Neural Network Rule Research Articles

Engine and Oil Condition Analysis Using a Hybrid Supervised Model With a Multilayer Neural Network and Expert Rules

Engine and Oil Condition Analysis Using a Hybrid Supervised Model With a Multilayer Neural Network and Expert Rules

Embedded Gateway Security Detection Technology Based on the Deep Neural Network Rule Extraction

Aiming at the network security problem of power system cable trench control industrial Internet system, we studied an intrusion detection method applied to the embedded industrial Internet of Things gateway. This method extracts rules from the DBN-DNN deep neural network to obtain intrusion detection models that are conducive to integration into embedded systems. We first use the DBN network to reduce the dimensionality of the data, then use the DNN to train the classification model, and extract the rules from the DNN’s neurons to form a rule tree for intrusion detection. The KDD CUP99 training database is used to verify the feasibility of the method, and the test is carried out in the embedded gateway. The results show that the detection method based on rule extraction used in this paper can ensure detection efficiency and accuracy compared to the traditional detection methods. At the same time, it saves more computing resources and is more conducive to integration in embedded gateway systems.

Ocular disease examination of fundus images by hybriding SFCNN and rule mining algorithms

ABSTRACT Eye is the organ of vision, which is necessary for daily routine, disorders can prevent it from working effectively. Human eye can be harmed, including age, strain, poor care, and using medications without a prescription. Choroid behind the retina are affected by retinal disorders. Small and big blood vessels compressed densely and affected by light, in order to perform multi-label categories of fundus images and properly handle the issue of interleaved overflow in fundus image lesions. We proposed Single population leapfrog optimization convolutional neural network algorithm (SFCNN) and Rule mining algorithm for self-closing computer investigation of ocular disease related to retinal images, which makes the work easier for ophthalmologists to state the state of the disease. For classification hybrid techniques SFCNN and EARMAM are proposed to predict the diseased image. When comparing with existing approach, proposed approach yields an accuracy of 98.75%, a sensitivity of 97%, and specificity of 98%.

Effective Diagnosis of Lung Cancer via Various Data-Mining Techniques

One of the leading cancers for both genders worldwide is lung cancer. The occurrence of lung cancer has fully augmented since the early 19th century. In this manuscript, we have discussed various data mining techniques that have been employed for cancer diagnosis. Exposure to air pollution has been related to various adverse health effects. This work is subject to analysis of various air pollutants and associated health hazards and intends to evaluate the impact of air pollution caused by lung cancer. We have introduced data mining in lung cancer to air pollution, and our approach includes preprocessing, data mining, testing and evaluation, and knowledge discovery. Initially, we will eradicate the noise and irrelevant data, and following that, we will join the multiple informed sources into a common source. From that source, we will designate the information relevant to our investigation to be regained from that assortment. Following that, we will convert the designated data into a suitable mining process. The patterns are abstracted by utilizing a relational suggestion rule mining process. These patterns have revealed information, and this information is categorized with the help of an Auto Associative Neural Network classification method (AANN). The proposed method is compared with the existing method in various factors. In conclusion, the projected Auto associative neural network and relational suggestion rule mining methods accomplish a high accuracy status.

Retraction Note: Classification of malignant melanoma and benign skin lesion by using back propagation neural network and ABCD rule

Retraction Note: Classification of malignant melanoma and benign skin lesion by using back propagation neural network and ABCD rule

IoT enabled healthcare system for predicting the diseases using feature optimization, decision tree, neural network, and fuzzy temporal rules

