Bio-inspired Computing: Theories and Applications

DNA strand replacement technology has the advantages of simple operation which makes it becomes a common method of DNA computing. A four bit binary number Complementer based on two-domain DNA strand displacement is proposed in this paper. It implements the function of converting binary code into complement code. Simulation experiment based on Visual DSD software is carried out. The simulation results show the correctness and feasibility of the logic model of the Complementer, and it makes useful exploration for further expanding the application of molecular logic circuit.

Avionics is confronted with transitioning from a federated avionics architecture to an Integrated Modular Avionics (IMA) architecture. IMA architectures utilize shared, configurable computing, communication, and I/O resources to increase system scalability. Therefore, resources scheduling becomes a critical issue for IMA. This paper focuses on the process scheduling. We use preemption threshold scheduling strategy to improve process scheduling performance, and propose a two-stage tabu algorithm to optimize the preemption threshold and the priority respectively. Firstly, we investigate a convergence criterion to stop iteration of level-i busy period which is used to calculate the worse-case response time. Secondly, we propose the difference analysis method based on weight to evaluate the optimal schedule. Finally, we propose TS-preemption threshold and priority optimization algorithm to obtain the near-optimal assignment of the priority and the preemption threshold. The experiment results of different sizes of process scheduling problems illustrate the validity and effectivity of the algorithm.

Self-reconfigurable robots are robots built by modules which can move in relationship to each other. This ability of changing its physical form provides the robots a high level of adaptability and robustness. Given an initial configuration and a goal configuration of the robot, the problem of self-regulation consists on finding a sequence of module moves that will reconfigure the robot from the initial configuration to the goal configuration. In this paper, we use a bio-inspired method for studying this problem which combines a cluster-flow locomotion based on cellular automata together with a decentralized local representation of the spatial geometry based on membrane computing ideas. A promising 3D software simulation and a 2D hardware experiment are also presented.

Multi-threshold segmentation is a powerful technique that is used for the processing of pattern recognition and computer vision. However, traditional, exhaustive search is computationally expensive when searching for thresholds. In order to solve such challenging problems, the fitness function is designed by the maximum entropy method, the optimal threshold of segmentation is found by using the parallel optimization mechanism of Flower Pollination algorithm (FPA), then a multi-threshold image segmentation algorithm based on FPA is proposed. The experimental results show that FPA is superior to the genetic algorithm (GA) and the shuffled frog leaping algorithm (SFLA).

As the complexity of ship arrangements increases, general arrangements optimization technology based on evolutionary algorithms has emerged, giving enormous potential to assist designers in enhancing the range of alternative arrangements and in expediting the design process. This paper presents a hybrid evolutionary algorithm to handle the multi-objective constrained arrangements optimization problem based on elitist non-dominated sorting strategy. To enhance the efficiency of optimization, a hybrid evolutionary algorithm that couples an NSGA-II with a stochastic local search technique is used to find feasible solutions rapidly and facilitate local optimization. However, the algorithm that can rapidly find feasible solutions is also expected to contribute to better optimization. It has also been observed that lack of diversity of potential solutions leads to a local optimal solution which means the coherent arrangements could not be discovered. Hence, a modified replacement strategy is proposed to overcome this drawback. The final experimental results illustrate that the algorithm is capable of generating coherent arrangements.

In order to improve the feature selection stability based on evolutionary algorithms, an evolutionary algorithms’ feature selection stability improvement system is proposed. Three Filter methods’ results are aggregated to provide the stability information, and feature selection stability and classification accuracy are adopted as two optimization objectives. Weighted sum, weighted product and biobjective optimization methods together are applied as the system’s optimization models. Ant colony optimization, particle swarm optimization and genetic algorithm are used as testing algorithms, and experiments are taken on two benchmark datasets. The results show that the proposed system can improve the stability of evolutionary algorithms’ feature selection efficiently and their classification performance simultaneously.

