meteorological satellite is a visible and infrared spectral imaging instrument comparable with Moderate Resolution Imaging Spectroradiometer (MODIS), which provides ample opportunities for regional crop mapping. However, current research may have overlooked the potential of combining FY-3D MERSI-II data with deep neural network (DNN) algorithms for winter wheat identification. This research utilizes a DNN method to train a nonlinear model with the synthetic normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) at a 250 m resolution from FY-3D MERSI-II as the eigenvalues, enabling to accurately identify spatial distribution of winter wheat in China’s main production regions and filling the gap in the application of FY-3D MERSI-II data. The results indicate that the winter wheat identification with the combination of NDVI and EVI achieved an accuracy of 95.32%, with a Kappa coefficient of 0.87 and a determination coefficient (R2) of 0.73, improving the overall accuracy by 1.27% and 1.07% compared to models trained solely using NDVI or EVI, respectively. In terms of spatial distribution, the overlap rate with the winter wheat dataset of China from MODIS at a 500 m resolution exceeded 95%, with an area bias of 2.67%. This study indicates that the medium resolution FY-3D MERSI-II remote sensing dataset combined with the DNN algorithm can accurately identify crop information over large areas. [ABSTRACT FROM AUTHOR]

Additive Manufacturing (AM) is a pivotal technology for transforming complex geometries with minimal tooling requirements. Among the several AM techniques, Wire Arc Additive Manufacturing (WAAM) is notable for its ability to produce large metal components, which makes it particularly appealing in the aerospace sector. However, precise control of the bead geometry, specifically bead width and height, is essential for maintaining the structural integrity of WAAM-manufactured parts. This paper introduces a methodology using a Deep Neural Network (DNN) model for forecasting the bead geometry in the WAAM process, focusing on gas metal arc welding cold metal transfer (GMAW-CMT) WAAM. This study addresses the challenges of bead geometry prediction by developing a robust predictive framework. Key process parameters, such as the wire travel speed, wire feed rate, and bead dimensions of the previous layer, were monitored using a Coordinate Measuring Machine (CMM) to ensure precision. The collected data were used to train and validate various regression models, including linear regression, ridge regression, regression, polynomial regression (Quadratic and Cubic), Random Forest, and a custom-designed DNN. Among these, the Random Forest and DNN models were particularly effective, with the DNN showing significant accuracy owing to its ability to learn complex nonlinear relationships inherent in the WAAM process. The DNN model architecture consists of multiple hidden layers with varying neuron counts, trained using backpropagation, and optimized using the Adam optimizer. The model achieved mean absolute percentage error (MAPE) values of 0.014% for the width and 0.012% for the height, and root mean squared error (RMSE) values of 0.122 for the width and 0.153 for the height. These results highlight the superior capability of the DNN model in predicting bead geometry compared to other regression models, including the Random Forest and traditional regression techniques. These findings emphasize the potential of deep learning techniques to enhance the accuracy and efficiency of WAAM processes. [ABSTRACT FROM AUTHOR]

This research work develops a new framework that combines patient feedback with evidence-based best practices across disease states to improve drug recommendations. It employs BERT as its free-text processing engine to deal with sentiment judgment and classification. The functionality of the system, named 'PharmaBERT', includes acceptance of drug review data as a comprehensive input, drug categorization when dealing with a wide range of treatments and fine-tuning the BERT-based model for gaining positive or negative sentiment towards specific medications. PharmaBERT categorizes various drugs and fine-tunes the BERT structure to perceive lots of possible sentiments for specific medications. Consequently, PharmaBERT brings all its training and optimization capabilities together and through this, the system reaches a higher accuracy of up to 91% thus showcasing the potency of the model in capturing patient sentiments. While being a BERT spinoff, PharmaBERT utilizes its own set of experienced techniques to comprehend and sense the health-related text input given by the patient, doctor, or pharmacist. It uses transfer learning, that is, it learns from language representations to adapt quickly to the intricacies of drug reviewing. Through PharmaBERT, healthcare professionals may expand their diagnoses with insights from patient feedback to constitute more neutral decisions. [ABSTRACT FROM AUTHOR]

This study develops a Convolutional Autoencoder (CAE) and deep neural network (DNN)-based model optimized for real-time signal processing and high accuracy in motor fault diagnosis. This model learns complex patterns from voltage and current data and precisely analyzes them in combination with DNN through latent space representation. Traditional diagnostic methods relied on vibration and current sensors, empirical knowledge, or harmonic and threshold-based monitoring, but they had limitations in recognizing complex patterns and providing accurate diagnoses. Our model significantly enhances the accuracy of power data analysis and fault diagnosis by mapping each phase (R, S, and T) of the electrical system to the red, green, and blue (RGB) channels of image processing and applying various signal processing techniques. Optimized for real-time data streaming, this model demonstrated high practicality and effectiveness in an actual industrial environment, achieving 99.9% accuracy, 99.8% recall, and 99.9% precision. Specifically, it was able to more accurately diagnose motor efficiency and fault risks by utilizing power system analysis indicators such as phase voltage, total harmonic distortion (THD), and voltage unbalance. This integrated approach significantly enhances the real-time applicability of electric motor fault diagnosis and is expected to provide a crucial foundation for various industrial applications in the future. [ABSTRACT FROM AUTHOR]

