コロキアムB発表

日時: 6月19日(水)3限(13:30~15:00)


会場: L1

司会: Doudou Fall
SUMAILA NIGO M, 2回目発表 知能コミュニケーション 中村 哲☆, 松本 裕治, 荒牧 英治, 若宮 翔子
Title:Web-based Epidemic and Pandemic-Prone Diseases Surveillance using Google Search Volume
Abstract:People leave traces about their wellbeing on the Internet, and these traces can be captured and used to derive actionable information, one of such is web-based disease surveillance. Existing methods toward web-based disease surveillance fail in certain contexts, e.g., in places with highly biased data, such as in developing countries. We propose a new approach based on Google Search Volume data that can produce reliable results in multiple contexts.
Language of the presentation: English
 
寺西 裕紀 D, 中間発表 自然言語処理学 松本 裕治, 中村 哲, 新保 仁, 進藤 裕之
title: Decomposed Local Models for Coordinate Structure Parsing
abstract: We propose a simple and accurate model for coordination boundary identification. Our model decomposes the task into three subtasks during training; finding a coordinator, identifying inside boundaries of a pair of conjuncts, and selecting outside boundaries of it. For inference, we make use of probabilities of coordinators and conjuncts in the CKY parsing to find the optimal combination of coordinate structures. Experimental results demonstrate that our model achieves state-of-the-art results, ensuring that the global structure of coordinations is consistent.
language of the presentation: Japanese
 
TRAN VAN HIEN M, 2回目発表 自然言語処理学 松本 裕治, 中村 哲一, 新保 仁, 進藤 裕之
title: *** Relation Classification Using Segmen-Level Attention-based CNN and Dependency-based RNN ***
abstract: *** Recently, relation classification has gained much success by exploiting deep neural networks. In this work, we propose a new mode effectively combining Segment-level Attention-based Convolutional Neural Networks (SACNNs) and Dependency-based Recurrent Neural Networks (DepRNNs). While SACNNs allow the model to selectively focus on the important information segment from the raw sequence, DepRNNs help to handle the long-distance relations from the shortest dependency path of the related entities. Experiments on the SemEval-2010 Task 8 dataset show that our model is comparable to the state-of-the-art without using any external lexical features. ***
language of the presentation: *** English ***
 
辰巳 守祐 M, 2回目発表 自然言語処理学 松本 裕治, 中村 哲一, 新保 仁, 進藤 裕之
title: Chemical domain Unsupervised Named Entity Recoginition (NER)
abstract: Unlike general domains, creating chemical domain corpus requires specialized knowledge. Therefore, it is expensive to create a corpus. We propose an unsupervised NER that does not require a manual annotation corpus. Specifically, we construct the NER system using the chemical dictionary and a large amount of scientific papers in the Distant supervision style.
language of the presentation: Japanese
 

会場: L2

司会: 藤本 大介
MAIPRADIT RUNGROJ M, 2回目発表 ソフトウェア工学 松本 健一, 笠原 正治, 石尾 隆, 畑 秀明
Title: Software Document Classification Using N-gram IDF and Automated Machine Learning

Abstract: In software development, there are many documents regarding software engineering activities. It’s a time-consuming task to manually classify the type of documents based on their contents, in order to overcome time-consuming problems an automated tool is required. We proposed a framework to automatically classify software document using N-gram IDF and Auto-sklearn. In this research, we apply the framework on two problems, identifying “on-hold” self-admitted technical debt and sentiment classification. In the first problem, our framework achieved F1 score of 0.73 compared to naive baseline which has 0.31 score. In the second problem, our framework achieved the highest f1 score in positive and negative sentences which is comparable with well-known sentiment classification tools. The result shows that our framework performs better working than traditional sentiment classification tools. In both cases our framework outperforms baselines.

Language of the presentation: English
 
MEI WENJIE M, 2回目発表 知能システム制御 杉本 謙二, 笠原 正治, 松原 崇充, 小蔵 正輝
title:Analysis and Control of Markov Jump Linear System
abstract:In this study, we are concerned with the stability property of continuous-time Markov jump linear systems, as well as H_2/H_{\infty} control for Markov jump linear systems with observation and state delay. We proposed a new criterion for analyzing the mean stability of Markov jump linear systems, and a mixed H_2/H_{\infty} controller for H_2/H_{\infty} control of Markov jump linear systems. We confirm the effectiveness of the proposed methods by numerical simulations.
language of the presentation: English
 
ZHU LINGWEI M, 2回目発表 ロボットラーニング 杉本 謙二, 小笠原 司, 松原 崇充
title:Scalable Reinforcement Learning for Vinyl Acetate Monomer Process
abstract: Chemical processes comprising a large number of components are challenging to control. A complex plant requires extensive knowledge from plant engineers and operators. However, training experts can be expensive both in terms of time and money, especially facing the problem of aging population. In this research, we leverage model-free Reinforcement Learning which is a branch of machine learning that learns via trial-and-error autonomously. By using RL, we achieve comparable performance to that of model-based performance operated by plant engineers, showing RL as a potential approach for autonomous control.
language of the presentation: English
 
CHEN ZHENG M, 2回目発表 インタラクティブメディア設計学 加藤 博一, 佐藤 嘉伸, 神原 誠之, Alexander Plopski
Title: Comparison of Image Similarity Measurements for Vertebrae Pose Estimation
Abstract: Image registration is a key step in medical imaging analysis, as it allows single or multiple images to be registered together in a common world coordinate system to acquire detail information from a consistent 3D volume. Pose estimation for medical imaging analysis is the problem of determining the transformation of an object in a 2D image which gives the 3D object. For our previous research, it uses image similarity method to compare the two of the image models and accomplish pose estimation after image registration. However, we need more accurate results than ones by existing methods and how to find the global optimum result is still a challenge. Therefore, my research is to compare different image similarity measurement methods and find an accurate method. Currently, I choose three similarity measurement methods ZNCC, Mutual Information and Structural Similarity, and try to make conclusion between these similarity measurement methods.
Language of the presentation: English