Artificial Neural Network Paradigms Research Articles

Two-step interpretable modeling of ICU-AIs

We present a novel methodology for integrating high resolution longitudinal data with the dynamic prediction capabilities of survival models. The aim is two-fold: to improve the predictive power while maintaining the interpretability of the models. To go beyond the black box paradigm of artificial neural networks, we propose a parsimonious and robust semi-parametric approach (i.e., a landmarking competing risks model) that combines routinely collected low-resolution data with predictive features extracted from a convolutional neural network, that was trained on high resolution time-dependent information. We then use saliency maps to analyze and explain the extra predictive power of this model. To illustrate our methodology, we focus on healthcare-associated infections in patients admitted to an intensive care unit.

Early energy performance analysis of smart buildings by consolidated artificial neural network paradigms

The assessment of energy performance in smart buildings has emerged as a prominent area of research driven by the increasing energy consumption trends worldwide. Analyzing the attributes of buildings using optimized machine learning models has been a highly effective approach for estimating the cooling load (CL) and heating load (HL) of the buildings. In this study, an artificial neural network (ANN) is used as the basic predictor that undergoes optimization using five metaheuristic algorithms, namely coati optimization algorithm (COA), gazelle optimization algorithm (GOA), incomprehensible but intelligible-in-time logics (IbIL), osprey optimization algorithm (OOA), and sooty tern optimization algorithm (STOA) to predict the CL and HL of a residential building. The models are trained and tested via an Energy Efficiency dataset (downloaded from UCI Repository). A score-based ranking system is built upon three accuracy evaluators including mean absolute percentage error (MAPE), root mean square error (RMSE), and percentage-Pearson correlation coefficient (PPCC) to compare the prediction accuracy of the models. Referring to the results, all models demonstrated high accuracy (e.g., PPCCs >89%) for predicting both CL and HL. However, the calculated final scores of the models (43, 20, 39, 38, and 10 in HL prediction and 36, 20, 42, 42, and 10 in CL prediction for the STOA, OOA, IbIL, GOA, and COA, respectively) indicated that the GOA, IbIL, and STOA perform better than COA and OOA. Moreover, a comparison with various algorithms used in earlier literature showed that the GOA, IbIL, and STOA provide a more accurate solution. Therefore, the use of ANN optimized by these three algorithms is recommended for practical early forecast of energy performance in buildings and optimizing the design of energy systems.

Predicting the flowability of UHPC and identifying its significant influencing factors using an accurate ANN model

In this research, a one-hidden layer artificial neural network paradigm (ANN) was created to forecast the slump flow of ultra-high-performance concrete (UHPC). To achieve this goal, 3,200 ANNs were evaluated to estimate the fresh UHPC’s slump flow utilizing 793 observations. The performance metrics measured on training and test data subsets were in the same order of magnitude, thereby pointing out the proper work of the k-fold validation procedure. The results of the connection weight approach analysis (CWA) indicated that water dosage had the highest positive importance in slump flow, preceding the superplasticizer volume ratio. Other factors that positively influenced slump flow were the water-to-powder ratio, the dosage of high-alkali glass powder, the water-to-binder ratio, and limestone concentration. The most negative influences on rheology were the high-alumina FC3R and metakaolin. The ANN accurately predicted the slump flow of UHPC, while the results of the CWA analysis were well-correlated with previous research.

Neuro-Heuristic Computational Intelligence Approach for Optimization of Electro-Magneto-Hydrodynamic Influence on a Nano Viscous Fluid Flow

In this investigative study, the electro-magneto hydrodynamic (EMHD) influence on a nano viscous fluid model is scrutinized by designing an artificial neural network (ANN) paradigm using a neuro-heuristic approach (NHA) through the combination of GAs (genetic algorithms) and one of the most efficient locally searching solver SQP (sequential quadratic programming), i.e., NHA-GA-SQP. The fluid flow for the proposed problem is initially interpreted in the form of PDEs and then utilization of suitable similarity transformation on these PDEs yields in terms of a stiff nonlinear system of ODEs. The numerical results of the suggested fluidic model based on the variation of its physically existing parameters are calculated through the NHA-GA-SQP solver to detect the variation in velocity, thermal gradient, and concentration during the fluid flow. A detailed analysis of obtained outcomes through the NHA-GA-SQP algorithm and their comparison with the reference results estimated via the Adams method are presented. The calculation of the proposed solver’s accuracy, stability, and consistency through various statistical operators is also involved in the current inspection.

