Image segmentation based on competitive learning

Image segment is a primary step in image analysis of unexploded ordnance (UXO) detection by ground p enetrating radar (GPR) sensor which is accompanied with a lot of noises and other elements that affect the recognition of real target size. In this paper we bring forward a new theory, that is, we look the weight sets as target vector sets which is the new cues in semi-automatic segmentation to form the final image segmentation. The experiment results show that the measure size of target with our method is much smaller than the size with other methods and close to the real size of target.

Improvement of Stochastic Competitive Learning for Social Network

As an unsupervised learning method,stochastic competitive learning is commonly used for community detection in social network analysis.Compared with the traditional community detection algorithms,it has the advantage of realizing the time-series community detection by simulating the community formation process.In order to improve the accuracy and solve the problem that several parameters in stochastic competitive learning need to be pre-set,the author improves the algorithms and realizes improved stochastic competitive learning by particle position initialization,parameter optimization and particle domination ability self-adaptive.The experiment result shows that each improved method improves the accuracy of the algorithm,and the F1 score of the improved algorithm is 9.07%higher than that of original algorithm.

Hawk and pigeon's intelligence for UAV swarm dynamic combat game via competitive learning pigeon-inspired optimization

For unmanned aerial vehicle(UAV)swarm dynamic combat,swarm antagonistic motion control and attack target allocation are extremely challenging sub-tasks.In this paper,the competitive learning pigeon-inspired optimization(CLPIO)algorithm is proposed to handle the cooperative dynamic combat problem,which integrates the distributed swarm antagonistic motion and centralized attack target allocation.Moreover,the threshold trigger strategy is presented to switch two sub-tasks.To seek a feasible and optimal combat scheme,a dynamic game approach combined with hawk grouping mechanism and situation assessment between sub-groups is designed to guide the solution of the optimal attack scheme,and the model of swarm antagonistic motion imitating pigeon’s intelligence is proposed to form a confrontation situation.The analysis of the CLPIO algorithm shows its convergence in theory and the comparison with the other four metaheuristic algorithms shows its superiority in solving the mixed Nash equilibrium problem.Finally,numerical simulation verifis that the proposed methods can provide an effective combat scheme in the set scenario.

Research of Dynamic Competitive Learning in Neural Networks

Introduce a method of generation of new units within a cluster and aalgorithm of generating new clusters. The model automatically builds up its dynamically growinginternal representation structure during the learning process. Comparing model with other typicalclassification algorithm such as the Kohonen's self-organizing map, the model realizes a multilevelclassification of the input pattern with an optional accuracy and gives a strong support possibilityfor the parallel computational main processor. The idea is suitable for the high-level storage ofcomplex datas structures for object recognition.

Kernel and graph： Two approaches for nonlinear competitive learning clustering

Competitive learning has attracted a signif- icant amount of attention in the past decades in the field of data clustering. In this paper, we will present two works done by our group which address the nonlin- early separable problem suffered by the classical com- petitive learning clustering algorithms. They are ker- nel competitive learning （KCL） and graph-based multi- prototype competitive learning （GMPCL）, respectively. In KCL, data points are first mapped from the input data space into a high-dimensional kernel space where the nonlinearly separable pattern becomes linear one. Then the classical competitive learning is performed in this kernel space to generate a cluster structure. To real- ize on-line learning in the kernel space without knowing the explicit kernel mapping, we propose a prototype de- scriptor, each row of which represents a prototype by the inner products between the prototype and data points as well as the squared length of the prototype. In GM- PCL, a graph-based method is employed to produce an initial, coarse clustering. After that, a multi-prototype competitive learning is introduced to refine the coarse clustering and discover clusters of an arbitrary shape. In the multi-prototype competitive learning, to gener- ate cluster boundaries of arbitrary shapes, each cluster is represented by multiple prototypes, whose subregions of the Voronoi diagram together approximately charac- terize one cluster of an arbitrary shape. Moreover, we introduce some extensions of these two approaches with experiments demonstrating their effectiveness.

Research on the Reform of the Course“Reading of Concrete Structure Plan and Construction Drawings”Under the Background of“Promoting Teaching and Learning Through Competitions”

The inherent teaching approach can no longer meet the demands of society.In this paper,current issues within the teaching landscape of architectural engineering technology in higher vocational colleges as well as the policies and teaching demands that formed the basis of this model were analyzed.The study shows the importance of the implementation of the teaching model“promoting teaching and learning through competitions.”This model puts emphasis on the curriculum and teaching resources,while also integrating the teaching process and evaluation with competition.These efforts aim to drive education reform in order to better align with the objectives of vocational education personnel training,while also acting as a reference for similar courses.

Creating Optimal Language Learning Environment

From the viewpoint of psycholinguistics, this paper concerns how to create an optimal language learning environment in language learning, to stimulate students enthusiasm to participate in classroom activities and to make language learning easier and more pleasant.