Video Coding for Machines (VCM) is gaining momentum in applications like autonomous driving, industry manufacturing, and surveillance, where the robustness of machine learning algorithms against coding artifacts is one of the key success factors. This work complements the MPEG/JVET standardization efforts in improving the resilience of deep neural network (DNN)-based machine models against such coding artifacts by proposing the following three advanced fine-tuning procedures for their training: (1) the progressive increase of the distortion strength as the training proceeds; (2) the incorporation of a regularization term in the original loss function to minimize the distance between predictions on compressed and original content; and (3) a joint training procedure that combines the proposed two approaches. These proposals were evaluated against a conventional fine-tuning anchor on two different machine tasks and datasets: image classification on ImageNet and semantic segmentation on Cityscapes. Our joint training procedure is shown to reduce the training time in both cases and still obtain a 2.4% coding gain in image classification and 7.4% in semantic segmentation, whereas a slight increase in training time can bring up to 9.4% better coding efficiency for the segmentation. All these coding gains are obtained without any additional inference or encoding time. As these advanced fine-tuning procedures are standard-compliant, they offer the potential to have a significant impact on visual coding for machine applications. [ABSTRACT FROM AUTHOR]

Email phishing is a social engineering scheme that uses spoofed emails intended to trick the user into disclosing legitimate business and personal credentials. Many phishing email detection techniques exist based on machine learning, deep learning, and word embedding. In this paper, we propose a new technique for the detection of phishing emails using word embedding (Word2Vec, FastText, and TF-IDF) and deep learning techniques (DNN and BiLSTM network). Our proposed technique makes use of only four header based (From, Returnpath, Subject, Message-ID) features of the emails for the email classification. We applied several word embeddings for the evaluation of our models. From the experimental evaluation, we observed that the DNN model with FastText-SkipGram achieved an accuracy of 99.52% and BiLSTM model with FastText-SkipGram achieved an accuracy of 99.42%. Among these two techniques, DNN outperformed BiLSTM using the same word embedding (FastText-SkipGram) techniques with an accuracy of 99.52%. [ABSTRACT FROM AUTHOR]

The COVID-19 pandemic has highlighted that effective early infection detection methods are essential, as they play a critical role in controlling the epidemic spread. In this work, we investigate the use of wearable sensors in conjunction with machine learning (ML) techniques for pandemic infection detection. We work on designing a wristband that measures various vital parameters such as temperature, heart rate, and SPO2, and transmits them to a mobile application using Bluetooth Low Energy. The accuracy of the wristband measurements is shown to be within 10% of the readings of existing commercial products. The measured data can be used and analyzed for various purposes. To benefit from the existing online datasets related to COVID-19, we use this pandemic as an example in our work. Hence, we also develop ML-based models that use the measured vital parameters along with cough sounds in order to determine whether a case is COVID-19 positive or not. The proposed models are shown to achieve remarkable results, exceeding 90% accuracy. One of our proposed models exceeds 96% performance in terms of accuracy, precision, recall, and F1-Score. The system lends itself reasonably for amendment to deal with future pandemics by considering their specific features and designing the ML models accordingly. Furthermore, we design and develop a mobile application that shows the data collected from the wristband, records cough sounds, runs the ML model, and provides feedback to the user about their health status in a user-friendly, intuitive manner. A successful deployment of such an approach would decrease the load on hospitals and prevent infection from overcrowded spaces inside the hospital. [ABSTRACT FROM AUTHOR]

Signal detection plays an essential role in massive Multiple-Input Multiple-Output (MIMO) systems. However, existing detection methods have not yet made a good tradeoff between Bit Error Rate (BER) and computational complexity, resulting in slow convergence or high complexity. To address this issue, a low-complexity Approximate Message Passing (AMP) detection algorithm with Deep Neural Network (DNN) (denoted as AMP-DNN) is investigated in this paper. Firstly, an efficient AMP detection algorithm is derived by scalarizing the simplification of Belief Propagation (BP) algorithm. Secondly, by unfolding the obtained AMP detection algorithm, a DNN is specifically designed for the optimal performance gain. For the proposed AMP-DNN, the number of trainable parameters is only related to that of layers, regardless of modulation scheme, antenna number and matrix calculation, thus facilitating fast and stable training of the network. In addition, the AMP-DNN can detect different channels under the same distribution with only one training. The superior performance of the AMP-DNN is also verified by theoretical analysis and experiments. It is found that the proposed algorithm enables the reduction of BER without signal prior information, especially in the spatially correlated channel, and has a lower computational complexity compared with existing state-of-the-art methods.

Abstract Video Coding for Machines (VCM) is gaining momentum in applications like autonomous driving, industry manufacturing, and surveillance, where the robustness of machine learning algorithms against coding artifacts is one of the key success factors. This work complements the MPEG/JVET standardization efforts in improving the resilience of deep neural network (DNN)-based machine models against such coding artifacts by proposing the following three advanced fine-tuning procedures for their training: (1) the progressive increase of the distortion strength as the training proceeds; (2) the incorporation of a regularization term in the original loss function to minimize the distance between predictions on compressed and original content; and (3) a joint training procedure that combines the proposed two approaches. These proposals were evaluated against a conventional fine-tuning anchor on two different machine tasks and datasets: image classification on ImageNet and semantic segmentation on Cityscapes. Our joint training procedure is shown to reduce the training time in both cases and still obtain a 2.4% coding gain in image classification and 7.4% in semantic segmentation, whereas a slight increase in training time can bring up to 9.4% better coding efficiency for the segmentation. All these coding gains are obtained without any additional inference or encoding time. As these advanced fine-tuning procedures are standard-compliant, they offer the potential to have a significant impact on visual coding for machine applications.

