Search

Extreme learning machine (ELM) is a new learning algorithm for the single hidden layer feedforward neural networks. Compared with the conventional neural network learning algorithm it overcomes the slow training speed and over-fitting problems. ELM is based on empirical risk minimization theory and its learning process needs only a single iteration. The algorithm avoids multiple iterations and local minimization. It has been used in various fields and applications because of better generalization ability, robustness, and controllability and fast learning rate. In this paper, we make a review of ELM latest research progress about the algorithms, theory and applications. It first analyzes the theory and the algorithm ideas of ELM, then tracking describes the latest progress of ELM in recent years, including the model and specific applications of ELM, finally points out the research and development prospects of ELM in the future.

In order to reduce the influence of illumination changes on the appearance of smart meter and improve the accuracy of smart meter appearance detection, based on the local two value algorithm theory, the algorithm is improved as the local two value algorithm of the rotation invariance. But rotation invariant local value operator with the sampling number of blocks increases the operator types will be a substantial increase, this limitation is bound to increase computing influence quantity calculation speed. Therefore, it will be done in a unified process, and the unified local two value algorithm is obtained. In the light of the smart meter in strong lighting and weak lighting condition, separately use the local two value algorithm theory and the unified local two value algorithm are separately used to detect the appearance of the smart meter. The analysis results show that the unified local two value algorithm can recognize the information of the smart meter, which has a certain application value for the smart meter.

Courses on algorithm analysis often include little programming, and do not emphasize application of advanced techniques to practical problems. On the other hand, students usually prefer hands-on learning, and may lack motivation to study theory. We augmented an algorithms course with a series of programming tasks, which involved application of the course material to image processing. These tasks motivated the students, and led to better understanding and retention of theoretical material. They also enabled the students to learn the basics of representing and manipulating images, along with the algorithm theory.

Obtaining a predictive understanding of cell behavior will require new computational methods, as well as adaptations of traditional techniques of optimization and system analysis. The authors discuss the challenges cell biologists face in their effort to control information transfer in cells.

This paper presents a qualitative theory for faults diagnosis. We extend Reiter's [1] theory to deal with the dynamic and continuous systems and offer a necessary assumption and the corresponding propositions. Then we propose an algorithm of detection of faults and extend it to the multi-faults case. As there is not enough information about multi-faults, multi-faults diagnosis is usually a partially observable problem. The STRIPS [2], a classic technique of automated planning, is chosen to build the system model in which the cause-effect information required to process the diagnosis has been integrated. It provides the reasoning ability for the multi-faults diagnosis when diagnosis is formalized as reasoning from effects to causes with causal knowledge. The advantages are that this approach allows building the model gradually with the information received instead of fully in one time and provides a corresponding diagnostic search strategy that has the powerful reasoning ability. Keywords-qualitative; multi-faults; diagnosis; cause-effect

This paper aims to evaluate the learning model for coordination schemes in multiagent systems (MAS) based on Cultural Algorithms. The model is applied to a case of study in industrial automation, related to the Agents-based System for Fault Management System. The instantiation occurs on the conversations that are defining in the MAS's coordination model, which are characterized by type of conversation that have been previously defined. A conversation can have sub-conversations, and in this case the sub-conversations are characterized by a particular type of conversation. Additionally in these conversations can occur some type of conflict, that can be solved by using different coordination mechanisms existing in the literature. For this, it is developed a model based on cultural algorithms, which is used by the MAS as a learning way in the process to determine which coordination mechanism is more suitable for a given conversation and a given scenario. The results show that the obtained model through this learning guides the MAS to determine which mechanism is better suited for a given conversation.

Technological improvements have led to situations where data sets are sufficiently rich that in the interests of processing speed it is desirable to throw away samples that provide little additional information. This is referred to here as data sparsification. The first contribution is a study of a recently proposed data sparsification scheme; ideas from vector quantisation are used to assess its performance. Informed by this study, a modification of the data sparsification algorithm is proposed and applied to the problem of estimating a kernel-based measure of independence of two datasets. (Given i.i.d. observations from two random variables, x and y, the underlying problem is to determine whether or not x and y are independent of each other.) The second contribution of this paper is to make recursive an existing algorithm for measuring independence and able to operate on both raw data and on sparsified data generated by the aforementioned data sparsification algorithm. Compared with the original algorithm, the recursive algorithm is significantly faster due to its lower memory and computational requirements.

Some aspects of cognitive processes are investigated in terms of a special theoretical extension of Algorithm Theory called Formal Technology (FT). FT has many signs of the “Metaphysics” introduced by Mario Bunge and it correlates to Aleksander Bogdanov's “Tektology”. FT differs from die latter by constructivity and exactness. Besides, many results in FT correlate with reality. In that sense FT may be considered as a mathematical development of Tektology.

A particle swarm optimization (PSO) approach towards the development of strategy co-evolution for multiple choices IPD game is presented. It is demonstrated that, birds can play IPD with multiple choices, and the co-evolutionary behaviors are influenced by social environment.