A multi-objective hybrid particle swarm algorithm is proposed to solve the problem that the current unmanned surface vessel (USV) global routing algorithm is easy to fall into the local optimal solution and the optimization target is single. The snap jump feature of simulated annealing algorithm is used to improve global search capability of particle swarm algorithm, and the three objective functions of path length, path smoothness and path security are used to optimize the path. The simulation result shows that the algorithm can improve the smoothness of the inflection point and the security of the path on the shortest path.

Predicting essential proteins is indispensable for understanding the minimal requirements of cellular survival and development. In recent years, many methods combined with the topological features of PPI networks have been proposed. However, most of these approaches ignored the intrinsic characteristics of biological attributes. This paper integrates Gene expression data, Subcellular localization and PPI networks to identify essential proteins, named GSP. We use local average connectivity and edge clustering coefficient unite with gene expression data to measure centralities of nodes. Compared with non-essential proteins, essential proteins appear more frequently in some subcellular localizations such as Nucleus and considering that different compartments play different roles, thus we integrate subcellular localization information to identify essential proteins. The computational experiment results on the yeast PPI networks show that the proposed method GSP outperforms other state-of-art methods including DC, EC, IC, SC, NC, LAC, PeC, WDC and UDoNC.

Polarimetric Synthetic Aperture Radar (PolSAR) has been meeting the requirements in acquiring images for all-day, free of light, weather and other reasons, so it is widely applied in military and civilian life. PolSAR images contain abundant information. Its processing and interpretation have played more and more important role in national defense construction and economic development. However, the classification accuracy for PolSAR images using conventional clustering-based methods is quite limited. In this paper, a novel semi-supervised classification method is proposed. The Simulated Annealing-Genetic Algorithm (SAGA) is designed to optimize the iterative mechanism for finding the optimal centers of Fuzzy C-means (FCM) clustering, which avoids the local optimum. This leads to more accurate divisions on each category. Experimental results on synthesized and real PolSAR images confirm the superior performance of the proposed algorithm compared with conventional methods.

This paper is an attempt to relax the condition of using the rules in a maximally parallel manner in the framework of spiking neural P systems with exhaustive use of rules. To this aim, we consider the minimal parallelism of using rules: if one rule associated with a neuron can be used, then the rule must be used at least once (but we do not care how many times). In this framework, we study the computational power of our systems as number generating devices. Weak as it might look, this minimal parallelism still leads to universality, even when we eliminate the delay between firing and spiking and the forgetting rules at the same time.

Deoxyribonucleic acid (DNA) computing provides a novel way of breaking through the limitations of traditional computation framework. Some complicated computational problems on small-scale have been solved. Model checking is a notable verification technique which is important to security-critical system. We employ DNA computing models and propose DNA algorithms for checking four elementary formulas of computation tree logic with past-time constructs in this paper. The model checking algorithms based on DNA computing are proved to be practicable and valid by simulations. The time complexity of the algorithms is reduced to linearity while the classical algorithm is PSPACE-complete. It indicates that a complexity computational problem is solved on DNA-computing based and the problems which can be solved by DNA computing are enriched. Meanwhile, it could be a benefit to diagnosis and treatment of genetic diseases at molecular level.

The performance of genetic algorithm (GA) critically depends on the rates of variation operation. In this paper, we propose a hybrid parameter adaptation scheme, which integrates the traditional adaptive and self-adaptive method, to dynamically control the crossover and mutation rate of GA during evolution. Such a scheme can take advantage of both adaptive and self-adaptive mechanisms, thus effectively setting the parameters of GA. The resulting GA has been applied for data clustering. Our results show that the proposed scheme is beneficial and the resulting GA outperforms the adaptive GA or self-adaptive GA for data clustering.

Population size in evolutionary algorithms (EAs) is critical for their performance. In this paper, we first give a comprehensive review of existing population control methods. Then, a few representative methods are selected and empirically compared on a range of well-known benchmark functions to show their pros and cons.