This research investigates machine learning models for predicting mental health consequences using survey data. The study employs a two-phase approach first, it utilizes TensorFlow for initial Deep Neural Network (DNN) model building, and then it applies Random Forest (RF), Naive Bayes classifier, and decision tree methods for comparative analysis. The DNN model demonstrates strong performance, achieving high accuracy in mental health prediction. Metrics such as testing time, precision, mean absolute error, and accuracy are compared to provide insight into the advantages and disadvantages of each model. While the DNN model excels in accuracy and precision, other models offer trade-offs in computational efficiency. The results clarify the role of machine learning in mental wellness evaluation and intervention, providing guidance for further research and real-world applications. This research enhances the discourse on predictive modeling for mental health outcomes, facilitating advancements in leveraging machine learning to improve mental health assessment and intervention strategies. [ABSTRACT FROM AUTHOR]

Abstract This paper presents the crossing scheme (X-scheme) for improving the performance of deep neural network (DNN)-based music source separation (MSS) with almost no increasing calculation cost. It consists of three components: (i) multi-domain loss (MDL), (ii) bridging operation, which couples the individual instrument networks, and (iii) combination loss (CL). MDL enables the taking advantage of the frequency- and time-domain representations of audio signals. We modify the target network, i.e., the network architecture of the original DNN-based MSS, by adding bridging paths for each output instrument to share their information. MDL is then applied to the combinations of the output sources as well as each independent source; hence, we called it CL. MDL and CL can easily be applied to many DNN-based separation methods as they are merely loss functions that are only used during training and do not affect the inference step. Bridging operation does not increase the number of learnable parameters in the network. Experimental results showed that the validity of Open-Unmix (UMX), densely connected dilated DenseNet (D3Net) and convolutional time-domain audio separation network (Conv-TasNet) extended with our X-scheme, respectively called X-UMX, X-D3Net and X-Conv-TasNet, by comparing them with their original versions. We also verified the effectiveness of X-scheme in a large-scale data regime, showing its generality with respect to data size. X-UMX Large (X-UMXL), which was trained on large-scale internal data and used in our experiments, is newly available at https://github.com/asteroid-team/asteroid/tree/master/egs/musdb18/X-UMX .

Sentiment analysis (SA) or opinion mining is a general dialogue preparation chore that intends to discover sentiments behind the opinions in texts on changeable subjects. Recently, researchers in an area of SA have been considered for assessing opinions on diverse themes like commercial products, everyday social problems and so on. Twitter is a region, wherein tweets express opinions, and acquire an overall knowledge of unstructured data. This process is more time-consuming and the accuracy needs to be improved. Here, the Chronological Leader Algorithm Hierarchical Attention Network (CLA_HAN) is presented for SA of Twitter data. Firstly, the input Twitter data concerned is subjected to a data partitioning phase. The data partitioning of input Tweets are conducted by Deep Embedded Clustering (DEC). Thereafter, partitioned data is subjected to MapReduce framework, which comprises of mapper and reducer phase. In the mapper phase, Bidirectional Encoder Representations from Transformers (BERT) tokenization and feature extraction are accomplished. In the reducer phase, feature fusion is carried out by Deep Neural Network (DNN) whereas SA of Twitter data is executed utilizing a Hierarchical Attention Network (HAN). Moreover, HAN is tuned by CLA which is the integration of chronological concept with the Mutated Leader Algorithm (MLA). Furthermore, CLA_HAN acquired maximal values of f-measure, precision and recall about 90.6%, 90.7% and 90.3%. [ABSTRACT FROM AUTHOR]

Rice is a staple crop in Asia, with more than 400 million tons consumed annually worldwide. The protein content of rice is a major determinant of its unique structural, physical, and nutritional properties. Chemical analysis, a traditional method for measuring rice's protein content, demands considerable manpower, time, and costs, including preprocessing such as removing the rice husk. Therefore, of the technology is needed to rapidly and nondestructively measure the protein content of paddy rice during harvest and storage stages. In this study, the nondestructive technique for predicting the protein content of rice with husks (paddy rice) was developed using near-infrared spectroscopy and deep learning techniques. The protein content prediction model based on partial least square regression, support vector regression, and deep neural network (DNN) were developed using the near-infrared spectrum in the range of 950 to 2200 nm. 1800 spectra of the paddy rice and 1200 spectra from the brown rice were obtained, and these were used for model development and performance evaluation of the developed model. Various spectral preprocessing techniques was applied. The DNN model showed the best results among three types of rice protein content prediction models. The optimal DNN model for paddy rice was the model with first-order derivative preprocessing and the accuracy was a coefficient of determination for prediction, Rp² = 0.972 and root mean squared error for prediction, RMSEP = 0.048%. The optimal DNN model for brown rice was the model applied first-order derivative preprocessing with Rp² = 0.987 and RMSEP = 0.033%. These results demonstrate the commercial feasibility of using near-infrared spectroscopy for the non-destructive prediction of protein content in both husked rice seeds and paddy rice. [ABSTRACT FROM AUTHOR]