Generative complex networks within a dynamic memristor with intrinsic variability

Artificial neural networks (ANNs) have gained considerable momentum in the past decade. Although at first the main task of the ANN paradigm was to tune the connection weights in fixed-architecture networks, there has recently been growing interest in evolving network architectures toward the goal of creating artificial general intelligence. Lagging behind this trend, current ANN hardware struggles for a balance between flexibility and efficiency but cannot achieve both. Here, we report on a novel approach for the on-demand generation of complex networks within a single memristor where multiple virtual nodes are created by time multiplexing and the non-trivial topological features, such as small-worldness, are generated by exploiting device dynamics with intrinsic cycle-to-cycle variability. When used for reservoir computing, memristive complex networks can achieve a noticeable increase in memory capacity a and respectable performance boost compared to conventional reservoirs trivially implemented as fully connected networks. This work expands the functionality of memristors for ANN computing.

A comparative study using response surface methodology and artificial neural network towards optimized production of melanin by Aureobasidium pullulans AKW

The effect of three independent variables (i.e., tyrosine, sucrose, and incubation time) on melanin production by Aureobasidium pullulans AKW was unraveled by two distinctive approaches: response surface methodology (i.e. Box Behnken design (BBD)) and artificial neural network (ANN) in this study for the first time ever using a simple medium. Regarding BBD, sucrose and incubation intervals did impose a significant influence on the output (melanin levels), however, tyrosine did not. The validation process exhibited a high consistency of BBD and ANN paradigms with the experimental melanin production. Concerning ANN, the predicted values of melanin were highly comparable to the experimental values, with minor errors competing with BBD. Highly comparable experimental values of melanin were achieved upon using BBD (9.295 ± 0.556 g/L) and ANN (10.192 ± 0.782 g/L). ANN accurately predicted melanin production and showed more improvement in melanin production by about 9.7% higher than BBD. The purified melanin structure was verified by scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDX), X-ray diffraction pattern (XRD), and thermogravimetric analysis (TGA). The results verified the hierarchical architecture of the particles as small compasses by SEM analysis, inter-layer spacing in the XRD analysis, maximal atomic % for carbon, and oxygen atoms in the EDX analysis, and the great thermal stability in the TGA analysis of the purified melanin. Interestingly, the current novel endophytic strain was tyrosine-independent, and the uniquely applied ANN paradigm was more efficient in modeling the melanin production with appreciate amount on a simple medium in a relatively short time (168 h), suggesting additional optimization studies for further maximization of melanin production.

Imagery in the entropic associative memory

The Entropic Associative Memory is a novel declarative and distributed computational model of associative memory. The model is general, conceptually simple, and offers an alternative to models developed within the artificial neural networks paradigm. The memory uses a standard table as its medium, where the information is stored in an indeterminate form, and the entropy plays a functional and operation role. The memory register operation abstracts the input cue with the current memory content and is productive; memory recognition is performed through a logical test; and memory retrieval is constructive. The three operations can be performed in parallel using very few computing resources. In our previous work we explored the auto-associative properties of the memory and performed experiments to store, recognize and retrieve manuscript digits and letters with complete and incomplete cues, and also to recognize and learn phones, with satisfactory results. In such experiments a designated memory register was used to store all the objects of the same class, whereas in the present study we remove such restriction and use a single memory register to store all the objects in the domain. In this novel setting we explore the production of emerging objects and relations, such that cues are used not only to retrieve remembered objects, but also related and imaged objects, and to produce association chains. The present model supports the view that memory and classification are independent functions both conceptually and architecturally. The memory system can store images of the different modalities of perception and action, possibly multimodal, and offers a novel perspective on the imagery debate and computational models of declarative memory.

Application of hybrid ANN paradigms built with nature inspired meta-heuristics for modelling soil compaction parameters

For monitoring soil compaction in civil engineering projects, the determination of optimum moisture content (OMC) and maximum dry density (MDD) are very significant. The present method of determining the OMC and MDD through laboratory testing is a time-consuming task and requires skilled workmanship. Thus, in order to replace the conventional approach of OMC and MDD determination in the laboratory, this study presents a hybrid intelligence model of artificial neural network (ANN) and grey wolf optimizer (GWO). These two computational intelligence techniques collaborate to construct the ANN-GWO model for estimating OMC and MDD of soils. A database of 126 experimental results with six influencing factors was employed in the current work. To compare the outcomes of the developed ANN-GWO model, four additional hybrid ANNs were built utilizing particle swarm optimization, slime mould algorithm, Harris hawks optimization, and salp swarm algorithm. Additional laboratory experiments were conducted to investigate the generalization ability of the constructed ANNs. Results for the testing dataset indicate that the ANN-GWO achieves the best accurate estimation in cases of OMC (RMSE = 0.0986, PFI = 1.2569, and R2 = 0.7273) and MDD (RMSE = 0.1017, PFI = 1.2318, and R2 = 0.7147) of soils. The ANN-GWO model yields superior results than other hybrid ANNs constructed in this work. Overall, the suggested ANN-GWO can assist geotechnical engineers to estimate the OMC and MDD of soils in civil engineering works. The developed MATLAB models are also attached as a supplementary material which can readily be used to determine soil compaction parameters along with their ranges for different soil types.

