Artificial Neural Network Topology Research Articles

Extraction of Phenolic Compounds with Antioxidant Activity from Strawberries: Modelling with Artificial Neural Networks (ANNs).

This research study focuses on the evaluation of the total phenolic compounds (TPC) and antioxidant activity (AOA) of strawberries according to different experimental extraction conditions by applying the Artificial Neural Networks (ANNs) technique. The experimental data were applied to train ANNs using feed- and cascade-forward backpropagation models with Levenberg-Marquardt (LM) and Bayesian Regulation (BR) algorithms. Three independent variables (solvent concentration, volume/mass ratio and extraction time) were used as ANN inputs, whereas the three variables of total phenolic compounds, DPPH and ABTS antioxidant activities were considered as ANN outputs. The results demonstrate that the best cascade- and feed-forward backpropagation topologies of ANNs for the prediction of total phenolic compounds and DPPH and ABTS antioxidant activity factors were the 3-9-1, 3-4-4-1 and 3-13-10-1 structures, with the training algorithms of trainlm, trainbr, trainlm and threshold functions of tansig-purelin, tansig-tansig-tansig and purelin-tansig-tansig, respectively. The best R2 values for the predication of total phenolic compounds and DPPH and ABTS antioxidant activity factors were 0.9806 (MSE = 0.0047), 0.9651 (MSE = 0.0035) and 0.9756 (MSE = 0.00286), respectively. According to the comparison of ANNs, the results showed that the cascade-forward backpropagation network showed better performance than the feed-forward backpropagation network for predicting the TPC, and the FFBP network, in predicting the DPPH and ABTS antioxidant activity factors, had more precision than the cascade-forward backpropagation network. The ANN technique is a potential method for estimating targeted total phenolic compounds and the antioxidant activity of strawberries.

Comparison between Artificial Neural Network and Rigorous Mathematical Model in Simulation of Industrial Heavy Naphtha Reforming Process

In this study, an artificial neural network (ANN) model was developed and compared with a rigorous mathematical model (RMM) to estimate the performance of an industrial heavy naphtha reforming process. The ANN model, represented by a multilayer feed forward neural network (MFFNN), had (36-10-10-10-34) topology, while the RMM involved solving 34 ordinary differential equations (ODEs) (32 mass balance, 1 heat balance and 1 momentum balance) to predict compositions, temperature, and pressure distributions within the reforming process. All computations and predictions were performed using MATLAB® software version 2015a. The ANN topology had minimum MSE when the number of hidden layers, number of neurons in the hidden layer, and the number of training epochs were 3, 10, and 100,000, respectively. Extensive error analysis between the experimental data and the predicted values were conducted using the following error functions: coefficient of determination (R2), mean absolute error (MAE), mean relative error (MRE), and mean square error (MSE). The results revealed that the ANN (R2 = 0.9403, MAE = 0.0062) simulated the industrial heavy naphtha reforming process slightly better than the rigorous mathematical model (R2 = 0.9318, MAE = 0.007). Moreover, the computational time was obviously reduced from 120 s for the RMM to 18.3 s for the ANN. However, one disadvantage of the ANN model is that it cannot be used to predict the process performance in the internal points of reactors, while the RMM predicted the internal temperatures, pressures and weight fractions very well.

Examination of artificial neural networks to predict wind-induced displacements of cable net roofs

This paper discusses the implementation of an Artificial Neural Network (ANN) model for predicting the wind-induced, vertical displacements of cable net roofs with a hyperbolic paraboloid shape. The ANN model was calibrated by training and comparing three preliminary ANN topologies. The first one was trained using both experimental pressure coefficients on similar building geometries and wind-induced vertical displacements estimated by Finite Element Method (FEM). This first set was subsequently used to calibrate a second ANN model, which was based on surrogate modeling of pressure coefficients, calculated by polynomial fitting of the experimental data. Further generalization of the pressure coefficients was obtained through parametrization of the polynomial fitting as a function of geometrical properties. Pressure loads were used to calibrate the third ANN, which was employed to synthetically generate vertical displacements of a varied geometrical sample. The ANN-based models exhibit remarkable results; the coefficient of determination between physically-estimated flexible roof displacements and approximated values was usually above 0.90 for all the tested geometries, suggesting that this type of surrogate model is capable of replicating complex, geometrically nonlinear structural behavior.