Summary: Cognitive Radio Ad Hoc Networks (CRAHNs) are an essential method for resolving conflicts between extreme spectrum scarcity and rapid traffic increase while maintaining high‐quality service for consumers. However, the coexistence of primary and secondary users represents a critical challenge for reasonable resource allocation in order to provide a sustaining system performance. Many approaches have been developed to efficiently allocate resources; however, these methods are currently limited by things like user collision, strange traffic networks, and high data transmission error rates. To address these constraints, this paper proposes a policy‐configured reinforcement learning‐based ad hoc network (AHN) model. To obtain the ideal policy configuration for the network, the system first models the cognitive radio (CR) network, in which nodes are initialised and grouped employing the Link Reliability K‐Means clustering Algorithm (LR‐KMA). The available spectrum was then detected and separated into multiple bands utilizing coherent‐based detection (CBD) and signal source identification employing the Parzen–Rosenblatt Window‐based Restricted Boltzmann Machine (PRW‐RBM). Next, the learning model for the resource allocation process employs the Weibull Distribution‐based Blue Monkey Optimization (WD‐BMO) approach to pick the relevant bands. Finally, the experimental results were analyzed in order to evaluate the proposed resource allocation model's performance in CRAHNs. When compared with previous findings, the proposed method improves resource utilization by 5%, the proposed model achieves a 7% higher throughput, and the PRW‐RBM's accuracy improves classification accuracy by 1.07%. [ABSTRACT FROM AUTHOR]

A deep learning (DL) approach is implemented to determine the dimensionless stress intensity factors of mode‐I (YI) and mode‐III (YIII), as well as the normalized T‐stress (T*) of the edge notch disk bend specimen. The deep neural network (DNN) method as the DL approach is used to model the relationship between the geometry parameters of the specimen a/t, S/R, and β as inputs, and YI, YIII, and T* as output variables. To this end, three datasets consisting of 176, 176, and 123 finite element method data points from the previous studies are extracted for YI, YIII, and T*, respectively. The developed DNN models predicted the YI, YIII, and T* with correlations of 0.97003, 0.96918, and 0.97047, respectively. Then partial dependence plot is performed to determine the relationship between YI, YIII, and T* and the geometry parameters, which is useful in predicting and optimizing YI, YIII, and T*. Highlights: Three datasets are prepared for the normalized geometry factors and T‐stress values.Deep neural networks are developed for the specimens with different dimensions.The sensitivity of the normalized values to the geometry parameters is assessed.The mathematical relationship between the geometric and normalized values is modeled. [ABSTRACT FROM AUTHOR]

Electromagnetic wave logging while drilling (LWD) technology is an important tool for evaluation of formation oil and gas content. It generally adopts multi-transmitter–receiver coil system structure and the symmetrical coil system arrangement with equal transmitter–receiver spacing can obtain the measurement results with borehole compensation. Here, we develop a method to realize wellbore compensation by deep learning for logging data inversion. This paper focuses on reasonable inversion of logging data through deep learning technology, which is combined with Levenberg–Marquardt (LM) algorithm, modular and fast construction of deep neural network (DNN) model. Under the condition of reducing the outermost transmitting coil, the logging data are inversed, and the inversion effect is evaluated. Our research shows that the combination of neural network and logging data can realize the measurement results with borehole compensation under the condition of reducing one transmitting coil, thereby shortening the instrument length to reduce drilling tool sticking risk and effectively reducing the LWD instrument structure complexity, high power and other problems. At the same time, the accuracy of logging data inversion is tested. The test results show that the DNN method can achieve high-precision inversion, and the average error is reduced by about 50% compared with the traditional algorithm such as linear regression. [ABSTRACT FROM AUTHOR]

Some applications, such as satellites, require ultralow power and high-radiation resilience. We developed a12Tsoft error-resilient SRAM cell, TA-Quatro, to deliver in-memory computing (IMC) for those applications. Based on our TA-Quatro cell, we implemented an IMC circuit to support binary weights and ternary activations in a single SRAM cell. Our simulation under 28 nm FD-SOI technology demonstrates that the TA-Quatro IMC circuit maintains good IMC stability at a scaled supply of 0.7Vand achieves ternary activation without needing analog-to-digital converters. These advancements significantly enhance the power efficiency of the proposed IMC circuit compared to state-of-the-art works. [ABSTRACT FROM AUTHOR]

Identifying the similarity between fine-grained images requires sophisticated techniques. This study presents a deep learning approach to the image similarity problem as an unsupervised learning task. The proposed autoencoder, built on a Deep Neural Network (DNN), autonomously learns image representations by computing cosine similarity distances between extracted features. This paper presents several applications, including training the autoencoder, transforming images, and evaluating the DNN model. In each instance, the generated images exhibit sharpness and closely resemble natural photographs, demonstrating the effectiveness and versatility of the proposed deep learning framework in computer vision tasks. The results suggest that the proposed approach is well-suited for tasks that require accurate image similarity assessments and image generation, highlighting its potential for various applications in image retrieval, data augmentation, and pattern recognition. This study contributes to the advancement of the computer vision field by providing a robust and efficient method for learning image representations and evaluating image similarity in an unsupervised manner. [ABSTRACT FROM AUTHOR]

Tropomyosin receptor kinases (TRKs) are important broad-spectrum anticancer targets. The oncogenic rearrangement of the NTRK gene disrupts the extracellular structural domain and epitopes for therapeutic antibodies, making small-molecule inhibitors essential for treating NTRK fusion-driven tumors. In this work, several algorithms were used to construct descriptor-based and nondescriptor-based models, and the models were evaluated by outer 10-fold cross-validation. To find a model with good generalization ability, the dataset was partitioned by random and cluster-splitting methods to construct in- and cross-domain models, respectively. Among the 48 models built, the model with the combination of the deep neural network (DNN) algorithm and extended connectivity fingerprints 4 (ECFP4) descriptors achieved excellent performance in both dataset divisions. The results indicate that the DNN algorithm has a strong generalization prediction ability, and the richness of features plays a vital role in predicting unknown spatial molecules. Additionally, we combined the clustering results and decision tree models of fingerprint descriptors to perform structure–activity relationship analysis. It was found that nitrogen-containing aromatic heterocyclic and benzo heterocyclic structures play a crucial role in enhancing the activity of TRK inhibitors. Workflow for generating predictive models for TRK inhibitors. [ABSTRACT FROM AUTHOR]

