Machine Learning Mechanisms Research Articles

Wearable Networked Sensing for Human Mobility and Activity Analytics: A Systems Study.

This paper presents implementation details, system characterization, and the performance of a wearable sensor network that was designed for human activity analysis. Specific machine learning mechanisms are implemented for recognizing a target set of activities with both out-of-body and on-body processing arrangements. Impacts of energy consumption by the on-body sensors are analyzed in terms of activity detection accuracy for out-of-body processing. Impacts of limited processing abilities in the on-body scenario are also characterized in terms of detection accuracy, by varying the background processing load in the sensor units. Through a rigorous systems study, it is shown that an efficient human activity analytics system can be designed and operated even under energy and processing constraints of tiny on-body wearable sensors.

Adaptive Detection of Design Flaws

Criteria for software quality measurement depend on the application area. In large software systems criteria like maintainability, comprehensibility and extensibility play an important role.My aim is to identify design flaws in software systems automatically and thus to avoid “bad” — incomprehensible, hardly expandable and changeable — program structures.Depending on the perception and experience of the searching engineer, design flaws are interpreted in a different way. I propose to combine known methods for finding design flaws on the basis of metrics with machine learning mechanisms, such that design flaw detection is adaptable to different views.This paper presents the underlying method, describes an analysis tool for Java programs and shows results of an initial case study.

Recognition and Utilization of Clausal Relations in Complex Sentences for Improving the Performance of Machine Translation Systems

Machine translation systems are inefficient when translating complex sentences. An important reason for this, among others, comes from the fact that the systems translate a complex sentence without taking into account the clausal relation between main and subordinate clauses. We devise a method for recognizing and utilizing the clausal relation to make the machine translation systems self-improve their performance. First, we attempt to find the clausal relation between the two clauses in a complex sentence through a machine learning mechanism. Second, utilizing this information, we try to pre-edit or modify the sentence so that the clausal relation is expressed overtly in the modified sentence. Experiments with a machine translation system have shown that our idea and methodology improve the performance of the system.

Research on the Construction of Fuzzy Classifier System for Multidimensional Pattern Classification Using Genetic Algorithms

This paper discusses the application and performance of multidimensional pattern classification problems using Michigan approach based on fuzzy genetics-based machine learning mechanism, and proposes a new approach. In the approach, each fuzzy if-then rule is handled as an individual, and a fitness value is assigned to it. The approach not only retrieves fuzzy if-then rules, but also tunes the membership functions of each dimension, meanwhile the selection mechanism based on the similarity of individuals is involved to reduce the high selective pressure, keep the diversity of population, and avoid the premature convergence problem consequently. Finally the experiments prove that the approach has a better correct classification rate and a better adaptability on multidimensional pattern classification problems.

土木構造物の景観設計支援システム開発のための3次元形状プリミティブの感性評価機構

In general, the landscape evaluation of civil infrastructures is done based on the approach which intends to link the KANSEI (sensitivity) evaluation of the whole landscape and dimensions of structures and natural environments. In this research, we focused on the details and propose a new approach, which intends to link the dimensions of 3D shape primitives and the KANSEI information of people watching the shapes and to cluster similar shapes in terms of KANSEI by using questionnaires and the fuzzy set theory. We applied this method to cone primitives and performed questionnaire and analysis. In addition, we designed a system which shows appropriate 3D primitives corresponding to the request of a designer and which has a machine learning mechanism for individual designers. We discussed the application of this methodology to lighting installations of a dam.

A computational model for distributed knowledge systems with learning mechanisms

This paper addresses the issues of machine learning in distributed knowledge systems, which will consist of distributed software agents with problem solving, communication and learning functions. To develop such systems, we must analyze the roles of problem-solving and communication capabilities among knowledge systems. To facilitate the analyses, we propose a computational model: LPC. The model consists of a set of agents with (a) a knowledge base for learned concepts, (b) a knowledge base for problem solving, (c) prolog-based inference mechanisms and (d) a set of beliefs on the reliability of the other agents. Each agent can improve its own problem-solving capabilities by deductive learning from the given problems, by memory-based learning from communications between the agents and by reinforcement learning from the reliability of communications between the other agents. An experimental system of the model has been implemented in Prolog language on a Window-based personal computer. Intensive experiments have been carried out to examine the feasibility of the machine learning mechanisms of agents for problem-solving and communication capabilities. The experimental results have shown that the multiagent system improves the performance of the whole system in problem solving, when each agent has a higher learning ability or when an agent with a very high ability for problem solving joins the organization to cooperate with the other agents in problem solving. These results suggest that the proposed model is useful in analyzing the learning mechanisms applicable to distributed knowledge systems.

Towards method-independent knowledge acquisition

Rapid prototyping and tool reusability have pushed knowledge acquisition research to investigate method-specific knowledge acquisition tools appropriate for predetermined problem-solving methods. We believe that method-dependent knowledge acquisition is not the only approach. The aim of our research is to develop powerful yet versatile machine learning mechanisms that can be incorporated into general-purpose but practical knowledge acquisition tools. This paper shows through examples the practical advantages of this approach. In particular, we illustrate how existing knowledge can be used to facilitate knowledge acquisition through analogy mechanisms within a domain and across domains. Our sample knowledge acquisition dialogues with a domain expert illustrate which parts of the process are addressed by the human and which parts are automated by the tool, in a synergistic cooperation for knowledge-base extension and refinement. The paper also describes briefly the EXPECT problem-solving architecture that facilitates this approach to knowledge acquisition.

LEFT — A learning tool for early floorplanning

In this paper a new, interactive approach to floorplanning is presented. The learning tool LEFT provides an environment for the creation of floorplan topologies. LEFT's input is a list of blocks to be placed on a two-dimensional area and a specification of their connections. The designer uses LEFT as an extended, graphical design editor and selects operators that perform single floorplanning steps. In this way a floorplan is created interactively. LEFT's learning component creates from observed design steps specific production rules and generalizes them by means of a new and optimized inductive machine learning mechanism. Each time a rule matches on the current state of the design process LEFT proposes the execution of the corresponding operator in the right-hand rule side of the rule. In this way LEFT works as an apprentice system that supports the creation of early floorplans, adapts to the design style of the user and increases its knowledge during the use of the system.