Co‐occurrence matrix has been applied successfully for echographic images characterization because it contains information about spatial distribution of grey‐scale levels in an image. The paper deals with the analysis of pixels in selected regions of interest of an US image of the liver. The useful information obtained refers to texture features such as entropy, contrast, dissimilarity and correlation extract with co‐occurrence matrix. The analyzed US images were grouped in two distinct sets: healthy liver and steatosis (or fatty) liver. These two sets of echographic images of the liver build a database that includes only histological confirmed cases: 10 images of healthy liver and 10 images of steatosis liver. The healthy subjects help to compute four textural indices and as well as control dataset. We chose to study these diseases because the steatosis is the abnormal retention of lipids in cells. The texture features are statistical measures and they can be used to characterize irregularity of tissues....

License Plate Recognition (LPR) systems commonly have framework of processing steps such as: Detection of number plate, Segmentation of plate characters and Recognition of each character. Number plate detection is a challenging task due to diversity of plate formats and environmental conditions during the image acquisition. Accuracy of character segmentation and recognition rely on the efficiency of plate detection. Various algorithms are developed for this work. Purpose of this paper is to categorize and brief them.

The change of illumination problem is still a challenging and difficult problem in face recognition under complex illumination condition. There were many proposals that dealing with illumination problem in face recognition in the past decade. In this paper, we classify the existing methods into three main categories and introduce the representative algorithms and theories, then analyze the advantages and disadvantages of these algorithms correspondingly. As a result of this study we conclude direction of future research.

This paper proposes an algorithm to enhance the performance of the spatial partitioning method for flocking behavior. Even when a moving entity (boid) in a flock changes its direction and location, its k-nearest neighbors (kNN), which influence its decision for the next direction, seldom change. Using this fact, this paper improves the performance by finding kNN of boids efficiently. A method to check that the new kNN is not changed from the previous kNN is proposed, and the correctness of the method is proven with two theorems. In order to minimize the cost of computing the new kNN, the method checks the fact that the new kNN did not change from the previous kNN. If the new kNN is not changed, the method copies the previous kNN to the new kNN instead of computing the new kNN. The proposed algorithm was implemented and its performance was compared with the original spatial partitioning method. The results of the comparison showed that the proposed algorithm outperformed the original method by about 57.7% with respect to the number of frames per second.

In this paper, we intend to formulate a new meta-heuristic algorithm, called Cuckoo Search (CS), for solving optimization problems. This algorithm is based on the obligate brood parasitic behaviour of some cuckoo species in combination with the Levy flight behaviour of some birds and fruit flies. We validate the proposed algorithm against test functions and then compare its performance with those of genetic algorithms and particle swarm optimization. Finally, we discuss the implication of the results and suggestion for further research.

In this paper we realize a brief review of the main social characteristics exhibited by the social insects (especially ants) that have been transferred in computer science as ideas for optimization and underline some of the approaches that scientists have used to solve real‐life problems. As well, we propose a modality to use pheromone information in the Ant System, considering the observation that biologists made, namely that the rate at which ants join and leave pheromone trails is not linear.

The re-ranking algorithm is a common method to use the results of subsequent stages, such as coreference resolution, to improve entity recognition. The nature of re-ranking is to select the most possible candidate from the entire candidate set. But if all of the candidates are incorrect, this method still can not give a right result. We propose a two-layer model to utilize the results of subsequent stages. This novel framework is able to overcome the disadvantage of re-ranking method, and correct the errors introduced by the first tagging. The experiments on the ACE2004 Chinese corpus show that the proposed framework are more effective than re-ranking method.

The paper presents the algorithmization of multi-factor method used to estimate the teaching quality at universities. The evaluation methods and procedures suggested in the paper are intended to make expertspsila evaluations as objective as possible. The generalized nets methods used to describe the process of algorithmization.

There are many emotion features. If all these features are employed to recognize emotions, redundant features may be existed. Furthermore, recognition result is unsatisfying and the cost of feature extraction is high. In this paper, a method to select speech emotion features based on contribution analysis algorithm of NN is presented. The emotion features are selected by using contribution analysis algorithm of NN from the 95 extracted features. Cluster analysis is applied to analyze the effectiveness for the features selected, and the time of feature extraction is evaluated. Finally, 24 emotion features selected are used to recognize six speech emotions. The experiments show that this method can improve the recognition rate and the time of feature extraction

The traditional problem of plan recognition can be solved in a straightforward manner by a graph-searching algorithm such as A*. A few simple extensions allow the algorithm to handle situations in which the user has an incorrect plan. However, it turns out that recognizing structural errors is computationally a much more complex problem than recognizing plans with simple constraint violations. Four algorithms are presented here, starting with a direct adaptation of A* for recognizing correct plans, and progressing to an optimized algorithm for recognizing violated, ill-formed plans.

The features of the algorithms abstract models are examined. The necessity of "algorithm" concept clarification and its usage for the synthesis and analysis of computer systems is shown. The new SH-model of algorithm and the universal SH-model of computer are proposed.