This paper presents the crossing scheme (X-scheme) for improving the performance of deep neural network (DNN)-based music source separation (MSS) with almost no increasing calculation cost. It consists of three components: (i) multi-domain loss (MDL), (ii) bridging operation, which couples the individual instrument networks, and (iii) combination loss (CL). MDL enables the taking advantage of the frequency- and time-domain representations of audio signals. We modify the target network, i.e., the network architecture of the original DNN-based MSS, by adding bridging paths for each output instrument to share their information. MDL is then applied to the combinations of the output sources as well as each independent source; hence, we called it CL. MDL and CL can easily be applied to many DNN-based separation methods as they are merely loss functions that are only used during training and do not affect the inference step. Bridging operation does not increase the number of learnable parameters in the network. Experimental results showed that the validity of Open-Unmix (UMX), densely connected dilated DenseNet (D3Net) and convolutional time-domain audio separation network (Conv-TasNet) extended with our X-scheme, respectively called X-UMX, X-D3Net and X-Conv-TasNet, by comparing them with their original versions. We also verified the effectiveness of X-scheme in a large-scale data regime, showing its generality with respect to data size. X-UMX Large (X-UMXL), which was trained on large-scale internal data and used in our experiments, is newly available at https://github.com/asteroid-team/asteroid/tree/master/egs/musdb18/X-UMX. [ABSTRACT FROM AUTHOR]

Dental age estimation is extensively employed in forensic medicine practice. However, the accuracy of conventional methods fails to satisfy the need for precision, particularly when estimating the age of adults. Herein, we propose an approach for age estimation utilizing orthopantomograms (OPGs). We propose a new dental dataset comprising OPGs of 27,957 individuals (16,383 females and 11,574 males), covering an age range from newborn to 93 years. The age annotations were meticulously verified using ID card details. Considering the distinct nature of dental data, we analyzed various neural network components to accurately estimate age, such as optimal network depth, convolution kernel size, multi-branch architecture, and early layer feature reuse. Building upon the exploration of distinctive characteristics, we further employed the widely recognized method to identify models for dental age prediction. Consequently, we discovered two sets of models: one exhibiting superior performance, and the other being lightweight. The proposed approaches, namely AGENet and AGE-SPOS, demonstrated remarkable superiority and effectiveness in our experimental results. The proposed models, AGENet and AGE-SPOS, showed exceptional effectiveness in our experiments. AGENet outperformed other CNN models significantly by achieving outstanding results. Compared to Inception-v4, with the mean absolute error (MAE) of 1.70 and 20.46 B FLOPs, our AGENet reduced the FLOPs by 2.7×. The lightweight model, AGE-SPOS, achieved an MAE of 1.80 years with only 0.95 B FLOPs, surpassing MobileNetV2 by 0.18 years while utilizing fewer computational operations. In summary, we employed an effective DNN searching method for forensic age estimation, and our methodology and findings hold significant implications for age estimation with oral imaging. [ABSTRACT FROM AUTHOR]

This comprehensive review explores the advancements in processing-in-memory (PIM) techniques and chiplet-based architectures for deep neural networks (DNNs). It addresses the challenges of monolithic chip architectures and highlights the benefits of chiplet-based designs in terms of scalability and flexibility. This review emphasizes dataflow-awareness, communication optimization, and thermal considerations in PIM-enabled manycore architectures. It discusses tailored dataflow requirements for different machine learning workloads and presents a heterogeneous PIM system for energy-efficient neural network training. Additionally, it explores thermally efficient dataflow-aware monolithic 3D (M3D) NoC architectures for accelerating CNN inferencing. Overall, this review provides valuable insights into the development and evaluation of chiplet and PIM architectures, emphasizing improved performance, energy efficiency, and inference accuracy in deep learning applications. [ABSTRACT FROM AUTHOR]

Employing a large dataset (at most, the order of n = 106), this study attempts enhance the literature on the comparison between regression and machine learning-based rent price prediction models by adding new empirical evidence and considering the spatial dependence of the observations. The regression-based approach incorporates the nearest neighbor Gaussian processes (NNGP) model, enabling the application of kriging to large datasets. In contrast, the machine learning-based approach utilizes typical models: extreme gradient boosting (XGBoost), random forest (RF), and deep neural network (DNN). The out-of-sample prediction accuracy of these models was compared using Japanese apartment rent data, with a varying order of sample sizes (i.e., n = 104, 105, 106). The results showed that, as the sample size increased, XGBoost and RF outperformed NNGP with higher out-of-sample prediction accuracy. XGBoost achieved the highest prediction accuracy for all sample sizes and error measures in both logarithmic and real scales and for all price bands if the distribution of rents is similar in training and test data. A comparison of several methods to account for the spatial dependence in RF showed that simply adding spatial coordinates to the explanatory variables may be sufficient. [ABSTRACT FROM AUTHOR]