SummaryDiabetes, cancer, and heart diseases are identified to be fatal for human beings due to the lack of awareness, erratic food habits, lack of physical work and so forth. Many disease prediction systems are introduced by various researchers. However, early diagnosis of the diseases and monitoring systems are not working effectively and it increases the death rate. Recently, Internet of Things (IoT) is incorporated to collect the instant data from patient and the system makes effective decisions. Many IoT enabled healthcare systems are available for predicting the specific diseases but such effective healthcare system for measuring the severity of diabetes, cancer and heart diseases is scarce. For this purpose, a new neuro‐fuzzy model is used in the proposed IoT enabled healthcare system for disease prediction, which combines fuzzy rules with temporal features. First, it categorizes the available features of the patient records. Second, a new temporal decision tree is introduced for learning the patient records carefully and predicting the record by using the available data. Third, the fuzzy temporal decision tree calculates the relationship between the available features and also selects them as contributed features. In addition, some of the features are identified as not useful for making a final decision, and these are to be removed. Fourth, the fuzzy rules are to be generated by using the Mamdani fuzzy membership function. Finally, a new framework is proposed which is the combination of neuro‐fuzzy temporal networks, fuzzy temporal decision trees, and fuzzy temporal rules. The proposed model is evaluated by conducting experiments and proved as better than the existing systems in terms of prediction accuracy, error rate, and efficiency.

Privacy-Preserving Efficient Fire Detection System for Indoor Surveillance

Residential fire is a proven hazard for human life and property. Vision based approaches for fire detection are superior to sensor based ones in terms of accuracy and alleviating false positives. Several frameworks that utilize vision-based monitoring in combination with convolutional neural network and other machine learning algorithms, such as support vector machine, K-mean clustering, logistic regression, neural network, and decision rules are available in literature for fire detection. While such frameworks are effective, they cannot be used in private spaces such as inside homes and offices as the privacy of occupants is compromised. In this article, a vision based fire detection framework for monitoring private spaces while preserving the privacy of the occupant is proposed. This is a novel endeavor as no other approach has looked at the issue of privacy preservation in fire detection with vision sensors. The framework utilizes a near infra-red camera to capture images in a manner that the privacy of occupants is preserved. To confirm that images captured with this camera do preserve occupants’ privacy, two random user surveys were conducted. For effective fire detection using these images, a novel system incorporating both spatial and temporal properties of fire is employed. Experiments were conducted and confirm the superiority of the proposed framework when compared with existing techniques in literature both in terms of performance and model size. In addition to this, the lightweight nature of the proposed system enables its effective use over resource-constrained environments as well. This is validated through a real-world prototypical implementation.

Three-Parameter Regulation Rules for the Long-Term Optimal Scheduling of Multiyear Regulating Storage Reservoirs

The perennial storage water level (PL), the water level at the end of wet season (WL), and the water level at the end of dry season (DL) are three critical water levels for multiyear regulating storage (MRS) reservoirs. Nevertheless, the three critical water levels have not been paid enough attention, and there is no general method that calculates them in light of developing regulating rules for MRS reservoirs. In order to address the issue, three-parameter regulation (TPR) rules based on the coordination between the intra- and interannual regulation effects of MRS reservoirs are presented. Specifically, a long-term optimal scheduling (LTOS) model is built for maximizing the multiyear average hydropower output (MAHO) of a multireservoir system. The TPR rules are a linear form of rule with three regulation parameters (annual, storage, and release regulation parameters), and use the cuckoo search (CS) algorithm to solve the LTOS model with three regulation parameters as the decision variables. The approach of utilizing the CS algorithm to solve the LTOS model with the WL and DL as the decision variables is abbreviated as the OPT approach. Moreover, the multiple linear regression (MLR) rules and the artificial neural network (ANN) rules are derived from the OPT approach-based water-level processes. The multireservoir system at the upstream of Yellow River (UYR) with two MRS reservoirs, Longyangxia (Long) and Liujiaxia (Liu) reservoirs, is taken as a case study, where the TPR rules are compared with the OPT approach, the MLR rules, and the ANN rules. The results show that for the UYR multireservoir system, (1) the TPR rules-based MAHO is about 0.3% (0.93 × 108 kW∙h) more than the OPT approach-based MAHO under the historical inflow condition, and the elapsed time of the TPR rules is only half of that of the OPT approach; (2) the TPR rules-based MAHO is about 0.79 × 108 kW∙h more than the MLR/ANN rules-based MAHO under the historical inflow condition, and the TPR rules can realize 0.1–0.4% MAHO more than the MLR and ANN rules when the reservoir inflow increases or reduces by 10%. According to the annual regulation parameter, the PLs of Long and Liu reservoirs are 2572.3 m and 1695.2 m, respectively. Therefore, the TPR rules are an easy-to-obtain and adaptable LTOS rule, which could reasonably and efficiently to determine the three critical water levels for MRS reservoirs.