Innovative binary sorption of Cobalt(II) and methylene blue by Sargassum latifolium using Taguchi and hybrid artificial neural network paradigms

The present investigation has been designed by Taguchi and hybrid artificial neural network (ANN) paradigms to improve and optimize the binary sorption of Cobalt(II) and methylene blue (MB) from an aqueous solution, depending on modifying physicochemical conditions to generate an appropriate constitution for a highly efficient biosorption by the alga; Sargassum latifolium. Concerning Taguchi's design, the predicted values of the two responses were comparable to actual ones. The biosorption of Cobalt(II) ions was more efficient than MB, the supreme biosorption of Cobalt(II) was verified in run L21 (93.28%), with the highest S/N ratio being 39.40. The highest biosorption of MB was reached in run L22 (74.04%), with a S/N ratio of 37.39. The R2 and adjusted R2 were in reasonable values, indicating the validity of the model. The hybrid ANN model has exclusively emerged herein to optimize the biosorption of both Cobalt(II) and MB simultaneously, therefore, the ANN model was better than the Taguchi design. The predicted values of Cobalt(II) and MB biosorption were more obedience to the ANN model. The SEM analysis of the surface of S. latifolium showed mosaic form with massive particles, as crosslinking of biomolecules of the algal surface in the presence of Cobalt(II) and MB. Viewing FTIR analysis showed active groups e.g., hydroxyl, α, β-unsaturated ester, α, β-unsaturated ketone, N–O, and aromatic amine. To the best of our knowledge, there are no reports deeming the binary sorption of Cobalt(II) and MB ions by S. latifolium during Taguchi orthogonal arrays and hybrid ANN.

Optimization of Heavy Metals Biosorption via Artificial Neural Network: A Case Study of Cobalt (II) Sorption by Pseudomonas alcaliphila NEWG-2.

The definitive screening design (DSD) and artificial neural network (ANN) were conducted for modeling the biosorption of Co(II) by Pseudomonas alcaliphila NEWG-2. Factors such as peptone, incubation time, pH, glycerol, glucose, K2HPO4, and initial cobalt had a significant effect on the biosorption process. MgSO4 was the only insignificant factor. The DSD model was invalid and could not forecast the prediction of Co(II) removal, owing to the significant lack-of-fit (P < 0.0001). Decisively, the prediction ability of ANN was accurate with a prominent response for training (R2 = 0.9779) and validation (R2 = 0.9773) and lower errors. Applying the optimal levels of the tested variables obtained by the ANN model led to 96.32 ± 2.1% of cobalt bioremoval. During the biosorption process, Fourier transform infrared spectroscopy (FTIR), energy-dispersive X-ray spectroscopy, and scanning electron microscopy confirmed the sorption of Co(II) ions by P. alcaliphila. FTIR indicated the appearance of a new stretching vibration band formed with Co(II) ions at wavenumbers of 562, 530, and 531 cm–1. The symmetric amino (NH2) binding was also formed due to Co(II) sorption. Interestingly, throughout the revision of publications so far, no attempt has been conducted to optimize the biosorption of Co(II) by P. alcaliphila via DSD or ANN paradigm.

Supervised Kohonen Self-Organizing Maps of Acute Asthma from Air Pollution Exposure.

There are unanswered questions with regards to acute respiratory outcomes, particularly asthma, due to environmental exposures. In contribution to asthma research, the current study explored a computational intelligence paradigm of artificial neural networks (ANNs) called self-organizing maps (SOM). To train the SOM, air quality data (nitrogen dioxide, sulphur dioxide and particulate matter), interpolated to geocoded addresses of asthmatics, were used with clinical data to classify asthma outcomes. Socio-demographic data such as age, gender and race were also used to perform the classification by the SOM. All pollutants and demographic traits appeared to be important for the correct classification of asthma outcomes. Age was more important: older patients were more likely to have asthma. The resultant SOM model had low quantization error. The study concluded that Kohonen self-organizing maps provide effective classification models to study asthma outcomes, particularly when using multidimensional data. SO2 was concluded to be an important pollutant that requires strict regulation, particularly where frail subpopulations such as the elderly may be at risk.

