Optimal operation of an industrial blast furnace (BF) ironmaking process largely depends on a reliable measurement molten iron quality (MIQ) indices, which are not feasible using the conventional sensors. This paper proposes novel data-driven robust modeling method for online estimation and control MIQ indices. First, nonlinear autoregressive exogenous (NARX) model is constructed indices to completely capture dynamics BF process. Then, considering that standard least-squares support vector...

BIRCH algorithm is a hierarchical clustering method which suitable for very large datasets especially. The traditional uses distance to control the shape of clusters. However, results effect non-spherical dataset not good, and in some cases clusters are divided into different In order break through such limitation, this paper presents an improved based on Link, draws link concept ROCK algorithm. Experiments show that can cluster any

CHAMELEON clustering algorithm uses dynamic modeling to find high quality clusters of different shapes, sizes and densities. When the constructs k-nearest neighbor graph G, k value is difficult select has a great influence on result. Using multi-level partitioned hMetic for segmenting large hyper graphs divide Gk coarse partition, which easily leads uncertainty results. Aiming at this shortcoming, paper proposes new condensed hierarchical method based AGNES algorithm, convenient efficient...

In a traffic system, drivers will always choose paths to maximize their own travel utilities, while in the other side, system manager endeavors minimize overall time and reach most efficiency. this paper, an induction mechanism based on game theory is proposed. According real data, publishes corrected information initially, then paths. Both of choosing are fulfilled before enter into which can avoid congestion drift intrinsically. Behaviors guided by theory. While achieving Nash equilibrium,...

We propose a novel algorithm based on clustering to extract rules from artificial neural networks. After networks Beijing trained and pruned successfully, inner-rules are generated by discrete activation values of hidden units. Then, weights between input units clustered decrease the complexity extraction. In phase, number can be adjusted dynamically according their corresponding The experimental results demonstrate that this is effective.

Complex networks are a useful tool to study WSNs. In this paper, we introduce the concept of overpayment from complex and propose an optimal path routing algorithm solve random problem after reaching equilibrium. Finally, present simulation our algorithm.

K-prototypes is an algorithm that deals with mixed data clustering. However, the clustering parameter k needs to be manually set and initial centroids are randomly selected, therefore, it can lead blindness of cluster number as well problems low accuracy unstable results. Regarding issue above, this paper proposes a strategy determine adaptively based on density distance, which clusters choose better than original algorithm. According experiments UCI datasets, superior traditional...

The combination of complex networks and game theory is one the most suitable ways to describe evolutionary laws various systems. In order explore evolution group cooperation in multiple social dilemmas, a model with double-layer network proposed here. Firstly, simulate multiplayer under identities, we combine public goods game. Secondly, make an individual’s strategy selection process more line practical context, new learning method that incorporates individual attributes designed here,...

This paper mainly introduces the coordination evolutionary game on complex network models. The structure and each individual strategy evolution are studied. Cooperative relationship of can be researched preferably by studying average payoff cooperation rate. graph strategies describes change among nodes. has changed nodes appear aggregation phenomenon comparing model parameters. coevolution.

According to the sociological principium, this paper designed a model aimed at social networks and implemented it using multi-agent system. Based on model, we established simulation system research evolution of colony in networks, analyzed effects by characteristics individuals achieved meaningful conclusions.

In traditional researches of evolutionary game theory, each individual takes a pure strategy or mixed which are both priori and fixed. Yet we find that smart-strategy, is based on individual's memory makes decision according to opponent's historical actions information, the evolutionarily stable (ESS) lose original standard stability. Specially, use some basic thoughts subliminal channel for reference then put forward innovative viewpoints how design smart-strategy. Through investigating...

ERA is the acronym of environment, reactive rules and agent; model a new effective multi-agent cooperation framework, which has been successfully applied in wide range areas. Multi-port Container Stowage (MPCS) more practical problem. However, multi-port makes container stowage problem difficult. This paper presents MPCS based on time series belief to solve The algorithm introduces concept into agents creates model. searching direction strength are determined by learning machine using...

Abstract Extracting text information from raw images has always been a hot and difficult problem in computer vision research. Due to blurred image, uneven illumination complex backgrounds, etc., the recognition of natural scene digit is achieve desired results. In this paper, an encoder-decoder neural network model proposed solve indefinite length sequences scene. The encoder convolutional (CNN), decoder long short-term memory (LSTM). accepts fixed-format image outputs fixed-length feature...

This paper proposes a modified version of opinion communication based on naming game to simulate the formation in real society. In version, freshness is brought measure memory level one for each agent, when various opinions exist social population. Main researches are devoted new replacing old after system already convergences an opinion, revealing that not easy govern whole network. Playing gaming BA network, we explore validate personal ability accept information, people number accepting...