Ant colony algorithm is a kind of bionic evolutionary algorithm, which is widely used in the field of optimization. Membrane computing is a new computing model, which has the characteristics of distributed, maximal parallelism and non-deterministic. Different with the most current researches that use ant colony algorithm as the sub-algorithm in the framework of the membrane algorithm, this paper considers the realizing ant colony algorithm completely by evolution rules, and we design new ant colony P system $$\varPi _{ACS}$$ΠACS, which includes the membrane structure and evolutionary rules. This paper not only provides a new way to realize the ant colony algorithm, but also lays a foundation for building a general framework for solving optimization problems in membrane computing.

In this paper, a new SN P system is investigated in order to compute the product of any two decimal natural numbers. Firstly, an SN P system with two input neurons is constructed, which can be used to compute the product of any two binary natural numbers which have specified lengths. Secondly, the correctness of the SN P system is proved theoretically. However, the system can only be used to compute the product of any two binary natural numbers, but the product of any two decimal natural numbers often need to be computed in practical application. Therefore, it is necessary to construct a coding SN P system which converts a decimal number into a binary number and to construct a decoding SN P system which converts a binary number to a decimal number. In the end, an new SN P system is constructed to compute the product of any two decimal natural numbers. An example test shows that the SN P system can be used to compute the product of any two decimal natural numbers. Therefore, this paper provides a new method for constructing the SN P system which can compute the product of any two natural numbers.

Standard particle swarm optimization 2011(SPSO2011, takes SPSO for short) was proposed to overcome problems that there is bias of the search area existing in the conventional PSO depending on rotational invariant property. The performance of SPSO is affected by the distribution of the center of the search range and the global search ability fades away during the iteration process. In this paper, in order to reinforce diversity-maintain ability as well as improve local search ability, a modified diversity-guided SPSO (DGAP-MSPSO) algorithm is proposed. A modified SPSO variant with average point method is first applied till the swarm loses its diversity thus to improve local search ability. Then, the search process turns to another new SPSO variant in which an enhanced diversity-maintain operator is used for global search. The DGAP-MSPSO switches alternately between two SPSO variants according to swarm diversity, thus its search ability is improved. Experimental results shows that our proposed algorithm, the DGAP-MSPSO algorithm, gets better performance on most test functions compared with other SPSO variants.

Chinese short text has become the main content of the Internet. Accurately extracting thematic terms is the basis of content analysis, query suggestions, document classification, and text clustering and other tasks for Chinese short text on the Internet. Since Chinese short text is short on the Internet, unbalanced and less of context information, the traditional text clustering model is not immediately appropriate. This paper presents a simple and generic theme model named Lexicon LDA for Chinese short text on the Internet, by using the sentence structure within the document, to enrich the context of the common Chinese word semantics. Words of each sentence which is divided by punctuation marks compose a word set. Unlike the previous method, the model distributes the theme for each word set, rather than for each document. When the data set presents a strong theme distribution, it can significantly improve the effect of the theme model through experiments. The conclusion is that extracting thematic terms of Chinese short text on the Internet is related both to the word itself and to the sentence where the word is located.

In the computational biology, DNA strand displacement technique is used to construct the logic gate model and molecular circuits. But with the increase of number of the reaction strands, the basic logic gates cannot meet the accuracy in the reaction process. This paper improved the logical unit OR gate, AND gate to solve this problem on the basic of the seesaw model and the mechanism of DNA strand displacement reaction. Since the basic circuit of the decoder is an array of AND gates, the improvement module is applied to the decoder. The molecular circuit of decoder is constructed to realize the dynamic link between the input signal and the output signal. It is concluded that the sensitivity and accuracy of the improved decoder in the molecular circuit is improved by the Visual DSD software. The improvement module laid the foundation for the development of molecular circuits.

In order to realize the optimization of ladle tracking of aluminium tapping, a mathematical model, which takes the grade of aluminium, the energy required for transportation and the optimum ratio of aluminium liquid into account, is established. The traditional method optimizes the impurity content and the transport distance based on a single objective optimization, but it requires the empirical values of the weight coefficients. The paper proposes a modified multi-objective optimization model with the elitist non-dominated sorting genetic algorithm (NSGA-II). In the crossover operator process and the mutation operator process, the separately improved methods are introduced based on the ladle tracking problem of aluminium tapping, which replaces the Simulated Binary Crossover (SBX) in the original NSGA-II algorithm into Partially Matched Crossover (PMX) based on natural number coding and uses exchange mutation (EM) operator. Finally, the practical production data of the aluminium factory is used to verify the validity and practicability of the method, and the results show that this method can obtain a feasible solution for the user to choose suitable solutions, and avoid the defects of selecting empirical weighting coefficient.