Copyright of Frontiers of Information Technology & Electronic Engineering is the property of Springer Nature and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In this paper, we investigate a novel power beacon (PB)-aided wireless sensor-powered non-orthogonal multiple-access (NOMA) Internet-of-Things (IoT) network under imperfect channel state information (CSI). Furthermore, the exact expression outage probability (OP) of two IoT users is derived to analyze the performance of the considered network. To give further insight, the expression asymptotic OP and diversity order are also expressed when the transmit power at the PB goes to infinity. Furthermore, a deep neural network (DNN) framework is proposed to concurrently forecast IoT users' OP in relation to real-time setups for IoT users. Additionally, when compared to the traditional analysis, our created DNN shows the shortest run-time prediction, and the outcomes predicted by the DNN model almost match those of the simulation. In addition, numerical results validate our analysis, simulation, and prediction through a Monte Carlo Simulation. Furthermore, the results show the impact of the main parameter on our proposed system. Finally, these findings show that NOMA performs better than the conventional orthogonal multiple-access (OMA) techniques. [ABSTRACT FROM AUTHOR]

The estimation of the soil volume exceeding a contamination threshold over decommissioned industrial sites is critical for the design of remediation strategies. In practice, the volume calculation is mostly based on preliminary sampling surveys and the use of interpolation methods. However, if the volume is not estimated correctly, this can lead to environmental and economic risks. Geostatistical-oriented methodologies have been developed to better assess the volume using uncertainty ranges. In our study, we propose a methodology entitled "Evol" to better estimate the volume and reduce the uncertainty ranges with a combination of classic non-parametrical interpolation techniques and deep learning. Evol consists of generating a synthetic model from a real polluted site, extracting descriptive variables (features) from multiple sample sets, and evaluating the error in the volume calculation. A Deep Neural Network model is then trained with the features to estimate the volume and uncertainty range for any sample set. Our methodology demonstrated high accuracy in error estimation, as evidenced by a low RMSE of 0.008 across most sample sets. Additionally, the confidence volume intervals produced by our approach were narrower than those generated by classic techniques, resulting in interval size reductions of up to 89%. [ABSTRACT FROM AUTHOR]

Automated diagnosis of cancer from skin lesion data has been the focus of numerous research. Despite that it can be challenging to interpret these images because of features like colour illumination changes, variation in the sizes and forms of the lesions. To tackle these problems, the proposed model develops an ensemble of deep learning techniques for skin cancer diagnosis. Initially, skin imaging data are collected and preprocessed using resizing and anisotropic diffusion to enhance the quality of the image. Preprocessed images are fed into the Fuzzy-C-Means clustering technique to segment the region of diseases. Stacking-based ensemble deep learning approach is used for classification and the LSTM acts as a meta-classifier. Deep Neural Network (DNN) and Convolutional Neural Network (CNN) are used as input for LSTM. This segmented images are utilized to be input into the CNN, and the local binary pattern (LBP) technique is employed to extract DNN features from the segments of the image. The output from these two classifiers will be fed into the LSTM Meta classifier. This LSTM classifies the input data and predicts the skin cancer disease. The proposed approach had a greater accuracy of 97%. Hence, the developed model accurately predicts skin cancer disease. [ABSTRACT FROM AUTHOR]

This study reports a methodology for predicting surface roughness using data-driven models with grinding force as the input data. Prior to the model training process, the critical grinding parameters for brass material were selected and optimized using the Taguchi method. The experimental grinding force data were then collected and preprocessed into three features: the raw feature as the baseline feature, the statistical feature, and the fast Fourier transform (FFT) feature. The data were imported into a linear regression model as the baseline model and a deep neural network (DNN) model as the proposed strategy. The widely used surface roughness (Ra) of the ground workpiece was experimentally measured and served as the performance index. The model's performance was evaluated based on the mean absolute percentage error (MAPE) between the predicted and measured Ra values. The validation of the Ra prediction revealed that, among all test combinations, the DNN model with four hidden layers and the FFT feature as the input had the best performance of surface roughness prediction, with a MAPE of 3.17%. The independent testing and evaluation of the DNN model with the FFT feature yielded a MAPE of 6.96%, indicating that the proposed strategy effectively predicted the surface roughness of the workpiece. This work also proposes an automatic regrinding strategy in which the grinding system automatically regrinds the workpiece if the predicted Ra of the workpiece in the previous grinding process exceeds the threshold. Experimental results confirmed that among 24 ground areas, two areas have roughness exceeding the threshold and need to be regrind, and the proposed strategy can correctly identify and regrind these two areas (100% success rate). After automatic regrinding, the workpiece exhibited a roughness lower than the set threshold. [ABSTRACT FROM AUTHOR]

A key requirement for product design is the ability to capture the physical features of structures quickly and efficiently. One way to achieve this purpose is to use machine learning to support topology optimization. However, due to the diversity of application requirements in product updates, the machine learning-based optimization model needs to be frequently rebuilt by collecting large amounts of data. In this work, Moving Morphable Component approach is combined with Deep Neural Network to achieve the quick topology optimization design of acoustic devices. A deep transfer learning approach to the predictive model is also developed for the model to be adapted to different design conditions. Thus, using the proposed approach, not only the originally designed acoustic structure can be quickly optimized, but also the serialized structures evolved from the source model can be easily obtained. Numerical examples illustrate the effectiveness and advantages of the proposed method. [ABSTRACT FROM AUTHOR]

