【受賞／表彰等】鳥海研究室の菊田俊平さん(M2)がComplex Networks 2019においてBest oral presentation awardを受賞しました。
==受賞日==
2019年12月12日
==日本語==
1.氏名 ：菊田 俊平
2.学部・研究科、学科（類・課程）・専攻等 と 学年
工学系研究科システム創成学専攻 鳥海研究室 修士2年
3.受賞した賞の名称と簡単な説明
Best oral presentation award
オーラルセッションにて最も優れたプレゼンテーションをした者に送られる賞。
4.受賞された研究・活動について
ネットワークにおける役割発見は、ノードの構造的な特徴に基づいて、ノードを構造的に類似したノード群に分類することを目的としたタスクです。
従来は教師なし学習が主流でしたが、教師情報が欠けていることで、結果の検証、手法の比較が難しいのが問題でした。
そこで転移学習に注目し、高い精度で別のネットワークから役割を転移させる手法を提案しました。
特に、ドメイン敵対的学習を用いて、二つの分布を類似させる学習を行い、ドメインシフト問題を緩和しました。
計算実験によって、従来手法よりも高い精度で役割が発見できることを示しました。
また、ドメインに依存しない特徴量生成によって精度が向上すること、ネットワークの大きさが異なっている場合でも、高い精度を維持できること、複雑な構造が混ざったグラフにおいても適用可能であることを示しました。
5.今後の抱負・感想
このような素晴らしい賞をいただき、大変光栄です。
これからも日々精進してまいります。
【Awards and Commendations】Shumpei Kikuta(M2), Toriumi labo, received Best oral presentation award at Complex Networks 2019
1. Name：Shumpei Kikuta
2. Faculty/Graduate School, Department (Stream / Program) /Major
Year:(B4, M1, M2, D1, D2, D3)
M2, Department of System Innovation,School of Engineering,Toriumi Laboratory
3.Name of award and short explanation about the award
Best oral presentation award
This award is sent to the researcher who made the best presentation.
4.About awarded research
Discovering the roles of nodes in a network is important for solving various social issues.
Role discovery aims to infer nodes’ roles from a network structure, and it has received considerable attention recently.
The conventional methods of role discovery mainly use unsupervised learning, but due to the lack of information, it is difficult to discover the roles we want or to ascertain the results.
In this paper, we attempt to improve accuracy through using supervised information.
Specifically, we adopt transfer learning using adversarial learning.
As a result of computational experiments, we show that the proposed model discovers a node’s role more effectively than do the conventional methods.
Furthermore, we found that domain-invariant features lead to higher accuracy, the proposed method discovers roles better even with different network sizes, and the proposed method works well even if the networks have nodes of various structures.
5.Your impression & future plan
I am honored to win such a wonderful award.
I will make an effort to improve this research to apply the proposed method to real-world datasets.
【Awards and Commendations】Shumpei Kikuta(M2), Toriumi labo, received Best oral presentation award at Complex Networks 2019
1. Name：Shumpei Kikuta
2. Faculty/Graduate School, Department (Stream / Program) /Major
Year:(B4, M1, M2, D1, D2, D3)
M2, Department of System Innovation,School of Engineering,Toriumi Laboratory
3.Name of award and short explanation about the award
Best oral presentation award
This award is sent to the researcher who made the best presentation.
4.About awarded research
Discovering the roles of nodes in a network is important for solving various social issues.
Role discovery aims to infer nodes’ roles from a network structure, and it has received considerable attention recently.
The conventional methods of role discovery mainly use unsupervised learning, but due to the lack of information, it is difficult to discover the roles we want or to ascertain the results.
In this paper, we attempt to improve accuracy through using supervised information.
Specifically, we adopt transfer learning using adversarial learning.
As a result of computational experiments, we show that the proposed model discovers a node’s role more effectively than do the conventional methods.
Furthermore, we found that domain-invariant features lead to higher accuracy, the proposed method discovers roles better even with different network sizes, and the proposed method works well even if the networks have nodes of various structures.
5.Your impression & future plan
I am honored to win such a wonderful award.
I will make an effort to improve this research to apply the proposed method to real-world datasets.