Recently, infectious diseases, such as Ebola fever and Middle East respiratory syndrome, have spread worldwide. To conduct a highly accurate infection screening, our group is working on the development of a non-contact and hand-held infection screening system that can detect infected individuals within 5 s. In this study, we propose a signal processing method to improve the measurement accuracy of the infection screening system. Body surface temperature, heartbeat, and respiration rates are detected by thermography and microwave radars. To evaluate the measurement accuracy, nine subjects (normal and pseudo-infection conditions) were tested with the proposed system in a laboratory. In this study, a linear discriminate function was used to detect pseudo-infection conditions. The detection accuracy was improved to 88.9%.

As a key part of aluminium smelting, the operational conditions of aluminium electrolytic cells are of great significance for the stability of the aluminium electrolysis process. As a result, developing a effective process monitoring and multiple fault diagnosis model is essential. Traditional multi-classification methods such as neural networks and multiple support vector machines (multi-SVM) have good effects. However, the connatural limitations of these methods limit the prediction accuracies. To solve this problem, a hierarchical method for multiple fault diagnosis based on kernel principal component analysis (KPCA) and support vector machines (SVM) is proposed in this paper. Firstly, test statistics, such as the comprehensive index $$ \phi $$ϕ, the squared prediction error (SPE), and Hotellings T-squared ($$ T^{2} $$T2), are used for fault detection. To separate faults preliminarily, traditional K-means clustering as transition layer is applied to the principal component scores. Next, anode effect is recognized and classified by the established SVM prediction model. Compared with multi-SVM-based classification methods, the proposed hierarchical method can diagnosis different faults with a higher precision. The prediction accuracy can reach about 90%.

Aiming at the deficiency of the low sensitivity of DNA encoding and chaotic encryption algorithms to text and key, and the limited encoding rules of DNA, etc. This paper presents a new image encryption algorithm based on DNA dynamic encoding and hyper-chaotic system. Firstly, the algorithm uses the SHA-3 algorithm to process the original image, generate a set of hash values, perform the dynamic encoding of the generated hash values and then carry out XOR operation with the original image, and then the generated hash values through Hamming distance processing to generate the initial value of the hyper-chaotic system. Secondly, the S-box is constructed by the sequence values generated by the hyper-chaotic system, and the XOR-shift manipulation is performed to the image by using the S-box. Finally, the image is scrambled by the hyper-chaotic Chen System. The simulation results and theoretical analysis show that the algorithm improves the sensitivity of key and the security of data transmission, and has better ability of anti-exhaustive attack, statistical attack and differential attack.

Through extensive application of DNA strand displacement technology in the field of molecular computing, we know that the DNA strand of the toehold domain and the branch migration domain are covalently connected to form logical gates in traditional DNA strand displacement circuits. In this paper, we will adopt a composite strand mechanism that the toehold domain and branch migration domain in different single strand form displacing complex, and construct logical gates with combinatorial strand displacement mechanism. After that, a logic gate model is constructed, and the mechanism is verified by the design and simulation of the logical molecular model of the encoder. When the DNA signal strand is input, the signal strand molecule can be output by combination of molecular specific hybridization reaction and intermolecular strand displacement reaction. The results of Visual DSD simulation show the feasibility and accuracy of the encoder logic calculation model designed in this article.

With the application of biosensors in environmental monitoring, these features of low sample concentrations and the need for real-time monitoring feedback in environmental monitoring, make the sensor requirements also increasing. High sensitivity, short response time and low cost are the environmental monitoring biological sensors goal. RNA has a high affinity capacity and sensitivity, and has better thermal stability after hybridization. Combined with the characteristics of nano-gold and magnetic particles in this paper, improving material and probe of electrode, propose design ideas of several biosensors with LNA, PNA for RNA viruses in water monitoring to improve biosensors in environmental monitoring of practicality.

