Deep Artificial Neural Network Architectures Research Articles

A PLS – SEM study to test the role of Social media in influencing Purchase Intention

In light of the growing importance of social networking marketing (SMM) to the profitability of tiny and medium-sized businesses (SMEs) and the relatively modest adoption rate of SMM among SMEs, this study seeks to determine which factors influence SMEs’ adoption of SMM. This study, unlike the majority of others, proposed a two-stage analysis combining the partial least squares (PLS) method with an artificial intelligence technique called an artificial neural network (ANN). Using a deep ANN architecture, the proposed model can make predictions with a 91% success rate. The marketing rate of social networking site adoption was found to be significantly affected by the strength of the relationship between perceived efficiency, user approval of use, perceived expenses, and encouragement from upper management, beneficial conditions, and vendor pressure. The findings of this study contribute to the expanding body of literature on online advertising by shedding light on the role played by technological, corporate, and ecological (TOE) variables in consumers’ adoption of social media promotional activities. Investment choices in digital marketing in comparable and non-competitive industries can benefit from the study’s findings, which can be used by policymakers as well as managers of SMM and consumer behavior.

Abstract 544: Evaluating The Diagnostic Ability Of Six Different Artificial Neural Networks From The Subcellular Microenvironment To The Clinical Manifestation

Peripheral arterial disease (PAD) is affecting a large number of patients in the USA and worldwide. The natural history of the disease leads to detrimental clinical implications including walking disability, pain and eventually limb loss. The clinical management of the disease is largely based on clinical criteria (e.g. pain, tissue loss). We have prepared muscle biopsy specimens from patients with PAD versus control subjects and utilizing electron-microscopy we captured in the subcellular level the mitochondrial architecture changes. Using the acquired data as input we tested the ability of six structurally different artificial neural networks (ANNs) to identify PAD patients. For the study we acquired human tissue biopsies from the gastrocnemius muscle. The specimens were processed and cut as ultrathin sections (70 nm) for analysis using a JEM1010 transmission electron microscope. The images were used to generate an image preprocessing database of 992 images and subsequently a custom python-based code was developed with Keras - TensorFlow platforms in order to perform the ANN analysis. We implementing six different deep artificial neural network architectures namely, VGG-19 , ResNet50 , Densenet201 , InceptionV3 , InceptionsResnetV2 and EfficientNetB7.The diagnostic accuracy was found to be respectively: VGG-19 0.81%, ResNet50 0.75%, Densenet201 0.81%, InceptionV3 0.79%, InceptionsResnetV2 0.78% and EfficientNetB7 0.87%.This study presents data for the ability of simple and more advanced ANNs to perform diagnostic predictions based on electron microscopy images of intracellular organelles. The ability to take advantage of applied artificial intelligence technologies and serve this information for clinical decision making could be a pathway for improved patient care. The presented work established insight in some of the currently available deep artificial neural networks, applied in a daily clinical problem and their ability to harness biological information. Our goal is to stimulate the development of higher computational capacity artificial intelligence systems in order to assist clinicians and support improved guidelines geared towards individualized precision medicine.

A conceptual framework for determining metaverse adoption in higher institutions of gulf area: An empirical study using hybrid SEM-ANN approach

The metaverse is a kind of imagined world with immersive digital spaces that increase, allowing a more interactive environment in educational settings. The metaverse is an expansion of the synchronous communication that embraces an effective number of users to share different experiences. The study aims to investigate the students' perceptions towards metaverse system for educational purposes in the Gulf area. The conceptual model comprises the adoption properties, namely trialability, observability, compatibility, and complexity, users' satisfaction, personal innovativeness, and Technology Acceptance Model (TAM) constructs. The novelty of the paper lies in its conceptual model that correlates both personal-based characteristics and technology-based features. In addition, the novel approach of hybrid analysis will be used in the current study to perform deep-learning-based analysis of structural equation modelling (SEM) and artificial neural network (ANN). Moreover, the importance-performance map analysis (IPMA) is used in the current study to evaluate the involved factors for their importance and performance. The study identified Perceived Usefulness (PU) to be an essential predictor of the factor of Users’ Intention to Use the Metaverse System (MS). The fact was discovered during ANN and IPMA analysis. Furthermore, this study is practically significant, as it helped the concerned authorities in educational sector in understanding the significance of each factor and allowed them to make efforts and plans according to the order of significance of factors. Another important implication of the study is methodological in nature. It validates that deep ANN architecture can offer deep insight into non-linear relationships shared by various factors of a theoretical model.

