Akihiro Yamaguchi

Multiple targets enclosure by robotic swarm

Target enclosure by autonomous robots is useful for many practical applications, for example, surveillance of disaster sites. Scalability is important for autonomous robots because a larger group is more robust against breakdown, accidents, and failure. However, since the traditional models have discussed only the cases in which minimum number of robots enclose a single target, there has been no study on the utilization of the redundant number of robots. In this paper, to achieve a highly scalable target enclosure model about the number of target to enclose, we introduce swarm based task assignment capability to Takayamas enclosure model. The original model discussed only single target environment but it is well suited for applying to the environments with multiple targets. We show the robots can enclose the targets without predefined position assignment by analytic discussion based on switched systems and a series of computer simulations. As a consequence of this property, the proposed robots can change their target according to the criterion about robot density while they enclose multiple targets.

A simple braking model for detecting incidents locations by smartphones

Recently, there have been strong demand and interest for developing methods to analyze driving data for extracting traffic safety information. In automobile research field, several methods for detecting sudden braking have been proposed; however, these methods cannot answer the question what is the causes of such sudden braking events. In previous research, we have proposed a method to estimate incidents locations which interfered with smooth driving by using smartphone, but the method just works for the cases of vehicle stop. In this paper, we propose a new braking model which can estimate incidents locations caused sudden braking for both cases of vehicle stop and non-stop. We take real world experiments in order to validate the incidents map result. The result shows that based on the proposed method, incidents map is accurately achieved.

Pseudorandom number generation using chaotic true orbits of the Bernoulli map

We devise a pseudorandom number generator that exactly computes chaotic true orbits of the Bernoulli map on quadratic algebraic integers. Moreover, we describe a way to select the initial points (seeds) for generating multiple pseudorandom binary sequences. This selection method distributes the initial points almost uniformly (equidistantly) in the unit interval, and latter parts of the generated sequences are guaranteed not to coincide. We also demonstrate through statistical testing that the generated sequences possess good randomness properties.

Individual recognition-free target enclosure model

Target enclosure by autonomous robots is useful for many practical applications, for example, surveillance of disaster sites. Scalability is important for autonomous robots because a larger group is more robust against breakdown, accidents, and failure. However, it is more difficult to operate a larger group of robots because their individual capacity for recognizing team-mates should be higher. In this paper, to achieve a highly scalable target enclosure model, we demonstrate a new condition for Takayama’s enclosure model. The original model requires a static relationship between agents. However, robots can form an enclosure even under a dynamic topology on the basis of a nearest neighbor graph; hence, they do not require recognition capability. We confirm this by an analytical discussion of switched systems and a series of computer simulations.

A New Multiple Label Propagation Algorithm for Linked Open Data

In recent years, Linked Open Data (LOD) has been attracting attention in the world since the LOD have features such as semantic web and ontology. Also, applications using the LOD have been increasing year after year in Japan and in all over the world. However, there are a problem that most of existing LOD are not linked with other LOD. Therefore, original features of the LOD have been lost. In this study, we propose a new Multiple Label Propagation algorithm for linking each other the existing LOD to bring out the original features by semi-supervised learning. This algorithm predicts semantic labels of individual data of the LOD using an assumption that adjacent nodes belong to a same or similar class. Therefore, the algorithm has new formula to consider semantic distance. Thereby, the algorithm is possible to extend links between neighborhood nodes (data). In addition, the algorithm is possible to correspond multi-label data. The results of experiments showed that the algorithm can give the label prediction values representing semantic distances, and can be processed much faster than a conventional algorithm.

Link prediction of LOD by multiple label propagation algorithm considering semantic distance

This paper presents link prediction of Linked Open Data (LOD) by Multiple Label Propagation Algorithm (MLPA). The current LOD do not have enough links. Therefore, the LOD have not been able to exert so much semantic characteristics. In order to solve this problem, we proposed the MLPA considering semantic distance. The MLPA can expand potential links of each data contained in the LOD. The experimental result of the MLPA shows the good performance and the validity of this algorithm is confirmed.

Feature Linking using Synchronized Responses in Chaotic Cellar Neural Networks for Visual Stimulus of Moving Objects

A feature-linking mechanism using the synchronized responses of neural assemblies was studied for chaotic cellar neural networks (Chaotic-CNN). Chaotic-CNNs consist of chaotic spike response neurons that show chaotic interspike-interval dynamics. In our scheme of feature linking, the features of the target objects are linked by the synchronized spike responses characterized by the temporal chaotic pattern of the spike sequence. In this paper, we analyze the synchronized spike responses invoked by the visual stimulus of moving bars. Consequently, the resulting neural assemblies have higher correlation for the visual stimulus of two bars moving in the same direction than in the opposite direction. We also discuss the possibility of feature linking using the chaotic synchronized responses.

Target Enclosure for Multiple Targets

Target enclosure by autonomous robots is useful for many practical applications, for example, surveillance of disaster sites. Scalability is important for autonomous robots because a larger group is more robust against breakdown, accidents, and failure. However, traditional models only discussed cases in which minimum number of robots enclose a single target so that there is no way to utilize the redundant number of robots. In this paper, to achieve a highly scalable target enclosure model about the number of target to enclose, we introduce swarm based task assignment capability to Takayama et al.’s enclosure model. Our robots can enclose without any global predefined position assignments for each target,, for example, a Hamiltonian cycle, a chain structure etc.. The behavior is shown by computer simulations.

Decomposition of Superimposed Chaotic Spike Sequences by Using the Bifurcating Neuron

In this study, decomposition of superimposed chaotic spike sequence was investigated from the view point of neural information coding. We construct simple network of bifurcating neuron and introduce the coupling model to decompose superimposed chaotic spike sequences. The decomposing performance was demonstrated by the numerical simulation and evaluated by the ratio of synchronized spikes. As a result, for the superimposed two chaotic spike sequences, approximately 90% of spikes were correctly decomposed.

Document Analysis of Survey on Employment Trends in Japan

This paper analyses the Survey on Employment Trends to propose methods for discovering the regularity of employment across different industrial sectors. Industrial sectors seem to have cooperative and competitive relationships, for example, through information and logistics. Therefore, some type of regularity in employment is expected across the industrial sectors. We examined the increases and decreases in employment for each industrial sector in Japan using the Survey on Employment Trends and found that most sectors were similar and a simple correlation analysis did not provide a satisfactory solution. Therefore, we developed the idea of applying the latent Dirichlet allocation (LDA) to the employment data; LDA is a method for finding the subjects (or topics) used in documents. This machine learning technique is based on the assumption that words reflect the topic; it helps to find topics from documents without predefining the keywords. We created a hypothetical document interlinking the results of the Survey on Employment Trends between 2000 and 2014, where the latent topic analysis was applied. In this study, we documented the numeric data from the Survey on Employment Trends by thresholding the numeric data to symbolize the data. As a result, we illustrated that there was a significant change in employment in the publication, food, mining, electricity, gas and water industries after 2009.