Brain-Inspired Learning on Neuromorphic Substrates

Neuromorphic hardware strives to emulate brain-like neural networks and thus holds the promise for scalable, low-power information processing on temporal data streams. Yet, to solve real-world problems, these networks need to be trained. However, training on neuromorphic substrates creates significant challenges due to the offline character and the required nonlocal computations of gradient-based learning algorithms. This article provides a mathematical framework for the design of practical online learning algorithms for neuromorphic substrates. Specifically, we show a direct connection between real-time recurrent learning (RTRL), an online algorithm for computing gradients in conventional recurrent neural networks (RNNs), and biologically plausible learning rules for training spiking neural networks (SNNs). Furthermore, we motivate a sparse approximation based on block-diagonal Jacobians, which reduces the algorithm's computational complexity, diminishes the nonlocal information requirements, and empirically leads to good learning performance, thereby improving its applicability to neuromorphic substrates. In summary, our framework bridges the gap between synaptic plasticity and gradient-based approaches from deep learning and lays the foundations for powerful information processing on future neuromorphic hardware systems.

3D path-following control of robotic penguin: an ETFLMPC approach

Path-following (PF) control is a critical technique in various underwater scenarios. This paper investigates the 3D path-following control problem of a robotic penguin propelled by median and paired fin (MPF), in which an event-triggered fuzzy linear model predictive control (ETFLMPC) method is presented. According to the discrete linear time-varying model, the LMPC controller based on line-of-sight (LOS) guidance and linear extended state observer (LESO) is further developed. Furthermore, using the appropriate back propagation neural network (BPNN) and fuzzy rules, the thrust is estimated, while the yaw moment is transformed into a controllable offset feathering angle of the pectoral fins. Then, an event-triggered condition is designed for reducing optimization frequency with strict feasibility and stability proof. Finally, numerical simulations are performed to demonstrate the effectiveness of the proposed control algorithm. Notably, this work contributes to the constrained underactuated 3D PF control issue of an MPF-propelled robotic penguin which has been rarely explored.

Spiking Neural Networks—Part II: Detecting Spatio-Temporal Patterns

Inspired by the operation of biological brains, Spiking Neural Networks (SNNs) have the unique ability to detect information encoded in spatio-temporal patterns of spiking signals. Examples of data types requiring spatio-temporal processing include logs of time stamps, e.g., of tweets, and outputs of neural prostheses and neuromorphic sensors. In this letter, the second of a series of three review papers on SNNs, we first review models and training algorithms for the dominant approach that considers SNNs as a Recurrent Neural Network (RNN) and adapt learning rules based on backpropagation through time to the requirements of SNNs. In order to tackle the non-differentiability of the spiking mechanism, state-of-the-art solutions use surrogate gradients that approximate the threshold activation function with a differentiable function. Then, we describe an alternative approach that relies on probabilistic models for spiking neurons, allowing the derivation of local learning rules via stochastic estimates of the gradient. Finally, experiments are provided for neuromorphic data sets, yielding insights on accuracy and convergence under different SNN models.

Image Hashtag Recommendations Using a Voting Deep Neural Network and Associative Rules Mining Approach.

Hashtag-based image descriptions are a popular approach for labeling images on social media platforms. In practice, images are often described by more than one hashtag. Due the rapid development of deep neural networks specialized in image embedding and classification, it is now possible to generate those descriptions automatically. In this paper we propose a novel Voting Deep Neural Network with Associative Rules Mining (VDNN-ARM) algorithm that can be used to solve multi-label hashtag recommendation problems. VDNN-ARM is a machine learning approach that utilizes an ensemble of deep neural networks to generate image features, which are then classified to potential hashtag sets. Proposed hashtags are then filtered by a voting schema. The remaining hashtags might be included in a final recommended hashtags dataset by application of associative rules mining, which explores dependencies in certain hashtag groups. Our approach is evaluated on a HARRISON benchmark dataset as a multi-label classification problem. The highest values of our evaluation parameters, including precision, recall, and accuracy, have been obtained for VDNN-ARM with a confidence threshold 0.95. VDNN-ARM outperforms state-of-the-art algorithms, including VGG-Object + VGG-Scene precision by 17.91% as well as ensemble–FFNN (intersection) recall by 32.33% and accuracy by 27.00%. Both the dataset and all source codes we implemented for this research are available for download, and our results can be reproduced.

