| 奥田 裕樹 | 1351022: M, 2回目発表 | 金谷 重彦, | 池田 和司, | MD.ALTAF-UL-AMIN, | 杉浦 忠男, | 小野 直亮 |
|
title: Clustering based differential gene expression analysis of secondary metabolic pathways in Euglena gracilis
abstract: Euglena gracilis, a unicellular protist that have photosynthesizing chloroplasts produces wax esters under anaerobic condition, many of which can be used as biofuel. De novo transcriptome analysis has become one of the main approaches for identifying the differentially expressed genes for various environments and understanding pathways of secondary metabolites in non-model organisms. In this study, transcriptome profiles from neutral and bleached Euglena gracilis strains with aerobic and anaerobic condition were investigated using massive parallel sequencing and we detected candidate genes potentially involved in the production of secondary metabolites by using clustering based on statistical test. language of the presentation: Japanese | ||||||
| 白石 磨貴男 | 1351058: M, 2回目発表 | 金谷 重彦, | 池田 和司, | MD.ALTAF-UL-AMIN, | 杉浦 忠男, | 小野 直亮 |
|
title: Bioinformatic Interpretation of Physicochemical Features in the Electron Transport Chain of Microorganisms by Protein Structures
abstract: The powerplant of a cell is the respiratory chain, an attribute of energy metabolism. This has long been studied, but its origin and contribution in evolutionary history of organisms is still a subject of discussion. Moreover, most of these studies focus on individual species and the comprehensive research of various microorganisms have been mostly ignored despite its evolutionary importance. Thus we study on this subject comprehensively by comparing reported measurements of the electron transport chain among molecular phylogenetic groups in bacteria and archaea. We found that phylogenetic clusters of proteins related with the electron transport chain can be characterized according to physicochemical properties. This result suggests effects of environmental constrains on the phylogenetic clusters. language of the presentation: Japanese | ||||||
| 近藤 雅芳 | 1351044: M, 2回目発表 | 松本 裕治☆６, | 池田 和司, | 山田 武士, | 新保 仁 | |
|
title: Visualizing Text Data in view of Lexical Ambiguity abstract: On this presentation, We propose the new approach of visualizing text data to consider word semantic differences and show the experience each to consider them or not with small text data. Our method assumes that both documents and words have latent coordinates in a two-demensional Euclidean space, or visualization space, and also words have the latent parameters which control the semantic diffrence for each words. Moreover, same word but different meanings has each coordinates depending on the number of its diffenrece meanings. And then documents and words is visualized by considering a generating process of documents as a mapping from the visualization space into the space of documents. A visualization can be obtained by fitting our model to text data using the Variational Bayesian Inference. In the experiments, we show that the proposed model can locate related documents closer together and well-matched words around them. language of the presentation: Japanese | ||||||
| 吉川 友也 | 1361013: D, 中間発表 | 松本 裕治☆６, | 池田 和司, | 山田 武士, | 新保 仁 | |
|
title: Latent Support Measure Machines for Bag-of-Words Data Classification
abstract: In many classification problems, the input is represented as a set of features, e.g. bag-of-words (BoW) representation of documents. Support vector machines (SVMs) are widely used tools for such classification problems. The performance of the SVMs is generally determined by whether kernel values between data points can be defined properly. However, SVMs for BoW representation have a major weakness that the co-occurrence of different but semantically similar words cannot be reflected in the kernel calculation. To overcome the weakness, this talk proposes a kernel-based discriminative classifier for BoW data, which is called the latent support measure machine (latent SMM). With the latent SMM, a latent vector is associated with each vocabulary term, and each document is represented as a distribution of the latent vectors for words appearing in the document. To represent the distributions efficiently, we use the framework of kernel embeddings that holds high order moment information of distributions. Then the latent SMM finds a separating hyperplane that maximizes margins between distributions of different classes while estimating latent vectors for words so as to improve the classification performance. The experiments show that the latent SMM achieves state-of-the-art accuracy on BoW text classification and is robust for its own hyper-parameters. language of the presentation: Japanese | ||||||
| 高木 隆志 | 1351063: M, 2回目発表 | 井上 美智子, | 中島 康彦, | 米田 友和, | 大和 勇太 |
|
title: Resonance noise reduction using variable clock frequency for high quality at-speed scan testing
abstract: Excessive power supply noise (PSN) during scan testing has become a serious issue in industry since it may cause undue yield losses. The type of PSN can be mainly classified as resistive (IR) noise, inductive (L*di/dt) noise, and resonance noise.Though various methods have been previously proposed to solve this issue, most of them only targets resistive or inductive noise. This work tackles the resonance noise issue by adjusting the clock frequency during shift operation to improve the quality of at-speed scan testing. language of the presentation: Japanese | |||||
| PHANNACHITTA PASSAKORN | 1361021: D, 中間発表 | 松本 健一, | 中島 康彦, | 門田 暁人 | |
|
title:
Analogy-P: Providing Probabilistic Inference for improving Analogy-based Software Effort Estimation
abstract: The performance of software effort estimation based on analogy reasoning (ABE) heavily relies upon (i) the measures that specify the similarity between software projects, and (ii) the adaptation methods that systematically derive solution from the retrieved similar projects. To date, the adaptation methods have been well-explored because of an ability to exploit a reliable estimation over the available data sets which mostly consists of less than 100 project cases. However, these methods are still long way to present a good accuracy for large data sets such as a well-known 499 project cases data set called China data set. My assumption towards the issue is a lack of feasible methods to reliably determine the similarity between project cases, when there are many projects to discuss. My Ph.D. dissertation is to introduce a utilizing of probabilistic inference between independent variables and the effort to improve the software project retrieval process called Analogy-P. The probabilistic inference is constructed from Learning-to-rank techniques, which have been well-perceived and continually developed in many other research domains such as information retrieval, where ranking is the principal problem. My highlighted preliminarily result from the approach using the Desharnais and China data set, showing that when sampling the probabilistic influence using Bayes' theorem over an optimized number of clusters of continuous variable, both data sets can achieve more accurate estimation than ABE with sophisticate adaptation methods, even though; the other optimization parameters in the Analogy-P experiment were set as standard ABE method. language of the presentation: English | |||||
| 湯月 亮平 | 1351110: M, 2回目発表 | 松本 健一, | 飯田 元, | 門田 暁人, | 畑 秀明 |
|
title: An Empirical Study of Fine-Grained Conflict Resolution in Merging abstract: Branching and merging are common activities in large-scale software development projects. Isolated development with branching enables developers to focus their effort on their specific tasks without bothering negative impact by other developers' changes. After the completion of tasks in branches, such branches should be integrated into common branches by merging. When conflicts occur in merging, developers need to resolve the conflicts, which are troublesome. To support conflict resolution in merging, we aim to understand how conflicts are resolved in practice from large-scale study. In this presentation, I introduce our techniques for identifying conflicts and detecting conflict resolution in method level and result which we obtain. language of the presentation:Japanese | |||||