This paper proposes distributed fuzzy P systems with promoters for multi-microgrids power balance, where the distributed P systems (dP systems) differ from other P systems with the ability to handle distributed input problems, which makes themselves more suitable for solving control problems. To make full use of the advantages of dP systems and provide a research idea for the power balance of multi-microgrids, promoters and fuzzy theory are introduced into dP systems to characterize a large amount of uncertain and inaccurate information. Moreover, the proposed distributed fuzzy P systems with promoters are applied to fulfill the power balance of multi-microgrids. Finally, the power balance in multi-microgrids as well as the balance between the multi-microgrids and their connecting grid is realized by three cases.

Cell-like P systems with symport/antiport rules (CSA P systems, for short) are a class of computational models in membrane computing, inspired by the way of transmembrane transport of substances through membrane channels between neighboring regions in a cell. In this work, we propose a variant of CSA P systems called cell-like P systems with symport/antiport rules and promoters (CSAp P systems, for short), where symport/antiport rules are regulated by multisets of promoters, and the computation power of CSAp P systems is investigated. Specifically, it is proved that CSAp P systems working in the maximally parallel mode, having any large number of membranes and promoters and using only symport rules of length 1 or antiport rules of length 2, are able to compute only finite sets of non-negative integers. Furthermore, we show that CSAp P systems with two membranes working in a sequential mode when having at most two promoters and using only symport rules of length 2, or having at most one promoter and using symport rules of length 1 and antiport rules of length 2, are Turing universal.

Chaotic circuit is an effective tool to observe and analyze chaotic phenomena, to verify chaos theory and to promote its application. The recent research work focuses on how to better analyze the basic circuit characteristics and to design application circuits. In this paper, a new chaotic system is proposed, whose dynamical behaviors are discussed with the change of the parameters in detail. The specific effects of different parameters on the system are also discussed. By adjusting these parameters of the proposed circuit, this nonlinear circuit can produce the different dynamical behaviors, such as, hyper chaotic behavior, periodic behavior, transient behavior, etc. In addition, the simulation results of Matlab can further prove the feasibility of this circuit.

DNA strand displacement acted as a useful tool is most widely used in the majority computing system. In this paper, a four-input fire alarm system device based on DNA strand displacement is designed. The whole reaction course is programmed and simulated in the software visual DSD, which presenting the superb simulation results with inputs and outputs through compiling the procedure for computation device. According to the results of the Visual DSD software, the method of DNA strand displacement by dual-rail circuits is feasible to achieve more complex logic computation. This investigation on the basis of DNA strand displacement by dual-rail circuits may have a great prospect for the development and application in the biological information processing, molecular computing, and so on.

This paper proposes a method for fault location in distribution networks with distributed generation based on an improved spiking neural P system with anti-spikes (IASNP system). In the IASNP system, firing mechanism, fuzzy logic, new types of neurons and a matrix algorithm are introduced. The IASNP system is used to model the distribution networks while its matrix algorithm locates faults by considering the causality between regions and the associated nodes. Finally, two cases, including a multi source distribution network and distribution network with distributed generation, are used to verify the validity and accuracy of the proposed method.

In recent years, DNA strand displacement technology has played a significant role in DNA computing. In this paper, a five-bit square-root digital logic circuit based on DNA strand displacement is designed by using a simple DNA reaction mechanism. The whole reaction process of logic circuit operations can be programmed and simulated through using the Visual DSD simulation software. According to the simulation results, the square-root logic circuit is feasible to achieve the desired logical computations. Through analyzing the simulation results, the feasibility of the designed circuit is demonstrated. And it is proves that DNA strand displacement may have a great potential and bright prospect in the construction of large-scale logic circuits.