Sustainability assessment of biomethanol production via hydrothermal gasification supported by artificial neural network

Global warming and climate change urge the deployment of close carbon-neutral technologies via the synthesis of low-carbon emission fuels and materials. An efficient intermediate product of such technologies is the biomethanol produced from biomass. Microalgae based technologies offer scalable solutions for the biofixation of CO2, where the produced biomass can be transformed into value-added fuel gas mixtures by applying thermochemical processes. In this study, the environmental and economic performances of biomethanol production are examined using artificial neural networks (ANNs) for the modelling of catalytic and noncatalytic hydrothermal gasification (HTG). Levenberg-Marquardt and Bayesian Regularisation algorithms are applied to describe the thermocatalytic transformation involving various types of feedstocks (biomass and wastes) in the training process. The relationship between the elemental composition of the feedstock, HTG reaction conditions (380 °C–717 °C, 22.5 MPa–34.4 MPa, 1–30 wt% biomass-to-water ratio, 0.3 min–60.0 min residence time, up to 5.5 wt% NaOH catalyst load) and fuel gas yield & composition are determined for Chlorella vulgaris strain. The ideal ANN topology is characterised by high training performance (MSE = 5.680E-01) and accuracies (R2 ≥ 0.965) using 2 hidden layers with 17-17 neurons. The process flowsheeting of biomass-to-methanol valorisation is performed using ASPEN Plus software involving the ANN-based HTG fuel gas profiles. Cradle-to-gate life cycle assessment (LCA) is carried out to evaluate the climate change potential of biomethanol production alternatives. It is obtained that high greenhouse gas (GHG) emission reduction (−725 kg CO2,eq (t CH3OH)−1) can be achieved by enriching the HTG syngas composition with H2 using variable renewable electricity sources. The utilisation of hydrothermal gasification for the synthesis of biomethanol is found to be a favourable process alternative due to the (i) variable synthesis gas composition, (ii) heat integration, and (iii) GHG emission mitigation possibilities.

Overview and Choice of Artificial Intelligence Approaches for Night-Time Adaptive Optics Reconstruction

Adaptive optics (AO) is one of the most relevant systems for ground-based telescopes image correction. AO is characterized by demanding computational systems that must be able to quickly manage large amounts of data, trying to make all the calculations needed the closest to real-time. Furthermore, next generations of telescopes that are already being constructed will demand higher computational requirements. For these reasons, artificial neural networks (ANNs) have recently become one alternative to commonly used tomographic reconstructions based on several algorithms as the least-squares method. ANNs have shown its capacity to model complex physical systems, as well as predicting values in the case of nocturnal AO where some models have already been tested. In this research, a comparison in terms of quality of the outputs given and computational time needed is presented between three of the most common ANN topologies used nowadays, to obtain the one that fits better these AO systems requirements. Multi-layer perceptron (MLP), convolutional neural networks (CNN) and fully convolutional neural networks (FCN) are considered. The results presented determine the way forward for the development of reconstruction systems based on ANNs for future telescopes, as the ones being under construction for solar observations.

Predicting energy cost of public buildings by artificial neural networks, CART, and random forest

The paper deals with modeling the cost of energy consumed in public buildings by leveraging three machine learning methods: artificial neural networks, CART, and random forest regression trees. Energy consumption is one of the major issues in global and national policies, therefore scientific efforts in creating prediction models of energy consumption and cost are highly important. One of the largest energy consumers in every state is its public sector, consisting of educational, health, public administration, military, and other types of public buildings. Recent technologies based on sensor networks and Big data platforms enable collection of large amounts of data that could be used to analyze energy consumption and cost. A real data from Croatian public sector is used in this paper including a large number of constructional, energetic, occupational, climate and other attributes. The algorithms for data pre-processing and modeling by optimizing parameters are suggested. Three strategies were tested: (1) with all available variables, (2) with a filter-based variable selection, and (3) with a wrapper-based variable selection which integrates Boruta algorithm and random forest. Prediction models of energy cost are created using two approaches: (a) comparative usage of artificial neural networks and two types of regression trees, CART and random forest, and (b) integration of RF-Boruta variable selection and machine learning methods for prediction. A cross-validation procedure was used to optimize the artificial neural network and regression tree topology, as well to select the most appropriate activation function. Along with creating a prediction model, the aim of the paper was also to extract the relevant predictors of energy cost in public buildings which are important in planning the construction or renovation of buildings. The results have shown that the second approach which integrates machine learning with Boruta method, where the random forest algorithm is used for both variable reduction and prediction modeling, has produced a higher accuracy of prediction than the individual usage of three machine learning methods. Such findings confirm the potential of hybrid machine learning methods which are suggested in previous research, but in favor of random forest method over CART and artificial neural networks. Regarding the variable selection, the model has extracted heating and occupational data as the most important, followed by constructional, cooling, electricity, and lighting attributes. The model could be implemented in public buildings information systems and their IoT networks within the concept of smart buildings and smart cities.

