| 芥子 育雄 | 1561201: D, 中間発表 | 知能コミュニケーション 中村 哲, 松本 裕治, 鈴木 優, 吉野 幸一郎 |
|
title: Semantically Readable Distributed Representation Learning and Its Expandability Using a Word Semantic Vector Dictionary
abstract: Distributed representations named word2vec and paragraph vectors, computed using simple neural networks with context information as features, have been widely used. The practical problem is that a large collection of documents is required for overcoming the sparsity of words. Also, the meaning of the features is difficult to understand. In contrast to the distributed representations obtained by learning, we proposed a word semantic vector dictionary, constructed using a human expert with context information as features. First, we proposed a method that encodes knowledge in an encyclopedia using the word semantic vector dictionary and associative image retrieval. Second, we proposed an integration method to learn words expanded using the dictionary with paragraph vector to solve the problem of word sparsity in Twitter and reputation information extraction from Twitter. The integration showed that the accuracy of sentiment analysis improves by learning context information even if words are sparse. Finally, we have proposed a method that gives a specific meaning to each node of a hidden layer by introducing the dictionary into the initial weights and by using paragraph vector models. We determined the readability in a user test. A total of 52.4% of the top five weighted hidden nodes were related to tweets. For the expandability of the method, we constructed a diverse sentiment analysis benchmark and enhanced the dictionary for the purpose of distributed representations. We also conducted a word similarity task using a Wikipedia corpus to test the domain-independence of the method. We found the objective and subjective evaluation support each hidden node maintaining a specific meaning. Thus, our method succeeds in improving readability. language of the presentation: Japanese | ||
| DO QUOC TRUONG | 1561205: D, 中間発表 | 知能コミュニケーション 中村 哲, 松本 裕治, Sakriani Sakti |
|
title: Sequence-to-sequence approaches for emphasis speech translation
abstract: Emphasis is an important piece of paralinguistic information that is used to express different intentions, attitudes, or convey emotion A recent work has tried to tackle this task by developing a method for mapping emphasis between languages utilizing conditional random fields (CRFs). Although CRFs allow for consideration of rich features and local context, they have difficulty in handling continuous variables, and cannot capture long-distance dependencies easily. In this paper, we proposed approaches for emphasis speech translation using sequence-to-sequence models that jointly trains and predicts words and emphasis in a unified architecture based on sequence-to-sequence models. The proposed model not only speeds up the translation pipeline but also allows us to perform joint training. Our experiments on the emphasis and word translation tasks showed that we could achieve comparable performance for both tasks compared with previous approaches while eliminating complex dependencies language of the presentation: English | ||
| ZHU ZEYU | 1651207: M, 1回目発表 | ネットワークシステム学 岡田 実, 林 雄一, 東野 武史, Duong Quang Thang |
|
title: *** Deployment Position Evaluation of 2-by-2 LCX-MIMO System over Linear Corridor Environment ***
abstract: *** Leaky coaxial (LCX) cables can be used as antennas and are widely applied in linear cell environment for wireless communication systems. A novel method proposed to use single LCX cable to achieve MIMO system. In this research, we confirmed the capability of proposed LCX cable and did some evaluation about the performance of systems deploying the LCX cable in different positions by measuring the throughput *** language of the presentation: *** English *** | ||
| ZHANG HAO | 1651212: M, 1回目発表 | ネットワークシステム学 岡田 実, 杉本 謙二, 東野 武史, Duong Quang Thang |
|
Title: 2-by-2 LCX-MIMO System in Real Indoor Linear-cell Environment
Abstract: Leaky coaxial (LCX) cable is widely deployed in Linear-cell environments. To improve the channel capacity, LCX cable can also build multiple-input-multiple-output (MIMO) system. In the past, one LCX cable was only treated as one antenna. Our study is focusing on that one LCX cable is treated as two antennas for increasing the efficiency and reducing the interference. This proposed LCX-MIMO can almost double the throughput comparing with traditional SISO system. In addition, we designed some experiments for confirming the feasibility of this proposal in the real world. Language of the presentation: English | ||
| 丸山 和輝 | 1551104: M, 2回目発表 | ネットワークシステム学 岡田 実, 佐藤 嘉伸, 東野 武史, 侯 亜飛, Duong Quang Thang |
| QIU JIE | 1651210: M, 1回目発表 | ロボティクス 小笠原 司 ☆ |
< > title: Person Re-idnetification via Temporal-enhanced Recurrent Convolutional Neural Networks abstract: As the number of surveillance cameras increases, we are able to use the data captured by these cameras to do some complicated tasks, such as looking for missing kids. Person reidentification is such kind of task which aims at finding specific person among gallery images. However, the changes of camera setting, appearance other noise are still big challenges. In this presentation, we introduce a novel method, which enhances the temporal information across time axis in order to capture to time related information like walking pattern for recurrent convolutional neural networks. The result shows that we outperform current state-of-the-art results for two commom person re-identification database. language of the presentation: *** English *** | ||