A5034077980- Natural Language Processing Techniques
- Image Retrieval and Classification Techniques
- Semantic Web and Ontologies
- Tropical and Extratropical Cyclones Research
- Remote-Sensing Image Classification
- Research Data Management Practices
- Handwritten Text Recognition Techniques
- Geographic Information Systems Studies
- Scientific Computing and Data Management
- Cell Image Analysis Techniques
- Image Processing and 3D Reconstruction
- Bayesian Methods and Mixture Models
- Topic Modeling
- Web Data Mining and Analysis
- Gene expression and cancer classification
- Generative Adversarial Networks and Image Synthesis
- Remote Sensing and Land Use
- Computer Graphics and Visualization Techniques
- Advanced Computational Techniques and Applications
- Ocean Waves and Remote Sensing
- Advanced Vision and Imaging
- Aesthetic Perception and Analysis
- Molecular Biology Techniques and Applications
- Digital Humanities and Scholarship
- Public Relations and Crisis Communication
Research Organization of Information and Systems
2010-2023
National Institute of Informatics
2012-2021
Open Data Institute
2018-2019
The Open University of Japan
2019
Hitotsubashi University
2014
Japan Science and Technology Agency
2012-2013
Data61
2013
University of Tehran
2011
Dynamic Systems & Research (United States)
2005
Otsuka (Japan)
1999
Much of machine learning research focuses on producing models which perform well benchmark tasks, in turn improving our understanding the challenges associated with those tasks. From perspective ML researchers, content task itself is largely irrelevant, and thus there have increasingly been calls for tasks to more heavily focus problems are social or cultural relevance. In this work, we introduce Kuzushiji-MNIST, a dataset Kuzushiji (cursive Japanese), as two larger, challenging datasets,...
Kuzushiji, a cursive writing style, had been used in Japan for over thousand years starting from the 8th century. Over 3 millions books on diverse array of topics, such as literature, science, mathematics and even cooking are preserved. However, following change to Japanese system 1900, Kuzushiji has not included regular school curricula. Therefore, most natives nowadays cannot read written or printed just 150 ago. Museums libraries have invested great deal effort into creating digital...
Abstract In this report, we propose a deep learning technique for high-accuracy estimation of the intensity class typhoon from single satellite image, by incorporating meteorological domain knowledge. By using Visual Geometric Group’s model, VGG-16, with images preprocessed fisheye distortion, which enhances typhoon’s eye, eyewall, and cloud distribution, achieved much higher classification accuracy than that previous study, even sequential-split validation. Through comparison t-distributed...
The spread of dengue fever depends mainly on the availability favourable breeding sites for its mosquito vectors around human dwellings.To investigate if various factors influencing habitats can be mapped from space, indices, such as container index, house index and Breteau were calculated Ministry Public health data collected three times annually in Phitsanulok, Thailand between 2009 2011.The most influential found to temperature, humidity, rainfall, population density, elevation land...
Abstract Kuzushiji, a cursive writing style, had been used in Japan for over thousand years starting from the eighth century. Over 3 million books on diverse array of topics, such as literature, science, mathematics and even cooking are preserved. However, following change to Japanese system 1900, Kuzushiji has not included regular school curricula. Therefore, most natives nowadays cannot read written or printed just 150 ago. Museums libraries have invested great deal effort into creating...
Weather is a typical topic of daily conversations, so it natural idea to use social data observe weather. Geotagging key for weather applications because highly localized phenomenon on the earth. Hence we developed software called GeoNLP toponym-based geotagging, and applied Twitter stream find toponyms (place names) from Japanese tweets talking about precipitation events. We observed that less than 10 percent contain toponym information, but can capture events each place. also show temporal...
Recognition of historical documents is a challenging problem due to the noised, damaged characters, and background. However, in Japanese documents, not only contains mentioned problems, pre-modern characters were written cursive are connected. Therefore, character segmentation-based methods do work well. This leads idea creating new recognition system. In this paper, we propose human-inspired document reading system recognize multiple lines documents. During reading, people employ eyes...
Abstract Stock market prediction has always been a difficult process, most of the rely solely on data corresponding stock market. Relationship gold and oil price with performance proven significant in some major world index. Prediction index using machine learning methods is expected to perform well, ability method predict nonlinear inputs. The were commonly able relatively well predicting values Japanese Nikkei 225 400 indexes.
ou non, émanant des établissements d'enseignement et de recherche français étrangers, laboratoires publics privés.
Despite frequent use of digital devices in everyday life, cost-effective measurement public health issues urban areas is still challenging. This study was, therefore, planned to extract land-use types using object-based and spatial metric approaches explore the dengue incidence relation surrounding environment near real-time Google Advanced Land Observation Satellite images. The characterised image showed useful classification an area with 77% accuracy 0.68 kappa. Geospatial analysis on data...
From classifying handwritten digits to generating strings of text, the datasets which have received long-time focus from machine learning community vary greatly in their subject matter. This has motivated a renewed interest building are socially and culturally relevant, so that algorithmic research may more direct immediate impact on society. One such area is history humanities, where better relevant models can accelerate across various fields. To this end, newly released benchmarks been...
An attributed relational graph (ARG) is introduced into our NOAA satellite image database system. The node and the branch of an ARG denotes a classified region spatial relationship between adjacent regions, respectively. Furthermore, few attributes node/branch help to express numerical shape features regions. Similarity retrieval thereby turns out be equivalent matching. similarity process system as follows: (1) select visual example query generate its structure, (2) calculate optimal...
This paper describes the key technologies used in a 1-Gb synchronous DRAM. DRAM was developed according to new cell-operating concept which ground-level (V/sub ss/) precharged bit line with negative word-line reset scheme enables nonboosted 2.1-V architecture. Total power consumption is less than that of conventional half-V/sub cc/ bit-line scheme. We also propose vernier-type, high-accuracy delay-locked-loop circuit realizing /spl plusmn/20-ps quantization errors for clock recovery and skew...
[1] We reproduce a magnetospheric reconfiguration under southward interplanetary magnetic field (IMF) conditions using numerical magnetohydrodynamic simulations. To investigate the relative displacements of geomagnetic conjugate points, we trace both footprints lines during positive and negative IMF By conditions. Several substormlike features, namely, formation near-Earth neutral line, fast earthward flow, tailward releases plasmoid, occur about 1 h after turning IMF. The surveyed line...
The development of data and model integration platforms has furthered scientific inquiry helped to solve pressing social environmental problems. While several e-infrastructure have been developed, the concept remains obscure, these produced few firm results. This article investigates on Data Integration Analysis System (DIAS) platform, using three case projects from water-related fields. We provide concrete examples by analyzing transfer analysis process, demonstrate what platform functions...
Following the increasing adoption of online social networks, users those services can now easily collect and transmit various types data instantaneously from real-world locations (e.g., geo-referenced pictures, videos, messages, etc.). This paper argues that one take advantage information is spread throughout networks like Twitter, in order to provide a realistic solution for situational awareness be useful decision makers during particular events. Through generation English captions...
In this study, we have developed an unstructured normal mode decomposition solver based on the finite element method for purpose of analysis natural basins. The can be used to provide site-specific storm hazard summary such as risk a given basin. This information useful guideline during events. two past events using modes derived from has been demonstrated in different locations. first event is tropical cyclone Linda (1997) Gulf Thailand and second one Melor (2009) Ise Bay, Japan. reveals...