Akira Namatame

New Frontiers in Artificial Intelligence

Neg-Raising (NR) verbs form a class of verbs with a clausal complement that show the following behavior: when a negation syntactically attaches to the matrix predicate, it can semantically attach to the embedded predicate. This paper presents an account of NR predicates within Tree Adjoining Grammar (TAG). We propose a lexical semantic interpretation that heavily relies on a Montague-like semantics for TAG and on higher-order types.

Intelligent and Evolutionary Systems

Artificial evolutionary systems are computer systems, inspired by ideas from natural evolution and related phenomena. The field has a long history, dating back to the earliest days of computer science, but it has only become an established scientific and engineering discipline since the 1990s, with packages for the commonest form, genetic algorithms, now widely available. Researchers in the Asia-Pacific region have participated strongly in the development of evolutionary systems, with a particular emphasis on the evolution of intelligent solutions to highly complex problems. The Asia-Pacific Symposia on Intelligent and Evolutionary Systems have been an important contributor to this growth in impact, since 1997 providing an annual forum for exchange and dissemination of ideas. Participants come primarily from East Asia and the Western Pacific, but contributions are welcomed from around the World. This volume features a selection of fourteen of the best papers from recent APSIES. They illustrate the breadth of research in the region, with applications ranging from business to medicine, from network optimization to the promotion of innovation.

STRUCTURAL CONNECTIONIST LEARNING WITH COMPLEMENTARY CODING

We propose a new learning algorithm, structural learning with the complementary coding for concept learning problems. We introduce the new grouping measure that forms the similarity matrix over the training set and show this similarity matrix provides a sufficient condition for the linear separability of the set. Using the sufficient condition one should figure out a suitable composition of linearly separable threshold functions that classify exactly the set of labeled vectors. In the case of the nonlinear separability, the internal representation of connectionist networks, the number of the hidden units and value-space of these units, is pre-determined before learning based on the structure of the similarity matrix. A three-layer neural network is then constructed where each linearly separable threshold function is computed by a linear-threshold unit whose weights are determined by the one-shot learning algorithm that requires a single presentation of the training set. The structural learning algorithm proceeds to capture the connection weights so as to realize the pre-determined internal representation. The pre-structured internal representation, the activation value spaces at the hidden layer, defines intermediate-concepts. The target-concept is then learned as a combination of those intermediate-concepts. The ability to create the pre-structured internal representation based on the grouping measure distinguishes the structural learning from earlier methods such as backpropagation.

Dynamic diffusion in evolutionary optimised networks

Diffusion is the process by which new products and practices are invented and successfully introduced into a society. This paper presents a possible explanation of this phenomenon in terms of a network of interacting agents whose decisions are determined by the action of their neighbours according to a probabilistic model. It is known that the maximum eigenvalue of the network decides a tipping point of a diffusion process by probabilistic model. The network with large maximum eigenvalue is susceptible to a diffusion process. Evolutionary optimisation is used to make the network in which the diffusion process will start more early than in other networks. Two properties are identified in which the network is suitable for fast diffusion. These are a power law of degree distribution and the phenomena in which hub nodes are connected very densely, it is called a rich-club phenomena. Finally, the results of numerical diffusion simulation are compared with other network topology to verify the performance of evolutionary optimised networks.

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.

Emergent Intelligence of Networked Agents

The study of intelligence emerged from interactions among agents has been popular. In this study it is recognized that a network structure of the agents plays an important role. The current state-of-the art in agent-based modeling tends to be a mass of agents that have a series of states that they can express as a result of the network structure in which they are embedded. Agent interactions of all kinds are usually structured with complex networks. The idea of combining multi-agent systems and complex networks is also particularly rich and fresh to foster the research on the study of very large-scale multi-agent systems. Yet our tools to model, understand, and predict dynamic agent interactions and their behavior on complex networks have lagged far behind. Even recent progress in network modeling has not yet offered us any capability to model dynamic processes among agents who interact at all scales on complex networks.This book is based on communications given at the Workshop on Emergent Intelligence of Networked Agents (WEIN 06) at the Fifth International Joint Conference on Autonomous Agents and Multi-agent Systems (AAMAS 2006), which was held at Future University, Hakodate, Japan, from May 8 to 12, 2006. WEIN 06 was especially intended to increase the awareness of researchers in these two fields sharing the common view on combining agent-based modeling and complex networks in order to develop insight and foster predictive methodologies in studying emergent intelligence on of networked agents. From the broad spectrum of activities, leading experts presented important paper and numerous practical problems appear throughout this book. The papers contained in this book are concerned with emergence of intelligent behaviors over networked agents and fostering the formation of an active multi-disciplinary community on multi-agent systems and complex networks.