Dynamic Modeling Using Artificial Neural Network of Bacillus Velezensis Broth Cross-Flow Microfiltration Enhanced by Air-Sparging and Turbulence Promoter.

Cross-flow microfiltration is a broadly accepted technique for separation of microbial biomass after the cultivation process. However, membrane fouling emerges as the main problem affecting permeate flux decline and separation process efficiency. Hydrodynamic methods, such as turbulence promoters and air sparging, were tested to improve permeate flux during microfiltration. In this study, a non-recurrent feed-forward artificial neural network (ANN) with one hidden layer was examined as a tool for microfiltration modeling using Bacillus velezensis cultivation broth as the feed mixture, while the Kenics static mixer and two-phase flow, as well as their combination, were used to improve permeate flux in microfiltration experiments. The results of this study have confirmed successful application of the ANN model for prediction of permeate flux during microfiltration of Bacillus velezensis cultivation broth with a coefficient of determination of 99.23% and absolute relative error less than 20% for over 95% of the predicted data. The optimal ANN topology was 5-13-1, trained by the Levenberg–Marquardt training algorithm and with hyperbolic sigmoid transfer function between the input and the hidden layer.

Prediction of the outlet flow temperature in a flat plate solar collector using artificial neural network.

In the current research, the efficiency of a solar flat plate collector (SFPC) was examined experimentally, while the system was modeled with an artificial neural network (ANN) under semi-arid weather conditions of Rafsanjan, Iran. Based on the backpropagation algorithm, a feedforward neural network was established to estimate and forecast the outlet flow temperature of SFPC. To identify the most appropriate model, the ANN topology hidden layer, the number of hidden neurons, iteration, and statistical indicators were analyzed. In the first ANN modeling (CASE I), five parameters, including solar radiation, inlet flow temperature, flow rate, ambient temperature, and wind speed, were applied in the input layer of the network, while the output flow temperature (subsequently efficiency) was in the output layer. In the second artificial neural network modeling (CASE II), the wind speed was omitted from the input of the ANN model. Results showed that the ANN with four inputs yields more accurate results for both estimation and prediction of outlet flow temperature.

Artificial neural network modelling for seedling regeneration in Gracilaria dura (Rhodophyta) under different physiochemical conditions

Agarophytic seaweeds have assumed prominence recently due to the development of innovative products, different marketing strategies as well as attracting new entrepreneurs and investors. Several domestic species have emerged as key players aptly supporting regional agar trade. Gracilaria dura is one such example and its commercial farming has been adopted by local Indian fisherman for diversification of their livelihood. This necessitated the adequate and continuous supply of viable seeds for sustainability of farming and subsequent processing sector. We herein reported data centric approach by adopting combined artificial neural network (ANN) model, particle swarm optimization (PSO) as well as response surface methodology (RSM) to optimize salinity, temperature, media concentration and weight to volume ratio to derive an accurate regeneration strategy in clonal seedlings. ANN topology of 4-16-1 and the combination of tangent-sigmoidal transfer function for hidden layer and linear function for output layer was found to be optimal with maximum R-value of 0.991. On employing optimized ANN model as a fitness function with PSO tool, the optimal physiochemical factors were 27 ppt salinity, 25 °C, 2.19 g L−1 DAP and 303 ml media volume. Further, the results of ANN model were experimentally validated and 33.54 ± 6.36% regeneration was observed. The prediction error in optimum regeneration rate by the ANN-PSO and RSM were 1.25% and 13.75%, respectively. The study demonstrated the efficacy of combined ANN-PSO method in solving the nonlinearity of the system. This manuscript for the first time used data centric approach by adopting artificial neuralnetworking (ANN) model and particle swarm optimization (PSO) algorithm in optimization of salinity, temperature, media concentration and weight to volume ratio to derive an accurate regeneration strategy for production of clonal seedlings in industrially important red alga Gracilaria dura. The out-come of the study would certainly help to augment and support the commercial seaweed farmers for making their livelihood lucrative and sustainable.