A 64K-Neuron 64M-1b-Synapse 2.64pJ/SOP Neuromorphic Chip With All Memory on Chip for Spike-Based Models in 65nm CMOS

Generally, computer-science-oriented artificial neural networks (ANNs) and neuroscience-oriented spiking neural networks (SNNs) are two main approaches to develop brain-spired non von Neumann computing systems. The goal of exploring complex artificial intelligence (AI) systems demands general neuromorphic hardware platforms compatible with both of them. However, as a result of the obvious differences in their fundamental mathematical expression and coding scheme, many neuromorphic platforms or deep neural network (DNN) accelerators accommodate only one of them. This brief presents a reconfigurable scalable neuromorphic chip based on digital leaky integrate-and-fire (LIF) neuron model targeting low-cost large-scale systems. By unifying ANN and SNN paradigms within a LIF neuron framework with point-to-point (P2P) communication, the chip can accommodate most popular neural networks. The chip adopts distributed on-chip memory architecture with a capacity of 64K neurons and 64M synapses. It achieves a peak throughput of 12.29 GSOP/s at 1.2 V, 192MHz and peak energy efficiency of 2.64 pJ/SOP at 890 mV, 24MHz. The results of implementations of a spike-based spatio-temporal memory model and ternary-weight event-based convolutional neural networks (CNNs) demonstrate outstanding compatibility of the chip.

How to Make a Class

Abstract It was not a cybernetician but a neoliberal economist who provided the first systematic treatise on connectionism or, as it would later be known, the paradigm of artificial neural networks. In his 1952 book The Sensory Order, Friedrich Hayek advanced a connectionist theory of the mind already far more advanced than the theory of symbolic artificial intelligence, whose birth is redundantly celebrated in 1956 with the exalted Dartmouth workshop. In this text Hayek provided a synthesis of Gestalt principles and considerations of artificial neural networks, even speculating about the possibility of a machine fulfilling a similar function of “the nervous system as an instrument of classification,” auguring what we call today a “classifier algorithm.” This article shows how Hayek’s connectionist theory of the mind was used to shore up a specific and ideological view of the market and schematically reconstructs Hayek’s line of argumentation from his economic paradigm backward to his theory of cognition. Eventually, in Hayek’s interpretation, connectionism provides a relativist cognitive paradigm that justifies the “methodological individualism” of neoliberalism.

A New Approach to Constructing a Decentralized Hierarchical Modular Network for Solving Complex Problems in the Paradigm of Training Artificial Neural Networks with a Teacher

A New Approach to Constructing a Decentralized Hierarchical Modular Network for Solving Complex Problems in the Paradigm of Training Artificial Neural Networks with a Teacher

An entropic associative memory

Natural memories are associative, declarative and distributed, and memory retrieval is a constructive operation. In addition, cues of objects that are not contained in the memory are rejected directly. Symbolic computing memories resemble natural memories in their declarative character, and information can be stored and recovered explicitly; however, they are reproductive rather than constructive, and lack the associative and distributed properties. Sub-symbolic memories developed within the connectionist or artificial neural networks paradigm are associative and distributed, but lack the declarative property, the capability of rejecting objects that are not included in the memory, and memory retrieval is also reproductive. In this paper we present a memory model that sustains the five properties of natural memories. We use Relational-Indeterminate Computing to model associative memory registers that hold distributed representations of individual objects. This mode of computing has an intrinsic computing entropy which measures the indeterminacy of representations. This parameter determines the operational characteristics of the memory. Associative registers are embedded in an architecture that maps concrete images expressed in modality specific buffers into abstract representations and vice versa. The framework has been used to model a visual memory holding the representations of hand-written digits. The system has been tested with a set of memory recognition and retrieval experiments with complete and severely occluded images. The results show that there is a range of entropy values, not too low and not too high, in which associative memory registers have a satisfactory performance. The experiments were implemented in a simulation using a standard computer with a GPU, but a parallel architecture may be built where the memory operations would take a very reduced number of computing steps.