Machine learning based trust management framework for vehicular networks

Establishing security metrics in vehicular networking is still being debated. The dynamic characteristics of vehicular networks, imposes challenges to realize an appropriate solution to organize and ensure reliable data transfer between the vehicular nodes. In order to ensure road safety, avoid/reduce traffic congestion, and to identify malicious vehicles, an efficient Trust Management System has to be implemented in real time scenarios. All existing applications in this area have focused on reliable data exchange and authentication process of vehicular nodes to forward messages. This study proposes a new entity centric trust framework using decision tree classification and artificial neural networks. Decision tree classification model is used to derive rules for trust calculation and artificial neural networks are used to self-train the vehicular nodes, when expected trust value is not met. This model uses multifaceted role and distance based metrics like Euclidean distance to estimate the trust. The proposed entity centric trust model, uses a versatile new direct and recommended trust evaluation strategy to compute trust values. The suggested model is simple, reliable and efficient in comparison to the other popular entity centric trust models. Results and comparative analyses are carried out to prove the better performance of the proposed model over other related approaches.

Classification of Malignant Melanoma and Benign Skin Lesion with the Aid of Using Back Propagation Neural Network and ABCD Rule

Classification of Malignant Melanoma and Benign Skin Lesion with the Aid of Using Back Propagation Neural Network and ABCD Rule

Some Quantum Neural Networks

We analyzed the classical aritificial neural networks (ANN) and quantum neural networks (QNN), in particular, Boltzmann machines, Hopfield machines, the rules of training quantum neural networks. Then we analyzed the case of implementing the XOR QNN. We proposed to use the Bernstein-Vazirani Algorithm and Grover’s Search Algorithm in QNN. We looked at the problem of least square solutions of the linear regression and proposed to use the quantum Gauss-Jordan Elimination Code to solve the LSP equation. This lets us to make the network works outperform the classical approaches.

Towards explicit representation of an artificial neural network model: Comparison of two artificial neural network rule extraction approaches

In the quest for interpretable models, two versions of a neural network rule extraction algorithm were proposed and compared. The two algorithms are called the Piece-Wise Linear Artificial Neural Network (PWL-ANN) and enhanced Piece-Wise Linear Artificial Neural Network (enhanced PWL-ANN) algorithms. The PWL-ANN algorithm is a decomposition artificial neural network (ANN) rule extraction algorithm, and the enhanced PWL-ANN algorithm improves upon the PWL-ANN algorithm and extracts multiple linear regression equations from a trained ANN model by approximating the hidden sigmoid activation functions using N-piece linear equations. In doing so, the algorithm provides interpretable models from the originally trained opaque ANN models. A detailed application case study illustrates how the generated enhanced-PWL-ANN models can provide understandable IF-THEN rules about a problem domain. Comparison of the results generated by the two versions of the PWL-ANN algorithm showed that in comparison to the PWL-ANN models, the enhanced-PWL-ANN models support improved fidelities to the originally trained ANN models. The results also showed that more concise rule sets could be generated using the enhanced-PWL-ANN algorithm. If a more simplified set of rules is desired, the enhanced-PWL-ANN algorithm can be combined with the decision tree approach. Potential application of the algorithms to domains related to petroleum engineering can help enhance understanding of the problems.