РЕЗУЛЬТАТЫ МНОГОФАКТОРНОЙ ОПТИМИЗАЦИЯ ПАРАМЕТРОВ ИССЛЕДУЕМЫХ ЭНЕРГОСБЕРЕГАЮЩИХ СИСТЕМ, ПРИМЕНЯЕМЫХ В ЛЕСОВОЗНОМ АВТОМОБИЛЬНОМ ТРАНСПОРТЕ

A promising direction has been substantiated that increases the efficiency of timber road transport during timber hauling, based on the development and use of various recuperative mechanisms in its design. The analysis of foreign literature has been carried out and the authors' developments and the obtained research results in this area are described. Parametric optimization of eight types of investigated recuperative energy-saving systems has been carried out on the basis of mathematical modeling. A diagram of options for optimizing the design parameters of energy-saving systems is presented, including one-factor, two-factor and multi-factor optimization. Algorithms for multifactorial optimization of design parameters and uniform distribution of a given number of points of a computer experiment in a multifactor space are presented. A mathematical apparatus for the uniform distribution of points in a multifactor space has been developed. The choice of neural network approximation for solving the problem of multifactor optimization is substantiated. An artificial neural network topology diagram is presented. An algorithm for studying the factor space using an approximating radial neural network has been developed. A method of multifactor optimization is proposed, which makes it possible to ensure a minimum of computer experiments that uniformly cover the multifactor space with highly detailed representation of response surfaces and ensuring the search for optimal regions of the factor space. Optimization of parameters of eight investigated recuperative energy-saving systems for timber road transport has been made.

Carbonation model for fly ash concrete based on artificial neural network: Development and parametric analysis

Control and prediction of the carbonation depth in reinforced concrete structures has great relevance for construction industry, since the carbonation process is directly related to the service life and durability of these structures. One challenge in carbonation modelling is to understand the complex relation between the main parameters of the phenomenon. An Artificial Neural Network (ANN) may overcome this challenge, finding solutions to these nonlinear and complex problems. In this study, an ANN with backpropagation algorithm is used in predicting the carbonation depth of concretes that contains fly ash addition. A total of 90 ANN topologies are implemented. It was observed in the training process that networks with two hidden layers are able to generate models with determination coefficient greater than 0.8. One of them is select as the one that best fit the problem. The optimized configuration provided smallest root mean square error associated with the best determination coefficient. Besides, the parametric study shown that the parameters that had most influence on the carbonation depth in fly ash-concretes were the cement consumption, fly ash content, CO2 rate and relative humidity. Besides, results indicate that the model can be applied to estimate the lifespan of concrete structures, and may be used as simulation tool in the development of engineering projects focused on durability.

Assessment of TiO2 band gap from structural parameters using artificial neural networks

Titanium dioxide is widely employed in photocatalysis reactions. Despite its variety of applications, it is still a challenge to modify it to work under visible light. Strategies have been investigated to increase the photocatalytic performance of TiO2. Usually, the aspects discussed are the ratio of crystalline phases, the crystal size, and the band gap value. However, these properties are generally empirically associated with each other. The present work was carried out to investigate the relation of those TiO2 properties on the band gap energy, also indexing its transition type. Due to the system complexity, all values were taken from the literature for artificial neural network (ANN) topologies development. The dataset was carefully selected and inspected with different types of ANN, training algorithms, transfer functions, and numbers of hidden neurons. The network was evaluated by the sum of squared error (SSE) and correlation coefficients (R2) for training and test data. The best model was 4-4-6-1 (corresponding to input, first hidden layer, second hidden layer, and output neurons, respectively). Although the network SSE (2.24) and R2 for training (84.3%) were well fitted, the R2 test could only adjust 50.7%. Therefore, questions about empirical reports of catalysis were raised and discussed, such as standards of characterization techniques of band gap measurement, and the influence of different crystalline phases in a single material.

British Sign Language Recognition via Late Fusion of Computer Vision and Leap Motion with Transfer Learning to American Sign Language.