In recent years, DNA strand displacement technology has been become an integral part of DNA computing, which is proved that the complement circuit is played an important role in computer circuits. In this paper, a four-bit complement logic circuit based on DNA strand displacement is designed and simulated. Through the analysis about the simulation results, which is proved that the designed circuit is reliable, and the four-bit complement logic circuit based on DNA strand displacement design is also shown that the DNA strand displacement has a bright future in the construction of large-scale logic circuits.

Extreme learning machine (ELM), which proposed for generalized single-hidden layer feedforward neural networks, has become a popular research topic due to its fast learning speed, good generalization ability, and ease of implementation. However, ELM faces redundancy and randomness in the hidden layer which caused by random mapping of features. In ELM, although evolutionary algorithms have archived impressive improvement, they have not considered the sparsity of the hidden layers. In this paper, a hybrid learning algorithm is proposed, termed EMO-ELM, which adopts evolutionary multi-objective algorithm to optimise two conflict objectives simultaneously. Furthermore, the proposed method can be used for supervised classification and unsupervised sparse feature extraction tasks. Simulations on many UCI datasets have demonstrated that EMO-ELM generally outperforms the original ELM algorithm as well as several ELM variants in classification tasks, moreover, EMO-ELM achieves a competitive performance to PCA in sparse feature extraction tasks.

In this paper, a single molecular multi-functional DNA cascade logic circuit and signal amplifier model was demonstrated by two single molecular multi-functional ions DNA probe(SMIP) to detect environment mercury and silver ion pollution, these two SMIP random coil structures turned into different hairpin-like structures with T-Hg$$^{2+}$$2+-T or C-Ag$$^{+}$$+-C via inputting mercury and silver ions, then, use the SMIP structure “OR” and “AND” logic gate and unimolecular mulfunctional DNA logic amplifier model (UMDA). Finally, we proved the feasibility of our model by PAGE and fluorescence alteration.

Recommender systems have received much attention due to their wide applications. Current recommender approaches typically recommend items to user based on the rating prediction. However, the predicted ratings cannot truly reflect users interests on items because the rating prediction is usually based on history data and does not consider the effect of time factor on uses interests (behaviors). In this paper, we propose a recommendation approach combining the matrix factorization and a recurrent neural network. In this approach, all the items rated by a user are considered as time series data. The matrix factorization is used to obtain latent vectors of those items. The recurrent neural network is taken as a time series prediction model and trained by the latent vectors of historical items, and then the trained model is used to predict the latent vector of the item to be recommended. Finally, a recommendation list is formed by mapping the latent vector into a set of items. Experimental results show that the proposed approach is able to produce an effective recommend list and outperforms those comparative approaches.

The multi-level approach is known to be a highly effective metaheuristic framework for tackling several types of combinatorial optimization problems, which is one of the best performing approaches for the graph partitioning problems. In this paper, we integrate the multi-level approach into the hypervolume-based multi-objective local search algorithm, in order to solve the bi-criteria max-cut problem. The experimental results indicate that the proposed algorithm is very competitive, and the performance analysis sheds lights on the ways to further improvements.

DNA computing is a hot research topic in recent years, molecular logic gate is an important foundation of DNA computer architecture and implementation. Local hairpin DNA chain substitution reaction can increase the reliability of molecular logic gates, Make the reaction more efficient and more thoroughly. In this paper, using local hairpin DNA strand displacement, the comparator circuit is coded and simulated base on the double logic circuit. The simulation results further confirmed the feasibility and effectiveness of the DNA strand displacement reaction in the study of biochemical logic circuits, The comparator circuit can be used for biological computer and building large-scale molecular logic circuits in the future.

With the increasing complexity of real-world optimization problems, many challenges appear to evolutionary algorithms (EAs). When solving these time-consuming or high-complexity problems, although EAs can guarantee the high quality of solutions, the intolerable time costs will influence their availabilities drastically. Thus, many attempts have been made to overcome that problem. With the rapid development of the distributed computing paradigm and platforms, such as the Message Passing Interface (MPI) and Open Multi-Processing (OpenMP), distributed computing has become readily available and affordable for realizing more powerful EAs. In order to find out whether these platforms have any particular difficulties or preference, whether one of them would be more suitable for EAs, we analyze the performance of different distributed EAs (DEAs) based on different distributed computing platforms, using differential evolution (DE) as an example. Finally, we find out that both MPI and OpenMP have their own superiorities and they can improve the speedup obviously. However, MPI is more suitable for computationally expensive problems and can achieve higher speedup than OpenMP.

