Over the past two decades, the feedforward neural network (FNN) optimization has been a key interest among the researchers and practitioners of multiple disciplines. The FNN optimization is often viewed from the various perspectives: the optimization of weights, network architecture, activation nodes, learning parameters, learning environment, etc. Researchers adopted such different viewpoints mainly to improve the FNN's generalization ability. The gradient-descent algorithm such as backpropagation has been widely applied to optimize the FNNs. Its success is evident from the FNN's application to numerous real-world problems. However, due to the limitations of the gradient-based optimization methods, the metaheuristic algorithms including the evolutionary algorithms, swarm intelligence, etc., are still being widely explored by the researchers aiming to obtain generalized FNN for a given problem. This article attempts to summarize a broad spectrum of FNN optimization methodologies including conventional and metaheuristic approaches. This article also tries to connect various research directions emerged out of the FNN optimization practices, such as evolving neural network (NN), cooperative coevolution NN, complex-valued NN, deep learning, extreme learning machine, quantum NN, etc. Additionally, it provides interesting research challenges for future research to cope-up with the present information processing era.
Deep learning is an important component of big-data analytic tools and intelligent applications, such as, self-driving cars, computer vision, speech recognition, or precision medicine. However, the training process is computationally intensive, and often requires a large amount of time if performed sequentially. Modern parallel computing systems provide the capability to reduce the required training time of deep neural networks. In this paper, we present our parallelization scheme for training convolutional neural networks (CNN) named Controlled Hogwild with Arbitrary Order of Synchronization (CHAOS). Major features of CHAOS include the support for thread and vector parallelism, non-instant updates of weight parameters during back-propagation without a significant delay, and implicit synchronization in arbitrary order. CHAOS is tailored for parallel computing systems that are accelerated with the Intel Xeon Phi. We evaluate our parallelization approach empirically using measurement techniques and performance modeling for various numbers of threads and CNN architectures. Experimental results for the MNIST dataset of handwritten digits using the total number of threads on the Xeon Phi show speedups of up to 103x compared to the execution on one thread of the Xeon Phi, 14x compared to the sequential execution on Intel Xeon E5, and 58x compared to the sequential execution on Intel Core i5.
In this paper with the aid of genetic algorithm and fuzzy theory, we present a hybrid job scheduling approach, which considers the load balancing of the system and reduces total execution time and execution cost. We try to modify the standard Genetic algorithm and to reduce the iteration of creating population with the aid of fuzzy theory. The main goal of this research is to assign the jobs to the resources with considering the VM MIPS and length of jobs. The new algorithm assigns the jobs to the resources with considering the job length and resources capacities. We evaluate the performance of our approach with some famous cloud scheduling models. The results of the experiments show the efficiency of the proposed approach in term of execution time, execution cost and average Degree of Imbalance (DI).
While being it extremely important, many Exploratory Data Analysis (EDA) systems have the inhability to perform classification and visualization in a continuous basis or to self-organize new data-items into the older ones (evenmore into new labels if necessary), which can be crucial in KDD - Knowledge Discovery, Retrieval and Data Mining Systems (interactive and online forms of Web Applications are just one example). This disadvantge is also present in more recent approaches using Self-Organizing Maps. On the present work, and exploiting past sucesses in recently proposed Stigmergic Ant Systems a robust online classifier is presented, which produces class decisions on a continuous stream data, allowing for continuous mappings. Results show that increasingly better results are achieved, as demonstraded by other authors in different areas. KEYWORDS: Swarm Intelligence, Ant Systems, Stigmergy, Data-Mining, Exploratory Data Analysis, Image Retrieval, Continuous Classification.