Finding the most interesting areas of an image is the aim of saliency detection. Conventional methods based on low-level features rely on biological cues like texture and color. These methods, however, have trouble with processing complicated or low-contrast images. In this paper, we introduce a deep neural network-based saliency detection method. First, using semantic segmentation, we construct a pixel-level model that gives each pixel a saliency value depending on its semantic category. Next, we create a region feature model by combining both hand-crafted and deep features, which extracts and fuses the local and global information of each superpixel region. Third, we combine the results from the previous two steps, along with the over-segmented superpixel images and the original images, to construct a multi-level feature model. We feed the model into a deep convolutional network, which generates the final saliency map by learning to integrate the macro and micro information based on the pixels and superpixels. We assess our method on five benchmark datasets and contrast it against 14 state-of-the-art saliency detection algorithms. According to the experimental results, our method performs better than the other methods in terms of F-measure, precision, recall, and runtime. Additionally, we analyze the limitations of our method and propose potential future developments. [ABSTRACT FROM AUTHOR]

To execute seamless computing along with information sharing, Cloud Computing (CC) plays an important role in measurable resource sharing. But, owing to its open and distributed framework, privacy is a major risk as it is open to attackers. To accomplish their malicious goals, cloud services are abused by intruders. In this paper, using Linear Sigmoid Singleton Deep Neural Network (LS2DNN)with Pearson correlation and Brownian motion induced K-Anonymity (PBKA)centered lightweight privacy preservation technique in the cloud server, an efficient federated learning-based Intrusion Detection System (IDS) is proposed. From the publically available sources, NSL-KDD datasets are downloaded; also, Feature Extraction (FE), null feature removal, and feature mapping are executed. Subsequently, for feature selection, Chebyshev Chaotic Mapping adapted Jaya Optimization Algorithm (C2MJOA) is wielded. Then, for classifying attacked and normal data, the LS2DNNclassifier is utilized.The incoming data from the data owners are privacy preserved via federated learning-centered Pearson correlation and Brownian motion-induced K-Anonymity (PBKA) while testing; moreover, the data is verified for intrusion. Finally, when compared with the prevailing models, the proposed model attained better results. [ABSTRACT FROM AUTHOR]

The second most prevalent and deadly disease in the world is lung cancer. It is become more difficult to detect lung cancer in its early stages in humans. A novel deep learning architecture is suggested in this study to identify lung cancer in its early stages. To distinguish between the diseased and healthy samples, the suggested deep learning structure facilitates in-depth analysis and the creation of superior feature maps. Two independent deep neural network models are combined in the proposed H-DNN (Hybrid deep neural networks) architecture. DNN-1 or SC-SDNN, the first neural network, is used to analyse spatial data, and DNN-2 or CC-FDNN, the second neural network, is used to evaluate frequency data. In order to prevent vanishing gradients, we also added the short connection in the first deep network and the chain connection in the second. Finally, to achieve a better result, we combine both neural networks. The first neural network uses spatial information from the original raw image as its training input, whereas the second neural network uses frequencies generated by wavelets. Our technique outperformed more traditional CNN and SVM classifiers, with a classification accuracy of 98.2%. [ABSTRACT FROM AUTHOR]

Aiming at the limitations of sensors for data acquisition and the inability to collect comprehensive data for the experimental system, the fast prediction modeling technology of the high flow check valve experimental system was studied. The rapid prediction model of the experimental system was established, and the results of the rapid prediction model were analyzed. First of all, the solid model was constructed and based on the structure of the experimental system of high flow check valve. At the same time, the working principle of the experimental system was combined to simplify it and carry out finite element analysis. Then, the key technology of fast prediction model was used to construct the experimental system database and realize the sample collection of the experimental system. By comparing the prediction accuracy of different machine learning algorithms, the random forest (RF) algorithm was selected to establish the relationship between the internal pressure and stress-strain of the experimental system. Finally, the results of the fast prediction model were analyzed, and the overall prediction experiment of the experimental system and the individual prediction experiment of the experimental system components were done. The results show that the fast prediction model established by the random forest algorithm has a goodness of fit (R²) of 0. 99, which is 68. 97% and 51. 47% higher compared to the deep neural network (DNN) algorithm and gradient boosted tree (GBDT) algorithm. Comparison tests between the overall prediction of the experimental system and the individual prediction of the experimental system components show that the overall prediction model has a faster prediction speed and an accuracy of 97. 43%. [ABSTRACT FROM AUTHOR]

The rapid advancement in AI requires efficient accelerators for training on edge devices, which often face challenges related to the high hardware costs of floating-point arithmetic operations. To tackle these problems, efficient floating-point formats inspired by block floating-point (BFP), such as Microsoft Floating Point (MSFP) and FlexBlock (FB), are emerging. However, they have limited dynamic range and precision for the smaller magnitude values within a block due to the shared exponent. This limits the BFP's ability to train deep neural networks (DNNs) with diverse datasets. This paper introduces the hybrid precision (HPFP) selection algorithms, designed to systematically reduce precision and implement hybrid precision strategies, thereby balancing layer-wise arithmetic operations and data path precision to address the shortcomings of traditional floating-point formats. Reducing the data bit width with HPFP allows more read/write operations from memory per cycle, thereby decreasing off-chip data access and the size of on-chip memories. Unlike traditional reduced precision formats that use BFP for calculating partial sums and accumulating those partial sums in 32-bit Floating Point (FP32), HPFP leads to significant hardware savings by performing all multiply and accumulate operations in reduced floating-point format. For evaluation, two training accelerators for the YOLOv2-Tiny model were developed, employing distinct mixed precision strategies, and their performance was benchmarked against an accelerator utilizing a conventional brain floating point of 16 bits (Bfloat16). The HPFP selection, employing 10 bits for the data path of all layers and for the arithmetic of layers requiring low precision, along with 12 bits for layers requiring higher precision, results in a 49.4% reduction in energy consumption and a 37.5% decrease in memory access. This is achieved with only a marginal mean Average Precision (mAP) degradation of 0.8% when compared to an accelerator based on Bfloat16. This comparison demonstrates that the proposed accelerator based on HPFP can be an efficient approach to designing compact and low-power accelerators without sacrificing accuracy. [ABSTRACT FROM AUTHOR]