Recognition of Plate Identification Numbers Using Convolution Neural Network and Character Distribution Rules

Recognition of Plate Identification Numbers Using Convolution Neural Network and Character Distribution Rules

A Process to Implement an Artificial Neural Network and Association Rules Techniques to Improve Asset Performance and Energy Efficiency

In this paper, we address the problem of asset performance monitoring, with the intention of both detecting any potential reliability problem and predicting any loss of energy consumption efficiency. This is an important concern for many industries and utilities with very intensive capitalization in very long-lasting assets. To overcome this problem, in this paper we propose an approach to combine an Artificial Neural Network (ANN) with Data Mining (DM) tools, specifically with Association Rule (AR) Mining. The combination of these two techniques can now be done using software which can handle large volumes of data (big data), but the process still needs to ensure that the required amount of data will be available during the assets’ life cycle and that its quality is acceptable. The combination of these two techniques in the proposed sequence differs from previous works found in the literature, giving researchers new options to face the problem. Practical implementation of the proposed approach may lead to novel predictive maintenance models (emerging predictive analytics) that may detect with unprecedented precision any asset’s lack of performance and help manage assets’ O&M accordingly. The approach is illustrated using specific examples where asset performance monitoring is rather complex under normal operational conditions.

A Neuro-Fuzzy Model for Online Optimal Tuning of PID Controllers in Industrial System Applications to the Mining Sector

This paper develops a model for optimal and online tuning of PID controllers and evaluates its performance. The proposed model is based on computational intelligence approaches and can be used in industrial plant operating processes. The proposed tuning model is called PID-neuro-fuzzy model, a formulation based on a structured artificial neural network and fuzzy rules. The neural network is used to perform an optimal adjustment of PID controller gains to ensure the operating point of the system required to reduce the time of accommodation and the steady-state error. A fuzzy system is incorporated into a real-time gain-scheduling scheme to compensate for the possible variations in plant parameters. The performance of the proposed model is evaluated based on its ability to deal with uncertainties and disturbances in the process. The efficiency of the model is investigated through computational simulations in five plants of two industrial processes in the mining sector: 1) solid bulk unloading processes by car dumper and 2) solid bulk resumption process by bucket wheel reclaimers. In conclusion, the main advantage of the proposed model is its adaptability in relation to variations in plant parameters during the operational process.

CONSTRUCTION OF A NEUROET NETWORK EXPERT SYSTEM FOR PROCESSING NAVIGATION DATA IN CONDITIONS RIVER е-NAVIGATION

Context. The diagnostics automatisation problem of network anomalies during navigation data processing from gauging stations to electronic chart system under river e-navigation was studied. The object of the study is a process of diagnostic in the dynamic expert system. The purpose of the research is creating of automated troubleshooting system with a help of neural system.Method. The diagnostic automatisation method of network anomalies with context-based intelligent navigation data processing usage was suggested. The main idea consists in modern data processing methods usage in the neural system. These methods are based on the fuzzy logic algorithm. Neural networks can be in operation during parameter fluctuations, that come from gauging stations. The cluster-rules set is displayed in the fuzzy neural system structure. There is no need to download all sample specifications in the electronic mapping system or to re-check the sample, it helps to speed up the process of network synthesis. Inthe navigation data processing system were used different expert systems and neural networks. Data processing system should find network anomalies and propose the ways of their decision size, the method allows to get different levels of sample specification. The method allows to minimize network error in the synthesized model. In addition to fractal method, also had been proposed a method for unknown regularities between the input and output data coming from the gauging stations. That is, the neural network can determine which signals are non-informative. With a help of input data classification from the gauging stations based on Kokhonen`s system, the space of the data stream splits into clusters of the same size and shape. By changing the cluster size, the method allows toget different levels of sample specification. The procedure of input signals classification helps to predict the increasing or decreasing of differential corrections towards depth, and recognize information from gauging stations. The method determines the encoding and decoding of navigation parameters by specifying the parametric function of the triangular shape. The rules formed by an expertknowledge were established. In order to keep intellectual system under current conditions should be used neural dynamic expert system model including use case. The mechanism of automated solution based on the search use case algorithm is defined. If there are not any use cases, the task solves with a help of neural network rules. The network nodes are neurons – particular facts that areconsequences of use case. Links between nodes of the network implement the rules. That is, a multilayer neural network of facts and rules is built up.Results. The experimental indexes of network anomalies during data processing from the gauging stations were followed up.Conclusions. The automatisation problem of network diagnostic anomalies with a help of flexible fuzzy neural network was solved. The conducted experiments confirmed the efficiency of the proposed methods. Further research may consist in the creation of an instrumental navigation method (river e-navigation).