Tsuyoshi Idé

Semi-supervised local Fisher discriminant analysis for dimensionality reduction

When only a small number of labeled samples are available, supervised dimensionality reduction methods tend to perform poorly because of overfitting. In such cases, unlabeled samples could be useful in improving the performance. In this paper, we propose a semi-supervised dimensionality reduction method which preserves the global structure of unlabeled samples in addition to separating labeled samples in different classes from each other. The proposed method, which we call SEmi-supervised Local Fisher discriminant analysis (SELF), has an analytic form of the globally optimal solution and it can be computed based on eigen-decomposition. We show the usefulness of SELF through experiments with benchmark and real-world document classification datasets.

Semi-supervised local fisher discriminant analysis for dimensionality reduction

When only a small number of labeled samples are available, supervised dimensionality reduction methods tend to perform poorly due to overfitting. In such cases, unlabeled samples could be useful in improving the performance. In this paper, we propose a semi-supervised dimensionality reduction method which preserves the global structure of unlabeled samples in addition to separating labeled samples in different classes from each other. The proposed method has an analytic form of the globally optimal solution and it can be computed based on eigendecompositions. Therefore, the proposed method is computationally reliable and efficient. We show the effectiveness of the proposed method through extensive simulations with benchmark data sets.

Computing Correlation Anomaly Scores Using Stochastic Nearest Neighbors

This paper addresses the task of change analysis of correlated multi-sensor systems. The goal of change analysis is to compute the anomaly score of each sensor when we know that the system has some potential difference from a reference state. Examples include validating the proper performance of various car sensors in the automobile industry. We solve this problem based on a neighborhood preservation principle - If the system is working normally, the neighborhood graph of each sensor is almost invariant against the fluctuations of experimental conditions. Here a neighborhood graph is defined based on the correlation between sensor signals. With the notion of stochastic neighborhood, our method is capable of robustly computing the anomaly score of each sensor under conditions that are hard to be detected by other naive methods.

Dot pattern generation technique using molecular dynamics

We have developed a new technique for generating homogeneously distributed irregular dot patterns useful for optical devices and digital halftoning technologies. To introduce irregularity, we use elaborately designed sequences called low-discrepancy sequences instead of pseudorandom numbers. We also use a molecular-dynamics redistribution method to improve the distribution of dots. Our method can produce arbitrary density distributions in accordance with a given design. The generated patterns are free from visible roughness as well as any moiré patterns when superimposed on other regular patterns. We demonstrate that our method effectively improves luminance uniformity and eliminates moiré patterns when used for a backlight unit of a liquid-crystal display.

Unsupervised change analysis using supervised learning

We propose a formulation of a new problem, which we call change analysis, and a novel method for solving the problem. In contrast to the existing methods of change (or outlier) detection, the goal of change analysis goes beyond detecting whether or not any changes exist. Its ultimate goal is to find the explanation of the changes.While change analysis falls in the category of unsupervised learning in nature, we propose a novel approach based on supervised learning to achieve the goal. The key idea is to use a supervised classifier for interpreting the changes. A classifier should be able to discriminate between the two data sets if they actually come from two different data sources. In other words, we use a hypothetical label to train the supervised learner, and exploit the learner for interpreting the change. Experimental results using real data show the proposed approach is promising in change analysis as well as concept drift analysis.

X10-based massive parallel large-scale traffic flow simulation

Optimizing city transportation for smarter cities can have a major impact on the quality of life in urban areas in terms of economic merits and low environmental load. In many cities of the world, transport authorities are facing common challenges such as worsening congestion, insufficient transport infrastructure, increasing carbon emissions, and growing customer needs. To tackle these challenges, it is highly necessary to have fine-grained and large-scale agent simulation for designing smarter cities. In this paper we propose a large-scale traffic simulation platform built on top of X10, a new distributed and parallel programming language. Experimental results demonstrate linear scalable performance in simulating large-scale traffic flows of the national Japanese road network and a hundred of cities of the world using thousands of CPU cores.

Why does subsequence time-series clustering produce sine waves?

The data mining and machine learning communities were surprised when Keogh et al. (2003) pointed out that the k-means cluster centers in subsequence time-series clustering become sinusoidal pseudo-patterns for almost all kinds of input time-series data. Understanding this mechanism is an important open problem in data mining. Our new theoretical approach (based on spectral clustering and translational symmetry) explains why the cluster centers of k-means naturally tend to form sinusoidal patterns.

A Model Study on Cluster Size Effects of Resonant X-Ray Emission Spectra.

Cluster size dependence of X-ray absorption spectra (XAS), X-ray photoemission spectra (XPS), and resonant X-ray emission spectra (RXES) are theoretically studied with a one-dimensional d - p model, which describes qualitatively effects of translational symmetry for nominally d 0 (or f 0 ) compounds such as TiO 2 (CeO 2 ). It is shown that RXES depends more sensitively on the cluster size than XAS and XPS, so that RXES is a useful probe in studying the duality between itinerant and localized characters of 3 d or 4 f electrons. From results calculated by changing the cluster size and parameter values such as p - d hybridization strength, d - d Coulomb interaction etc., it is explained why the experimental Ce 4 f -3 d RXES of CeO 2 is well reproduced by calculations with a single-cation impurity Anderson model, but the Ti 3 d -2 p RXES of TiO 2 is not well reproduced.

Toward simulating entire cities with behavioral models of traffic

Resilient transportation systems enable quick evacuation, rescue, distribution of relief supplies, and other activities for reducing the impact of natural disasters and for accelerating the recovery from them. The resilience of a transportation system largely relies on the decisions made during a natural disaster. We developed an agent-based traffic simulator for predicting the results of potential actions taken with respect to the transportation system to quickly make appropriate decisions. For realistic simulation, we govern the behavior of individual drivers of vehicles with foundational principles learned from probe-car data. For example, we used the probe-car data to estimate the personality of individual drivers of vehicles in selecting their routes, taking into account various metrics of routes such as travel time, travel distance, and the number of turns. This behavioral model, which was constructed from actual data, constitutes a special feature of our simulator. We built this simulator using the X10 language, which enables massively parallel execution for simulating traffic in a large metropolitan area. We report the use cases of the simulator in three major cities in the context of disaster recovery and resilient transportation.

A novel dot‐pattern generation to improve luminance uniformity of LCD backlight

We report on a novel theoretical approach to generate irregular dot patterns, providing an integrated solution to difficulties peculiar to collimated -type backlight units. By applying this technology to a light guide and a diffuser film, the