Learning with ANIMA

The paper develops a semi-formal model of learning which modifies the traditional paradigm of artificial neural networks, implementing deep learning by means of a key insight borrowed from the works of Marvin Minsky: the so-called Principle of Non-Compromise. The principle provides a learning mechanism which states that conflicts in the processing of data to be integrated are a mark of unreliability or irrelevance; hence, lower-level conflicts should lead to higher-level weight-adjustments. This internal mechanism augments the external mechanism of weight adjustment by back-propagation, which is typical for the standard models of machine learning. The text is structured as follows: (§1) opens the discussion by providing an informal overview of real-world decision-making and learning; (§2) sketches a typology of decision architectures: the individualistic approach of classical decision theory, the general aggregation mechanism of social choice theory, the local aggregation mechanism of agent-based modeling, and the intermediate hierarchical model of Marvin Minsky's “Society of Mind”; (§3) sketches the general outline of ANIMA – a new model of decision-making and learning that borrows insights from Minsky's informal exposition; (§4) is the bulk of the paper; it provides a discussion of a toy exemplification of ANIMA which lets us see the Principle of Non-Compromise at work; (§5) lists some possible scenarios for the evolution of a model of this kind; (§6) is the closing section; it discusses some important differences between the way ANIMA was construed here and the typical formal rendering of learning by means of artificial neural networks and deep learning.

Face Recognition on Bag Locking Mechanism

With the emergent of biometric technology, people are no longer afraid to keep their important things in the safe box or room or even facility. This is because; human beings have unique features that distinguish them with other people. The scheme is based on an information theory approach that decomposes face images into a small set of characteristic feature images called ‘Eigenfaces’, which are actually the principal components of the initial training set of face images. In this report, thorough explanation on design process of face recognition on bags locking mechanism will be elucidated. The results and analysis of the proposed design prototype also presented and explained. The platform for executing the algorithm is on the Raspberry Pi. There are two artificial intelligent techniques applied to manipulate and processing data which is fuzzy logic and neural networks. Both systems are interdependent with each other, so that it can calculate and analyse data precisely. The receive image from the camera is analysed through the Eigenfaces algorithm. The algorithm is using Principal Component Analysis (PCA) method which comprise of artificial neural network paradigm and also statistical paradigm.

PENERAPAN METODE BACKRPOPAGATION UNTUK IDENTIFIKASI HURUF HIJAIYAH TULISAN TANGAN

Hijaiyah letters are Arabic spelling letters that are the original language of the Qur'an. Just like other types of letters, the hijaiyah has certain shapes and characteristics that will form a certain pattern. By using the concept of artificial neural networks, can dibanguun a system that can recognize the pattern by doing the previous training. One of the most commonly used meotodes in artificial neural network paradigms is the crawling or backpropagation buffer. This hijaiyah letters identification system is built using the handwritten hijaiyah image data of 150 images. The feature or feature taken from the image is the binary value of the letter pattern and the number of objects contained in the letters. Prior to the feature extraction process, the image first passes the preprocessing stage consisting of color binerization, object widening, cropping, and resizing. The result obtained by backpropagation method is the system is able to recognize handwriting hijaiyah pattern well. All training data have been correctly identified, while as many as 150 test data can be identified as 77 letters with an accuracy of 51.33%. This accuracy value is obtained with the architectural arrangement of the number of hidden layer neurons = 60, minimum error = 0.001 and maximum iteration = 10000.keyword:backpropagation, biner, hijaiyah, , pattern, preprocessing

Hybrid neural network and bi-criteria tabu-machine: comparison of new approaches to maximum clique problem

This paper presents two new approaches to solving a classical NP-hard problem of maximum clique problem (MCP), which frequently arises in the domain of information management, including design of database structures and big data processing. In our research, we are focusing on solving that problem using the paradigm of artificial neural networks. The first approach combines the artificial neuro-network paradigm and genetic programming. For boosting the convergence of the Hopfield neural network (HNN), we propose a specific design of the genetic algorithm as the selection mechanism for terms of the HNN energy function. The second approach incorporates and extends the tabu-search heuristics improving performance of network dynamics of so-called tabu machine. Introduction of a special penalty function in tabu machine facilitates better evaluation of the search space. As a result, we demonstrate the proposed approaches on well-known experimental graphs and formulate two hypotheses for further research.

Hybrid neural network and bi-criteria tabu-machine: comparison of new approaches to maximum clique problem

This paper presents two new approaches to solving a classical NP-hard problem of maximum clique problem (MCP), which frequently arises in the domain of information management, including design of database structures and big data processing. In our research, we are focusing on solving that problem using the paradigm of artificial neural networks. The first approach combines the artificial neuro-network paradigm and genetic programming. For boosting the convergence of the Hopfield neural network (HNN), we propose a specific design of the genetic algorithm as the selection mechanism for terms of the HNN energy function. The second approach incorporates and extends the tabu-search heuristics improving performance of network dynamics of so-called tabu machine. Introduction of a special penalty function in tabu machine facilitates better evaluation of the search space. As a result, we demonstrate the proposed approaches on well-known experimental graphs and formulate two hypotheses for further research.