Determinants of intention to use medical smartwatch-based dual-stage SEM-ANN analysis

The current study is based on an integrated research model developed by combining constructs from the Technology Acceptance Model (TAM) and other features affecting smartwatch effectiveness, such as content richness and user satisfaction (SAT). TAM is used to locate factors influencing the adoption of the smartwatch (ASW). Most importantly, the current study focuses on factors influencing smartwatch acceptance and use in the medical area, facilitating and enhancing the effective role of doctors and patients. The present study's conceptual framework examines the close association between two-term TAM variables of perceived ease of use (PEU) and perceived usefulness (PU) and the constructs of user satisfaction and content richness. It also incorporates the flow theory (EXP) to measure the effectiveness of the smartwatch. The study also uses the flow theory to assess involvement and control over ASW. The study used a sample of 489 respondents from the medical field, including doctors, nurses, and patients. The study employed a hybrid analysis method combining Structural Equation Modeling (SEM) and an Artificial Neural Network (ANN) based on deep learning. The study also used Importance-Performance Map Analysis (IPMA) to determine the relevance and performance of the variables influencing ASW. Based on the ANN and IPMA analyses, user satisfaction is the most crucial predictor of intention to use a medical smartwatch. Applying the structural equation model to the sample shows that SAT, PU, PEU, and EXP significantly influence intention to use a medical smartwatch. The study also revealed that content richness is an important factor that enhances users' PU. The current study could enable healthcare provider practitioners and decision-makers to identify factors for prioritisation and to strategise their policies accordingly. Methodologically, this study indicates that a “deep ANN architecture” can determine the non-linear associations between variables in the theoretical model. Overall, the study finds that smartwatches are in high demand in the medical field and are useful in information transmission between doctors and their patients.

A deep artificial neural network architecture for mesh free solutions of nonlinear boundary value problems

Seeking efficient solutions to nonlinear boundary value problems is a crucial challenge in the mathematical modelling of many physical phenomena. A well-known example of this is solving the Biharmonic equation relating to numerous problems in fluid and solid mechanics. One must note that, in general, it is challenging to solve such boundary value problems due to the higher-order partial derivatives in the differential operators. An artificial neural network is thought to be an intelligent system that learns by example. Therefore, a well-posed mathematical problem can be solved using such a system. This paper describes a mesh free method based on a suitably crafted deep neural network architecture to solve a class of well-posed nonlinear boundary value problems. We show how a suitable deep neural network architecture can be constructed and trained to satisfy the associated differential operators and the boundary conditions of the nonlinear problem. To show the accuracy of our method, we have tested the solutions arising from our method against known solutions of selected boundary value problems, e.g., comparison of the solution of Biharmonic equation arising from our convolutional neural network subject to the chosen boundary conditions with the corresponding analytical/numerical solutions. Furthermore, we demonstrate the accuracy, efficiency, and applicability of our method by solving the well known thin plate problem and the Navier-Stokes equation.

In-depth comparison of deep artificial neural network architectures on seismic events classification

One of the most challenging tasks in volcanic data analysis is the classification of seismic events. By knowing them, it is possible to take decisions in advance, providing benefits for the neighboring societies as, for instance, how such events may impact users' life and cropland areas. Although there are several approaches to perform such task, Deep Neural Networks (DNN) have been barely considered to deal with seismic signals, as discussed in our related work. In this sense, we started our research with a wide set of experiments to analyze the DNN performance while discriminating seismic activities through two common network architectures: a 2D Convolutional Neural Networks (CNN) and the Long Short-Term Memory (LSTM) network. In attempt to draw a parallel of DNN and classical supervised learning strategies, we extended our study by comparing those two architectures against the Multilayer Perceptron (MLP), which would be the simplest and most common baseline to take into account. Our MLP network was implemented after combining a basic architecture with elements from DNN models as dropout, flatten, and batch normalization layers. From this first analysis, we confirmed the need of additional neural network layers to obtain good classification results for seismic events, given the MLP required more DNN-based pipeline operations to improve its overall performance, making it comparable to the previous results obtained by our research group. As a natural extension, we designed a new DNN architecture by assessing additional network layers specifically devoted to extract features from seismic signals, thus improving the overall classification of Volcano activities involving the following events: Volcano-Tectonic (VT), Long Period (LP), Tremor (TR) and Tectonic (TC). Such design motivated the study of current DNN architectures tackling similar problems to classify raw signals from which SoundNet was taken as the most prominent candidate, thus leading to a new CNN architecture here proposed and referred to as SeismicNet. As our final contribution, SeismicNet provided classification results among the best in the literature without demanding explicit signal pre-processing steps though.

Multi-Channel Deep Feature Learning for Intrusion Detection

Networks had an increasing impact on modern life since network cybersecurity has become an important research field. Several machine learning techniques have been developed to build network intrusion detection systems for correctly detecting unforeseen cyber-attacks at the network-level. For example, deep artificial neural network architectures have recently achieved state-of-the-art results. In this paper a novel deep neural network architecture is defined, in order to learn flexible and effective intrusion detection models, by combining an unsupervised stage for multi-channel feature learning with a supervised one exploiting feature dependencies on cross channels. The aim is to investigate whether class-specific features of the network flows could be learned and added to the original ones in order to increase the model accuracy. In particular, in the unsupervised stage, two autoencoders are separately learned on normal and attack flows, respectively. As the top layer in the decoder of these autoencoders reconstructs samples in the same space as the input one, they could be used to define two new feature vectors allowing the representation of each network flow as a multi-channel sample. In the supervised stage, a multi-channel parametric convolution is adopted, in order to learn the effect of each channel on the others. In particular, as the samples belong to two different distributions (normal and attack flows), the samples labelled as normal should be more similar to the representation reconstructed with the normal autoencoder than that of the attack one, and viceversa. This expected dependency will be exploited to better disentangle the differences between normal and attack flows. The proposed neural network architecture leads to better predictive accuracy when compared to competitive intrusion detection architectures on three benchmark datasets.