In this work, we show that a late fusion approach to multimodality in sign language recognition improves the overall ability of the model in comparison to the singular approaches of image classification (88.14%) and Leap Motion data classification (72.73%). With a large synchronous dataset of 18 BSL gestures collected from multiple subjects, two deep neural networks are benchmarked and compared to derive a best topology for each. The Vision model is implemented by a Convolutional Neural Network and optimised Artificial Neural Network, and the Leap Motion model is implemented by an evolutionary search of Artificial Neural Network topology. Next, the two best networks are fused for synchronised processing, which results in a better overall result (94.44%) as complementary features are learnt in addition to the original task. The hypothesis is further supported by application of the three models to a set of completely unseen data where a multimodality approach achieves the best results relative to the single sensor method. When transfer learning with the weights trained via British Sign Language, all three models outperform standard random weight distribution when classifying American Sign Language (ASL), and the best model overall for ASL classification was the transfer learning multimodality approach, which scored 82.55% accuracy.

Cortico-Hippocampal Computational Modeling Using Quantum Neural Networks to Simulate Classical Conditioning Paradigms.

Most existing cortico-hippocampal computational models use different artificial neural network topologies. These conventional approaches, which simulate various biological paradigms, can get slow training and inadequate conditioned responses for two reasons: increases in the number of conditioned stimuli and in the complexity of the simulated biological paradigms in different phases. In this paper, a cortico-hippocampal computational quantum (CHCQ) model is proposed for modeling intact and lesioned systems. The CHCQ model is the first computational model that uses the quantum neural networks for simulating the biological paradigms. The model consists of two entangled quantum neural networks: an adaptive single-layer feedforward quantum neural network and an autoencoder quantum neural network. The CHCQ model adaptively updates all the weights of its quantum neural networks using quantum instar, outstar, and Widrow–Hoff learning algorithms. Our model successfully simulated several biological processes and maintained the output-conditioned responses quickly and efficiently. Moreover, the results were consistent with prior biological studies.

Artificial Neural Network Topology Optimization using K-Fold Cross Validation for Spray Drying of Coconut Milk

In this study, the development of an optimized topology neural network model for spray drying coconut milk is investigated using K-fold cross validation technique. Performance between standalone ANN and ANN with K-fold cross validation is compared, as K-fold cross validation method is integrated into neural network to overcome the limitations of restricted dataset. With inlet temperature (140 °C-180 °C), concentration of maltodextrin and sodium caseinate (0 w/w %- 10w/w %) are established as the input parameters, while moisture content (3.64%-5.1%), outlet temperature (76.5 °C-104.5 °C) and surface free fat percentage (0.35%-34.51%) are the output parameters for the neural network. Experimental data from the spray drying process is used to develop the neural network. Selection from the best training algorithm (gradient descent backpropagation, gradient descent with momentum, resilience backpropagation, conjugate gradient backpropagation with Polak-Riebre restarts, conjugate gradient backpropagation with Fletcher-Reeves, scaled conjugate gradient, Broyden-Flectcher-Goldfard-Shanno backpropagation algorithm and Levenberg-Marquardt backpropagation), transfer function (tansig, logsig, purelin and satlin), number of training runs (1000-5000), number of hidden layers (1-3) and nodes (5-15) have significant effect on the performance of the ANN models based on the lowest MSE values and R2 values. Overall, the optimum topology ANN model with k-fold cross validation outperformed the recorded lowest MSE value of 0.064 and highest R2 value of 0.855 compared to the optimum standalone ANN model with MSE value of 0.082 and R2 value of 0.832. The optimum ANN with K-fold cross validation implements the Levenberg-Marquart training algorithm with hyperbolic tangent sigmoid transfer function using 4500 times training runs with optimal topology configuration of 3-8-2-3. Result concludes that the developed neural network using K-fold cross validation represents the spray drying process as a highly reliable model with high degree of accuracy.

Lifting Activity Assessment Using Kinematic Features and Neural Networks

Work-related low-back disorders (WLBDs) can be caused by manual lifting tasks. Wearable devices used to monitor these tasks can be one possible way to assess the main risk factors for WLBDs. This study aims at analyzing the sensitivity of kinematic data to the risk level changes, and to define an instrument-based tool for risk classification by using kinematic data and artificial neural networks (ANNs). Twenty workers performed lifting tasks, designed by following the rules of the revised NIOSH lifting equation, with an increasing lifting index (LI). From the acquired kinematic data, we computed smoothness parameters together with kinetic, potential and mechanical energy. We used ANNs for mapping different set of features on LI levels to obtain an automatic risk estimation during these tasks. The results show that most of the calculated kinematic indexes are significantly affected by changes in LI and that all the lifting condition pairs can be correctly distinguished. Furthermore, using specific set of features, different topologies of ANNs can lead to a reliable classification of the biomechanical risk related to lifting tasks. In particular, the training sets and numbers of neurons in each hidden layer influence the ANNs performance, which is instead independent from the numbers of hidden layers. Reliable biomechanical risk estimation can be obtained by using training sets combining body and load kinematic features.