Labelled splicing P systems are distributed parallel computing models, where sets of strings that evolve by splicing rules are labelled. In this work, we consider labelled splicing systems with the following modifications: (i) The strings in the membranes are present in arbitrary number of copies; (ii) the rules in the regions are finite in number. Results on the language family generated by the labelled splicing system in comparison with the language families of the Chomsky hierarchy, including recursively enumerable languages, are obtained, by involving only either one or two membranes in the P systems considered.

Artificial neural networks are a computational system, and usually, backpropagation algorithm is used for learning a task, because of its simplicity. However, backpropagation algorithm is likely to converge to a local minimum or saddle point, so that a global minimum may not be found. Differential Evolution (DE) is a simple yet powerful global optimization algorithm for solving multi-dimensional continuous functions. In this paper, we propose a new DE algorithm by combining two excellent DE algorithms, Adaptive Cauchy DE (ACDE) and Self-adaptive DE (SaDE). ACDE shows promising performance by using the Cauchy distribution based on control parameter adaptation. However, ACDE uses only one mutation strategy. SaDE adapts mutation strategies automatically, which shows its effectiveness. Therefore, we extend ACDE with the strategy adaptation of SaDE for enhancing the global optimization performance. The result indicates that the extended ACDE performs better than standard DE not only on conventional benchmark problems but also for training neural networks.

Deep belief network (DBN) has become one of the most popular techniques for short-term load forecasting. The transfer functions play a vital role on the effective of DBN. In this study, different combinations of three commonly used transfer functions, i.e., logsig, purelin and tansig, in a DBN are examined. Experimental results show that a combination of purelin and tansig transfer functions produces the best load forecasting, and is therefore recommended to use.

Cloud service resource allocation is an essential task in cloud computing. The cloud service resource allocation problem is modeled as an optimization problem, and is solved via different particle swarm optimization (PSO) variants in this paper. The aim of our method is to minimize the delay and the price at the same time. Based on the experimental results, it could be conducted that the good performance could be achieved via PSO algorithms. The future research is to utilize PSO algorithms on solving more real-world problems, especially with other quality of service problems.

In order to improve the accuracy of synthetic aperture radar (SAR) image change detection, a novel unsupervised non-parametric method for change detection is described. The method treats the prior data and the observed data as two independent events and adopts a simple algorithm to realize and validate the effectiveness of the proposed method. Firstly, the prior distribution is obtained by MRF with Potts model of the initial classification result by k-means with the prior data. Secondly, the fuzzy probability is obtained through fusing gray value and texture feature fuzzy membership of the observed data. Meanwhile, the fuzzy probability is regarded as the data likelihood probability. Finally, by using the Bayesian formula and the independent distribution criteria to calculate the maximum a posteriori (MAP) probability, change detection can be regarded as the product of two probabilities of two independent events. Simulation results show that the proposed method effectively combines the gray and texture information of difference image, overcomes the shortcomings of using probability statistic model and parameter estimation, reduces the influence of speckle noise of SAR image and improves the accuracy of image change detection.

We realize the Hilbert-Mumford stability of products of Grassmannians with fixed total dimension base on Mumford’s computation in [8] in the computer algebra system Maxima. The problem is reduced to be a discrete algorithm and by some techniques of strings in Maxima, the coding is universal and effective. This code can prove the classical results for point sets and lines in projective spaces in Geometric Invariant Theory.