Integrating new technology in medical diagnostics presents significant potential for improving precision and productivity. To detect oral tumor cells, this study presents a unique method that uses carbon/graphene dot (CGD) and a Customized Dual Deep Neural Network (CDDNN), known as CGD-CDDNN. Given their distinct electrical and optical characteristics, graphene quantum dots (GQDs) are used as quantum dots in quantum optics to simulate the behaviour of carbon and graphene. The proposed CDDNN architecture has two routes for processing various characteristics associated with tumor cells and is specifically designed to detect oral tumors. This customized deep-learning model's goal is to recognize complex relationships and patterns in a variety of data sources. Because the network is dual, information may be processed in parallel, and complimentary aspects can be integrated more quickly. To take advantage of the unique qualities of CGDs, quantum optics concepts are incorporated into the model, offering a quantum-level comprehension of light-matter interactions in the context of oral cancer cell identification. The CGD-CDDNN system is trained on a dataset of labelled pictures of oral cancer cells using quantum-inspired embeddings and sophisticated optimisation methods. The validation of the proposed model demonstrates its effectiveness in detecting oral tumour cells, showcasing superior performance compared to conventional methods. The study emphasises the potential of quantum optics with carbon/graphene dot models and customised dual deep neural networks in revolutionising the landscape of medical imaging and tumour diagnostics. The CGD-CDDNN framework enhances accuracy and provides valuable insights into the quantum-level interactions within biological systems, paving the way for innovative applications in precision medicine. [ABSTRACT FROM AUTHOR]

In this article, deep neural network (DNN) and chain parameter methods are used to analyze the crosstalk coupling of a star‐quad and a twisted bundle of twisted wire pairs (TBTWP) cables for different excitations, respectively. A DNN is used to extract the per‐unit‐length (p.u.l) resistance, inductance, capacitance, and conductance (RLCG) parameters at any position for the star‐quad and TBTWP cables. The induced common mode (CM) and differential mode (DM) currents due to crosstalk are calculated by combining the obtained RLCG parameters with the chain parameter method. The obtained results show good agreement with that of FEKO software. Moreover, it is found that the star‐quad cable can reduce the external coupling more effectively than that of the TBTWP cable. Finally, measurements are performed to further validate the accuracy of our proposed algorithm. [ABSTRACT FROM AUTHOR]

Purpose: Identifying unknown parameters has long been significant in the field of engineering, as they can be used for fault diagnosis and the development of numerical models. This study aims to determine the unknown parameters of mechanical models by applying deep neural networks (DNNs) using frequency–response functions (FRFs). Methods: The proposed network consists of several DNNs that estimate unknown parameters. The inputs of the DNNs requires the initial values of the unknowns. The estimated parameters by the DNNs were used to calculate the outputs. The magnitude and phase of the FRF obtained from an experiment were used as target data. To optimize the performance of the DNNs, Bayesian optimization is employed to search for appropriate weight factors in the loss function and hyperparameters in the DNNs. Additionally, the estimated parameters were adopted as the optimal parameters after the difference between the output and target data satisfied the stopping criterion. Results: The performance of the DNN-based method was confirmed for lumped and distributed parameter models with noise, which could significantly challenge the parameter identification. The proposed method showed high accuracy within 5% for both models and was validated in terms of convergence by simulations and experiments. Furthermore, we compared the proposed method with the conventional optimization methods across various noise levels and initial values. Conclusion: The proposed method successfully estimated unknown parameters under severe noise and initial conditions compared with the other methods. [ABSTRACT FROM AUTHOR]

This paper introduces a mobile cloud-based predictive model for assisting Parkinson's disease (PD) patients. PD, a chronic neurodegenerative disorder, impairs motor functions and daily tasks due to the degeneration of dopamine-producing neurons in the brain. The model utilizes smartphones to aid patients in collecting voice samples, which are then sent to a cloud service for storage and processing. A hybrid deep learning model, trained using the UCI Parkinson's Telemonitoring Voice dataset, analyzes this data to estimate the severity of PD symptoms. The model's performance is noteworthy, with accuracy, sensitivity, and specificity metrics of 96.2 %, 94.15 %, and 96.15 %, respectively. Additionally, it boasts a rapid response time of just 13 s. Results are delivered to users via smartphone alert notifications, coupled with a knowledge base feature that educates them about PD. This system provides reliable home-based assessment and monitoring of PD and enables prompt medical intervention, significantly enhancing the quality of life for patients with Parkinson's disease.

In a recent paper by Sajindra et al. [1], the soil nutrient levels, specifically nitrogen, phosphorus, and potassium, in organic cabbage cultivation were predicted using a deep learning model. This model was designed with a total of four hidden layers, excluding the input and output layers, with each hidden layer meticulously crafted to contain ten nodes. The selection of the tangent sigmoid transfer function as the optimal activation function for the dataset was based on considerations such as the coefficient of correlation, mean squared error, and the accuracy of the predicted results. Throughout this study, the objective is to justify the tangent sigmoid transfer function and provide mathematical justification for the obtained results. • This paper presents the comprehensive methodology for the development of deep neural network for predict the soil nutrient levels. • Tangent Sigmoid transfer function usage is justified in predictions. • Methodology can be adapted to any similar real-world scenarios.