Extending self-organizing maps for supervised classification of remotely sensed data

An extended self-organizing map for supervised classification is proposed in this paper. Unlike other traditional SOMs, the model has an input layer, a Kohonen layer, and an output layer. The number of neurons in the input layer depends on the dimensionality of input patterns. The number of neurons in the output layer equals the number of the desired classes. The number of neurons in the Kohonen layer may be a few to several thousands, which depends on the complexity of classification problems and the classification precision. Each training sample is expressed by a pair of vectors ： an input vector and a class codebook vector. When a training sample is input into the model, Kohonen＇s competitive learning rule is applied to selecting the winning neuron from the Kohouen layer and the weight coefficients connecting all the neurons in the input layer with both the winning neuron and its neighbors in the Kohonen layer are modified to be closer to the input vector, and those connecting all the neurons around the winning neuron within a certain diameter in the Kohonen layer with all the neurons in the output layer are adjusted to be closer to the class codebook vector. If the number of training sam- ples is sufficiently large and the learning epochs iterate enough times, the model will be able to serve as a supervised classifier. The model has been tentatively applied to the supervised classification of multispectral remotely sensed data. The author compared the performances of the extended SOM and BPN in remotely sensed data classification. The investigation manifests that the extended SOM is feasible for supervised classification.

Controlling Remote Robots Based on Zidan’s Quantum Computing Model

In this paper,we propose a novel algorithm based on Zidan’s quantum computing model for remotely controlling the direction of a quantumcontrolledmobile robot equippedwith n-movements.The proposed algorithm is based on the measurement of concurrence value for the different movements of the robot.Consider a faraway robot that moves in the deep space(e.g.,moves toward a galaxy),and it is required to control the direction of this robot from a ground station by some person Alice.She sends an unknown qubitα|0)+β|1)via the teleportation protocol to the robot.Then,the proposed algorithm decodes the received unknown qubit into an angleθ,that determines the motion direction of the robot,based on the concurrence value.The proposed algorithm has been tested for four and eight movements.Two simulators have been tested;IBM Quantum composer and IBM’s system,The two simulators achieved the same result approximately.The motion of any part of the robot is considered,if it has a pre-existing sensor system and a locomotive system,.We can use this technique in many places like in space robots(16 directions).The results show that the proposed technique can be easily used for a huge number of movements.However,increasing the number of movements of the robot will increase the number of qubits.

Discriminative training of GMM-HMM acoustic model by RPCL learning

This paper presents a new discriminative approach for training Gaussian mixture models(GMMs)of hidden Markov models(HMMs)based acoustic model in a large vocabulary continuous speech recognition(LVCSR)system.This approach is featured by embedding a rival penalized competitive learning(RPCL)mechanism on the level of hidden Markov states.For every input,the correct identity state,called winner and obtained by the Viterbi force alignment,is enhanced to describe this input while its most competitive rival is penalized by de-learning,which makes GMMs-based states become more discriminative.Without the extensive computing burden required by typical discriminative learning methods for one-pass recognition of the training set,the new approach saves computing costs considerably.Experiments show that the proposed method has a good convergence with better performances than the classical maximum likelihood estimation(MLE)based method.Comparing with two conventional discriminative methods,the proposed method demonstrates improved generalization ability,especially when the test set is not well matched with the training set.

Application of a soft competition learning method in document clustering

Hard competition learning has the feature that each point modifies only one cluster centroid that wins. Correspondingly, soft competition learning has the feature that each point modifies not only the cluster centroid that wins, but also many other cluster centroids near this point. A soft competition learning method is proposed. Centroid all rank distance （CARD）, CARDx, and centroid all rank distance batch K-means （CARDBK） are three clustering algorithms that adopt the proposed soft competition learning method. Among them the extent to which one point affects a cluster centroid depends on the distances from this point to the other nearer cluster centroids, rather than just the rank number of the distance from this point to this cluster centroid among the distances from this point to all cluster centroids. In addition, the validation experiments are carried out in order to compare the three soft competition learning algorithms CARD, CARDx, and CARDBK with several hard competition learning algorithms as well as neural gas （NG） algorithm on five data sets from different sources. Judging from the values of five performance indexes in the clustering results, this kind of soft competition learning method has better clustering effect and efficiency, and has linear scalability.

A Necessary Condition about the Optimum Partition on a Finite Set of Samples and Its Application to Clustering Analysis

This paper presents another necessary condition about the optimum parti-tion on a finite set of samples. From this condition, a corresponding generalized sequential hao f k-means (GSHKM) clustering algorithm is built and many well-known clustering algorithms are found to be included in it. Under some assumptions the well-known MacQueen's SHKM (Sequential Hard K-Means)algorithm, FSCL (Frequency Sensitive Competitive Learning) algorithm and RPCL (Rival Penalized Competitive Learning) algorithm are derived. It is shown that FSCL in fact still belongs to the kind of GSHKM clustering algth rithm and is more suitable for producing means of K-partition of sample data,which is illustrated by numerical experiment. Meanwhile, some improvements on these algorithms are also given.