Optimization of artificial neural network topology for membrane bioreactor filtration using response surface methodology

<p>The optimization of artificial neural networks (ANN) topology for predicting permeate flux of palm oil mill effluent (POME) in membrane bioreactor (MBR) filtration has been investigated using response surface methodology (RSM). A radial basis function neural network (RBFNN) model, trained by gradient descent with momentum (GDM) algorithms was developed to correlate output (permeate flux) to the four exogenous input variables (airflow rate, transmembrane pressure, permeate pump and aeration pump). A second-order polynomial model was developed from training results for natural log mean square error of 50 developed ANNs to generate 3D response surfaces. The optimum ANN topology had minimum ln MSE when the number of hidden neurons, spread, momentum coefficient, learning rate and number of epochs were 16, 1.4, 0.28, 0.3 and 1852, respectively. The MSE and regression coeffcient of the ANN model were determined as 0.0022 and 0.9906 for training, 0.0052 and 0.9839 for testing and 0.0217 and 0.9707 for validation data sets. These results confirmed that combining RSM and ANN was precise for predicting permeates flux of POME on MBR system. This development may have significant potential to improve model accuracy and reduce computational time.</p>

Net-Net AutoML Selection of Artificial Neural Network Topology for Brain Connectome Prediction

Brain Connectome Networks (BCNs) are defined by brain cortex regions (nodes) interacting with others by electrophysiological co-activation (edges). The experimental prediction of new interactions in BCNs represents a difficult task due to the large number of edges and the complex connectivity patterns. Fortunately, we can use another special type of networks to achieve this goal—Artificial Neural Networks (ANNs). Thus, ANNs could use node descriptors such as Shannon Entropies (Sh) to predict node connectivity for large datasets including complex systems such as BCN. However, the training of a high number of ANNs for BCNs is a time-consuming task. In this work, we propose the use of a method to automatically determine which ANN topology is more efficient for the BCN prediction. Since a network (ANN) is used to predict the connectivity in another network (BCN), this method was entitled Net-Net AutoML. The algorithm uses Sh descriptors for pairs of nodes in BCNs and for ANN predictors of BCNs. Therefore, it is able to predict the efficiency of new ANN topologies to predict BCNs. The current study used a set of 500,470 examples from 10 different ANNs to predict node connectivity in BCNs and 20 features. After testing five Machine Learning classifiers, the best classification model to predict the ability of an ANN to evaluate node interactions in BCNs was provided by Random Forest (mean test AUROC of 0.9991 ± 0.0001, 10-fold cross-validation). Net-Net AutoML algorithms based on entropy descriptors may become a useful tool in the design of automatic expert systems to select ANN topologies for complex biological systems. The scripts and dataset for this project are available in an open GitHub repository.

Influence of random topology in artificial neural networks: A survey

Abstract Due to the fully-connected complex structure of Artificial Neural Networks (ANNs), systems based on ANN may consume much computational time, energy and space. Therefore, intense research has been recently centered on changing the topology and design of ANNs to obtain high performance. To explore the influence of network structure on ANNs complex systems topologies have been applied in these networks to have more efficient and less complex structures while they are more similar to biological systems at the same time. In this paper, the methodology and results of some recent papers are summarized and discussed in which the authors investigated the efficacy of random complex networks on the performance of Hopfield associative memory and multi-layer ANNs compared with ANNs with small-world, scale-free and regular structures.

Characterizing Computation in Artificial Neural Networks by their Diclique Covers and Forman-Ricci Curvatures

The relationships between the structural topology of artificial neural networks, their computational flow, and their performance is not well understood. Consequently, a unifying mathematical framework that describes computational performance in terms of their underlying structure does not exist. This paper makes a modest contribution to understanding the structure-computational flow relationship in artificial neural networks from the perspective of the dicliques that cover the structure of an artificial neural network and the Forman-Ricci curvature of an artificial neural network’s connections. Special diclique cover digraph representations of artificial neural networks useful for network analysis are introduced and it is shown that such covers generate semigroups that provide algebraic representations of neural network connectivity.