The rapid e-commerce growth has made both business community and customers face a new situation. Due to intense competition on one hand and the customer's option to choose from several alternatives business community has realized the necessity of intelligent marketing strategies and relationship management. Web usage mining attempts to discover useful knowledge from the secondary data obtained from the interactions of the users with the Web. Web usage mining has become very critical for effective Web site management, creating adaptive Web sites, business and support services, personalization, network traffic flow analysis and so on. The study of ant colonies behavior and their self-organizing capabilities is of interest to knowledge retrieval/management and decision support systems sciences, because it provides models of distributed adaptive organization, which are useful to solve difficult optimization, classification, and distributed control problems, among others. In this paper, we propose an ant clustering algorithm to discover Web usage patterns (data clusters) and a linear genetic programming approach to analyze the visitor trends. Empirical results clearly shows that ant colony clustering performs well when compared to a self-organizing map (for clustering Web usage patterns) even though the performance accuracy is not that efficient when comparared to evolutionary-fuzzy clustering (i-miner) approach. KEYWORDS: Web Usage Mining, Swarm Intelligence, Ant Systems, Stigmergy, Data-Mining, Linear Genetic Programming.
Security of computers and the networks that connect them is increasingly becoming of great significance. Computer security is defined as the protection of computing systems against threats to confidentiality, integrity, and availability. There are two types of intruders: the external intruders who are unauthorized users of the machines they attack, and internal intruders, who have permission to access the system with some restrictions. Due to the fact that it is more and more improbable to a system administrator to recognize and manually intervene to stop an attack, there is an increasing recognition that ID systems should have a lot to earn on following its basic principles on the behavior of complex natural systems, namely in what refers to self-organization, allowing for a real distributed and collective perception of this phenomena. With that aim in mind, the present work presents a self-organized ant colony based intrusion detection system (ANTIDS) to detect intrusions in a network infrastructure. The performance is compared among conventional soft computing paradigms like Decision Trees, Support Vector Machines and Linear Genetic Programming to model fast, online and efficient intrusion detection systems.
Decision-making is a process of choosing among alternative courses of action for solving complicated problems where multi-criteria objectives are involved. The past few years have witnessed a growing recognition of Soft Computing (SC) technologies that underlie the conception, design and utilization of intelligent systems. In this paper, we present different SC paradigms involving an artificial neural network trained using the scaled conjugate gradient algorithm, two different fuzzy inference methods optimised using neural network learning/evolutionary algorithms and regression trees for developing intelligent decision support systems. We demonstrate the efficiency of the different algorithms by developing a decision support system for a Tactical Air Combat Environment (TACE). Some empirical comparisons between the different algorithms are also provided.
This paper presents a comparative study of six soft computing models namely multilayer perceptron networks, Elman recurrent neural network, radial basis function network, Hopfield model, fuzzy inference system and hybrid fuzzy neural network for the hourly electricity demand forecast of Czech Republic. The soft computing models were trained and tested using the actual hourly load data for seven years. A comparison of the proposed techniques is presented for predicting 2 day ahead demands for electricity. Simulation results indicate that hybrid fuzzy neural network and radial basis function networks are the best candidates for the analysis and forecasting of electricity demand.
The costs of fatalities and injuries due to traffic accident have a great impact on society. This paper presents our research to model the severity of injury resulting from traffic accidents using artificial neural networks and decision trees. We have applied them to an actual data set obtained from the National Automotive Sampling System (NASS) General Estimates System (GES). Experiment results reveal that in all the cases the decision tree outperforms the neural network. Our research analysis also shows that the three most important factors in fatal injury are: driver's seat belt usage, light condition of the roadway, and driver's alcohol usage.
The academic literature suggests that the extent of exporting by multinational corporation subsidiaries (MCS) depends on their product manufactured, resources, tax protection, customers and markets, involvement strategy, financial independence and suppliers' relationship with a multinational corporation (MNC). The aim of this paper is to model the complex export pattern behaviour using a Takagi-Sugeno fuzzy inference system in order to determine the actual volume of MCS export output (sales exported). The proposed fuzzy inference system is optimised by using neural network learning and evolutionary computation. Empirical results clearly show that the proposed approach could model the export behaviour reasonable well compared to a direct neural network approach.