Michiharu Maeda

Qubit neuron according to quantum circuit for XOR problem

This paper presents a novel learning model with qubit neuron according to quantum circuit for XOR problem and describes the influence to learning by reducing the number of neurons. Our approach has a 3-qubit neuron including a work qubit in the input layer, which employs gradient descent for learning. For improving the learning efficiency, furthermore, a momentum term is added to gradient descent. For the number of neurons in the output layer, the convergence rate and the average iteration for learning are examined. Especially it is exhibited that our model can learn for one neuron in the output layer. Experimental results are presented in order to show that our approach is effective in the convergence rate and the average iteration.

Reduction of artificial bee colony algorithm for global optimization

Abstract This paper presents a reduction of artificial bee colony algorithm for global optimization. Artificial bee colony algorithm is an optimization technique which refers to the behavior of honeybee swarms, and a multi-point search approach which finds a best solution using multiple bees. For avoiding local minima, a number of bees are initially prepared and their positions are updated by artificial bee colony algorithm. Bees sequentially reduce to reach a predetermined number of them grounded in the evaluation value and artificial bee colony algorithm continues until the termination condition is met. In order to show the effectiveness of the proposed algorithm, we examine the best value by using test functions compared to existing algorithms. Furthermore the influence of best value on the initial number of bees for our algorithm is discussed.

Bagging and AdaBoost algorithms for vector quantization

In this paper, we propose VQ methods based on ensemble learning algorithms Bagging and AdaBoost. The proposed methods consist of more than one weak learner, which are trained in parallel or sequentially. In Bagging, the weak learners are trained in parallel by using randomly selected data from a given data set. The output for Bagging is given as the average among the weak learners. In AdaBoost, the weak learners are sequentially trained. The first weak learner is trained by using randomly selected data from a given data set. For the second and later weak learners, the probability distribution of learning data is modified so that each weak learner focuses on data involving higher error for the previous weak one. The output for AdaBoost is given as the weighted average among the weak learners. The presented simulation results show that the proposed methods can achieve a good performance in shorter learning times than conventional ones such as K-means and NG.

A relaxation algorithm influenced by self-organizing maps

A relaxation algorithm influenced by self-organizing maps for image restoration is presented in this study. Self-organizing maps have been hitherto studied for the ordering process and the convergence phase of weight vectors. As another approach of self-organizing maps, a novel algorithm of image restoration is proposed. The present algorithm creates a map containing one unit for each pixel. Utilizing pixel values as input, the image inference is carried out by self-organizing maps. Then, an updating function with a threshold is introduced, so as not to respond to a noisy input sensitively. Therefore, the inference of original image proceeds appropriately since any pixel is influenced by surrounding pixels corresponding to the neighboring setting. In the restoration process, the effect of the initial threshold and the initial neighborhood on accuracy is examined. Experimental results are presented in order to show that the present method is effective in quality.

A multiple vector quantization approach to image compression

This paper investigates the effectiveness of a parallelized approach to VQ based image compression. In particular, we consider an image compression method using multiple VQs. The method, called MVQ, generates multiple independent codebooks to compress an image by using a neural network algorithm. In the image restoration, MVQ restores low quality images from the multiple codebooks, and then combines the low quality ones into a high quality one. Further, we present an effective coding scheme for codebook indexes to overcome the in-efficiency of MVQ in compression rate. Our simulation results show that the MVQ method outperforms a conventional single-VQ method when the compression rate is smaller than some values.

Set-based particle swarm optimization with status memory for knapsack problem

Set-based particle swarm optimization (S-PSO) operates on discrete space. S-PSO can solve combinatorial optimization problem with high quality and is successful to apply to the large-scale problem. In S-PSO, a velocity is a set with possibility and a position is a candidate solution. In this paper, we present a novel algorithm of set-based particle swarm optimization with status memory (S-PSOSM) to decide the position based on the previous position for solving knapsack problem. Some operators are redefined for S-PSOSM. S-PSOSM is a simple algorithm because the state of probability reduces. In addition, the weight of S-PSOSM is discussed. S-PSOSM shows high qualities in experimental results.

Reduction methods of fuzzy inference rules with neural network learning algorithm

Describes reduction methods of the rule unit with fuzzy neural networks. The approaches are presented with a reducing mechanism of the rule unit which use three parameters, central value, width of the membership function in the antecedent part, and real number in the consequent part, constituted according to a fuzzy neural system. These methods indicate that a different technique exists besides the reduction approach. Experimental results are presented in order to show that the effectiveness is different in the proposed techniques for average inference error and learning iteration.

Restoration Model with Inference Capability of Self-Organizing Maps

This paper presents a restoration model with inference capability of self-organizing maps. Self-organizing maps have been studied principally for the ordering process and the convergence phase of weight vectors. As a novel approach of self-organizing maps, a restoration model for a defective image is proposed. The model creates a map containing one unit for each pixel. Utilizing pixel values as input, the inference for lost pixels is conducted by self-organizing maps. The inference of an original image proceeds appropriately since any pixel is influenced by neighboring pixels corresponding to the neighboring setting. Consequentially, images with high quality are constituted by restoring lost pixels. Experimental results are presented in order to show that our approach is effective in quality for restoration of lost pixels.

A learning model in qubit neuron according to quantum circuit

This paper presents a novel learning model in qubit neuron according to quantum circuit and describes the influence to learning with gradient descent by changing the number of neurons. The first approach is to reduce the number of neurons in the output layer for the conventional technique. The second is to present a novel model, which has a 3-qubit neuron including a work qubit in the input layer. For the number of neurons in the output layer, the convergence rate and the average iteration for learning are examined. Experimental results are presented in order to show that the present method is effective in the convergence rate and the average iteration for learning.

A Novel Approach of Set-Based Particle Swarm Optimization with Memory State

This paper presents a novel approach of set-based particle swarm optimization with memory state for solving traveling salesman problem. Particle swarm optimization achieves the social model of bird flocking and fish schooling and solves continuous optimization problem. Set-based particle swarm optimization functions in discrete space by using a set and solves combinatorial optimization problem with successfully applying to the large-scale problem. Our approach selects the best position among different positions from the current generation for creating a solution according to a velocity. In order to show the effectiveness of our approach, numerical experiments are presented for traveling salesman problem compared to existing algorithms.