We implement the "circle and helix" algorithm, which is exact and of filtered-backprojection (FBP) type. The purpose is to minimize the time delay and dynamic differences between the two scans required for exact cone beam reconstruction.

We discuss an incremental method which learns a concept definition from examples and counter-examples. We present two bottom-up algorithms: MG and MGI that learn characteristic disjunctive definitions; MG is a non-incremental algorithm and MGI is incremental. Some experiments are performed, and a comparison with incremental top-down algorithms shows that MGI is faster, requires fewer instances to learn and found better concept definitions. >

The structure of the fitness landscape on which genetic programming operates is examined. The landscapes of a range of problems of known difficulty are analyzed in an attempt to determine which landscape measures correlate with the difficulty of the problem. The autocorrelation of the fitness values of random walks, a measure which has been shown to be related to perceived difficulty using other techniques, is only a weak indicator of the difficulty as perceived by genetic programming. All of these problems show unusually low autocorrelation. Comparison of the range of landscape basin depths at the end of adaptive walks on the landscapes shows good correlation with problem difficulty, over the entire range of problems examined. >

A new optimization algorithm is presented. It is inspired by the natural behavior of a human or animal scout in exploring unknown geographical regions, and in their ability to exploit information coming from past experience. The algorithm has been applied to the maximum clique problem, and favorably compared with other more specialized technique.

Recently a number of approaches solving the long object problem with spiral scan appeared in the literature. Unlike the algorithms which employ circular scans for solving the long object problem, with only spiral scan it is necessary to scan regions of the object adjacent to the ROI in order to reconstruct the ROI without contamination; the spiral path required beyond the ROI is referred to as overscan in the literature. Overscan exposes the regions in the long object adjacent to the ROI to radiation as well as incurs additional computation time In this paper we present a number of methods to reduce the undesirable overscan, with emphasis on the local ROI algorithm.

Considers the trustworthiness investigation problem for mathematical models of the structure of linguistic, information, and algorithmic supply of projecting works automation systems. Ascertainments of mathematical models trustworthiness before their practical realization, and approval of models guarantees recognition and correction of syntactic, semantic, and algorithmical mistakes.

Superresolution offers the potential to improve system performance by increasing the resolution. We have developed a suite of five algorithms, applicable to 1-D and 2-D and to the discrete and distributed cases. We also have shown specific applications of superresolution for air-to-ground surveillance, data resolution enhancement, SAR ATR and FOPEN ATR. We are developing algorithms tuned for remote sensing based on a mosaic type approach.

It has gradually been recognised that Bayesian algorithms are more widely applicable and reliable than ad hoc algorithms. Advantages include the use of explicit and realistic stochastic models making it easier to understand the working behind the algorithm and allowing confidence statements about conclusions. The authors propose a method, within a Bayesian framework, to assimilate information from images obtained from different modalities at different resolutions. The algorithm is used with a pair of images, from which a fused high resolution image and improved data reconstructions are simultaneously obtained. The authors illustrate their method by 2 examples, the first fuses a pair of SPECT and CT phantom images and the second a pair of MR brain scan images, obtained from different acquisition techniques. The authors provide a pseudo-comparison of the latter example with a commercially available package called ANALYZE. However, the phantom images from physical experiment given here provide a true validation and performance of the model.

The purpose of this report is to present a brief overview of a computer-based method designed to automatically approximate the epicardial and endocardial borders of the heart for echocardiographic images acquired from the parasternal transthoracic short-axis view. The only user input required is the end diastolic (ED) and end systolic (ES) frame numbers. The method was tested off-line on a developmental database acquired retrospectively from 55 patient studies (2 cycles/patient). The measurements provided by the computer-based method were comparable to those made by 3 expert observers.

Wang and Feng (IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 28, no. 5, p 846, May 2006) pointed out that the deduction in (Z. Lin and H. Y. Shum, IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 26, no. 1, pp. 83-97, Jan. 2004) overlooked the validity of the perturbation theorem used in (Z. Lin and H. Y. Shum, IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 26, no. 1, pp. 83-97, Jan. 2004). In this paper, we show that, when the perturbation theorem is invalid, the probability of successful superresolution is very low. Therefore, we only have to derive the limits under the condition that validates the perturbation theorem, as done in (Z. Lin and H. Y. Shum, IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 26, no. 1, pp. 83-97, Jan. 2004).

When and if the statement in the title is proved, the person who proves it will win a million dollars and become an instant megastar. He or she will certainly appear on the front page of The New York Times and should expect invitations from late-night talk shows and to be the subject of a PBS Nova special. For a long time—a very long time—this fundamental open question about the relationship between P and NP has been the large white hippo blocking the main hallway of the Academy of Complexity Theory.

Algorithm theories represent the structure common to a class of algorithms, such as divide-and-conquer or backtrack. An algorithm theory for a class A provides the basis for design tactics—specialized methods for designing A -algorithms from formal problem specifications. We illustrate this approach with recent work on the theory of global search algorithms and briefly mention several others. Several design tactics have been implemented in the kids / cypress system and have been used to semiautomatically derive many algorithms.