This paper presents a predictive control approach based on ant colony optimization algorithm for critical maneuvering of unmanned surface vehicle in high sea environment. The algorithm uses the generalized predictive control to get the predictive course value. In the process of the algorithm, the ant colony algorithm is used to obtain the optimal control sequence of the rudder angle. The obtained simulation results show that the algorithm solves the problem of overshoot of course controller, and realizes the precise control of USV course in the case of large disturbance of wind and wave, then solves the saturation nonlinear problem of unmanned surface vessel in extreme sea condition.

Multipath signal is often considered an interference that must be removed. The coherence between multipath and direct component makes it difficult to use conventional direction-of-arrival (DOA) estimation methods in a smart antenna system. This study demonstrates a new multipath signal DOA estimation technique of the L-shaped array. The proposed algorithm first converts the two-dimensional DOA estimation to the DOA estimation of uniform linear array, and apply the independent component analysis algorithm to obtain the steering vectors with multipath information. Then, based on the special structure of the obtained steering vectors and spatial sparsity of the multipath signals, the algorithm uses the solution of the sparse signal reconstruction problem in the compressive sensing theory, and search the space spectrums to acquire the synthesis angles for each direct component and multipath component. Finally according to the geometric relationship to obtained the azimuth and elevation angles. Comparative simulation tests and analysis prove the effectiveness of the proposed algorithm in estimation accuracy.

One-class collaborative filtering with implicit feedback has attracted much attention, mainly due to the widespread of implicit data in real world. Pairwise methods have been shown to be the state-of-the-art methods for one-class collaborative filtering, but the assumption that users prefer observed items to unobserved items may not always hold. Besides, existing pairwise methods may not perform well in terms of Top-N recommendation. In this paper, we propose a new approach called EBPR, which relaxes the former simple pairwise preference assumption by further exploiting the hidden connection in observed items and unobserved items. EBPR can also be used as a basic method and has the extensive applicability, i.e., when combining our model with former pairwise methods, better performance can also be achieved. Empirical studies show that our algorithm outperforms the state-of-the-art methods on four real-world datasets.

In this paper, the optimal brain surgeon (OBS) strategy is introduced to improve the iterative rule of non-negative matrix factorization (NMF) algorithm for intrusion detection, which is called OBS-NMF algorithm. A new convergence condition and criterion function are proposed to improve the performance of the OBS-NMF algorithm. Then the proposed method is applied in the HIDS and NIDS, the experimental results show that our method can obtain higher accuracy and better stability than the NMF algorithm, and achieves satisfying detection performance. The improved OBS-NMF algorithm is also suitable for real-time intrusion detection.

In this paper, the immune mechanism is used to solve the problem of fuzzy target extraction in infrared thermal trace images. First, imitating the innate immunity, the maximum variance method was used to initially segment the image to obtain the decided target and background region and the undecided fuzzy region of the target. Then, inspired by the antigen-presenting mechanism, this paper constructed the feature set for each pixel in the image, which contained the gray information, the temperature information with their statistical values. Next, in order to express the characteristic changes of the antigen invading normal cells, the paper adopted the local preserving algorithm to do the mapping to obtain new feature states. Finally, in view of effector T-cell mechanism in adaptive immunity, the paper used the new feature states to measure the distances between fuzzy region and the target and background region, so as to decide the classification of the pixels of the fuzzy region. Through this process, the extraction of the fuzzy target in the thermal trace image is completed.

Object motion blur results when the object in the scene moves during the recording of a single exposure, either due to too rapid movement or long exposure, leaving streaks of the moving object in the image and thus degrading its quality. In this paper, we present a method to solve the object motion blur problem in images with clear static background. Specifically, we propose an object motion deblurring algorithm that uses a convolutional neural network with six convolutional layers to deblur the image. Taking advantages of the strong ability of feature learning in convolutional neural networks, our method can remove the blurring effect of fast-moving object while keeping the clear background untouched. It is well known that neural networks are best driven by large data sets and more data means more benefits for training convolutional neural networks; therefore, we generated training set of 144,000 images and test set of 32,400 images. Through carefully designed training process, our model learned the ability of deblurring the blurred object while keeping the clear background. The experiment results show that our approach can generate superior results to a representative image deblurring algorithm that treats the same blurred object and clear background.