Evolving neural network models

Neural networks in nature are not designed but evolved, and they should learn their structure through the interaction with their environment. The paper introduces the notion of an adaptive neural network model with reflection. We show how reflection can implement adaptive processes, and how adaptive mechanisms are actualized using the concept of reflection. Learning mechanisms must be understood in terms of their specific adaptive functions. We introduce an adaptive function which makes the network able to adjust its internal structure by itself to by modifying its adaptive function and associated learning parameters. We then provide the model of emergent neural networks. We show that the emergent neural network model is especially suitable for constructing large scale and heterogeneous neural networks with the composite and recursive architectures, where each component unit is modeled to be another neural network. Using the emergent neural network model, we introduces the concepts of composition and recursion for integrating heterogeneous neural network modules which are trained individually.

Management of systemic risks and cascade failures in a networked society

The qualification of risks lies in their systemic nature. A risk to one subsystem may present an opportunity to another subsystem. A systemic risk is the possibility that an event will trigger a loss of confidence in a substantial portion of the system serious enough to have adverse consequences on system performance. A systemic risk therefore impacts the integrity of the whole system. In this chapter, we use networks to address risks. A network, which is a collection of nodes with links between them, can be a useful representation of a system. For instance, network analysis can explain certain systemic risks in financial systems by modeling interactions among financial institutions. A network approach is also useful in mitigating large blackouts caused by cascade failure. In this case, instead of looking at the details of particular blackouts, we investigate the cascade dynamics as a series of blackouts using risk propagation models. We use a general framework of diffusion or contagion models to describe risk propagation in networks and investigate how network topology impacts risk propagation patterns. At the macroscopic level, systemic risk is measured as the fraction of failed nodes. We divide our discussion into two classes of diffusion. First, we discuss so-called progressive diffusion processes. Many diffusion processes are progressive in the sense that once a node switches from one state to another, it remains in that state in all subsequent time steps. The second class is non-progressive diffusion where, as time progresses, nodes can switch back and forth from one state to the other, depending on the states of their neighbors. Networks increase interdependency,creating challenges formanaging risks. This is especially apparent in areas such as financial institutions and enterprise risk management, where the actions of a single actor in an interconnected network can impact all the other actors in the network. The network is only as strong as its weakest link, and trade-offs are most often connected to a function that models system performance management. In this context, there is a class of problems, ranging

Evolving Failure Resilience in Scale-Free Networks

Today our society tends to become more and more dependent on large scale (global) infrastructure networks. In many cases, attacks on a few important nodes of such systems lead to irreparable local or, worse, global damages. Thus, designing resilient networks rather than reducing the effects of some unexpected attacks becomes a must. As the most resilient network, regarding any kind of attacks, should be a full-connected graph, it is obvious that implementing such a network is a utopia. This paper proposes an original multi-objective method for optimizing complex networks’ structure, taking into account the implementation costs. A micro genetic algorithm is used in order to improve networks’ resilience to targeted attacks on HUB nodes while keeping the implementation costs as low as possible.

Massive Multi-Agent Simulation in 3D

In this paper, we discuss our challenge on how to give the creatures and ability to follow spatial restriction while keeping the complexity low enough to still allow for real-time simulation of the herd. Our methodologies extend the pioneering work by Reynolds’ flocking algorithm. We extend how the herd can move in natural-looking paths. Also, we show like the creatures to travel smoothly in 3D space with speed regulation in curve.