スライド 1

Size: px
Start display at page:

Download "スライド 1"

Transcription

1 CBI 学会 2018 年大会フォーカストセッション FS-10 計算毒性学における人工知能 そして進化する 化学データサイエンス への展開 株式会社インシリコデータ 湯田浩太郎

2 ケモメトリックス (chemometrics): 計量化学 ( 化学計量学 ) とは 数理科学 統計学 機械学習 パターン認識 データマイニングなどの手法により ( 広義の ) 化学分野における諸問題を解決しようとする分野である ケモインフォマティクス (cheminformatics): 化学情報学は コンピュータと情報化技術を用いて化学領域の問題に適用する方法論である ケモデータサイエンス 化学データサイエンス (chemodatascience, chemical datascience): 化学データサイエンスとは 化学分野の諸問題についてデータサイエンス手法を適用して解決を目指す学問分野である

3 講演討論内容 1. 計算毒性学における人工知能の歴史 2. データサイエンスと IT 技術進歩に伴う変化 1 データサイエンスの構成技術 2 コンピューターの進歩によるデータサイエンスの変化 3 データサイエンスから 化学データサイエンス へ 3. 機械学習型人工知能適用上での解決すべき点 4. 次世代型データ解析手法開発の必要性 1KY 法の開発 2 テーラーメードモデリングの開発 7. 化学ビッグデータの構築上での問題点 8. まとめ

4 1. 計算毒性学における人工知能の歴史 1.1 計算毒性学 ( 安全性評価 ) 分野での 二大適用技術 多変量解析パターン認識 人工知能 化学 ( 化合物 )

5 1. 計算毒性学における人工知能の歴史 計算毒性学分野では人工知能が適用されてきた 当初から現在まで ルールベース型人工知能 当初より現在まで人工知能が適用され 実用システムや研究システムが多数開発済み 現在注目中機械学習型 ( ニューラルネットワーク ) 人工知能 機械学習型人工知能 システムとしての実績はないが 今後の展開が期待されている

6 現在展開されている毒性予測関連システム HazardExpert: CompuDrug Ltd. RIPT: Rule Induction for Predictive Toxicology DEREK: Deductive Estimation of Risk from Existing Knowledge:Lhasa Ltd. 代謝予測関連支援システム ( 人工知能 ) COMPACT,MetaSite,SPORCalc, ENO,MetabolExpert,META,Meteor 生分解性予測関連支援システム ( 人工知能 ) BESS,CATABOL,UMBBD, PPS,META

7 計算毒性学における人工知能 機械学習型人工知能の全体的歴史と主要トピックス 人工知能に注目させたトピックス : 成功事例 人工知能に注目させたトピックス : 失敗事例 人工知能の歴史 過古の多変量解析 / パターン認識と人工知能との関係 現在における多変量解析 / パターン認識と人工知能との関係

8 人工知能に注目させたトピックス 成功事例 AlphaGo( アルファ碁 ) が人間に打ち勝って世界一になった人間がコンピュータに勝てる最後の分野の神話が崩れた 碁の学習アルゴリズムに深層学習を適用していた 失敗事例 留意点 : 学習に使われた対局数が数千万という数に達している人工知能のチャットボット Tay がヘイト発言を乱発したサンプル数が少ない場合は成果を期待しにくい Twitter 等の内容からヘイト発言を学習してしまった

9 人工知能に注目させたトピックス 成功事例 AlphaGo( アルファ碁 ) が人間に打ち勝って世界一になった人間がコンピュータに勝てる最後の分野の神話が崩れた 留意点 : 人工知能における学習サンプルの重要性学習内容により人工知能は大きく変化する碁の学習アルゴリズムに深層学習を適用していた 失敗事例 人工知能のチャットボット Tay がヘイト発言を乱発した Twitter 等の内容からヘイト発言を学習してしまった

10 人工知能の歴史 人工知能言語 1958:LISP(List Processor) 1972:Prolog 1994:Python ルールベース型人工知能 * 第五世代コンピュータ ( 日本 ) 多変量解析 / パターン認識 重回帰 パーセプトロン PCA クラスタリング 他 機械学習発展 新アプローチニューラルネットワーク 遺伝的アルゴリズム ファジイ 新世代人工知能 深層学習開発 / 展開

11 過古の多変量解析 / パターン認識と人工知能との関係 多変量解析 / パターン認識と人工知能は互いに独立している 多変量解析 / パターン認識 様々なデータ解析手法 人工知能 クラス分類フィッティングマッピングクラスタリング他 既存のルール ルールベース型

12 現在における多変量解析 / パターン認識と人工知能との関係 多変量解析 / パターン認識と人工知能は 機械学習により繋がっている 多変量解析 / パターン認識 人工知能 様々なデータ解析手法 クラス分類フィッティングマッピングクラスタリング他 ビッグデータ 機械学習 既存のルール ( 特化型 汎用型 ) 発見された新ルール 深層学習 (DNN, CNN,RNN)

13 ディープラーニングの種類 深層学習 ( deep learning) 法 畳み込みニューラルネットワーク : CNN(Convolutional Neural Network) 画像 動画認識やレコメンダシステム 自然言語処理等 再帰型ニューラルネットワーク : RNN(Recurrent Neural Network) 音声の波形 動画 文章等の時系列データを扱う 敵対的生成モデル : GAN(Generative Adversarial Network) 自分で学習モデルを生成し それを評価し修正することで 自動学習するネットワークモデル

14 化学分野での人工知能の歴史と種類 二種類の人工知能 知識整理および適用型 ルールベース型人工知能 解決すべき問題点 : 目的解決に適したルール作成 ルール間の階層 衝突回避 エキスパートの存在必要 発見型および要因解析型 機械学習型人工知能ニューラルネットワーク深層学習 解決すべき問題点 : データ解析上の問題点過剰適合 偶然相関 クラス分布 欠損データ 他 解析手法の特性 / 限界 解析結果の解釈

15 講演討論内容 1. 計算毒性学における人工知能の歴史 2. データサイエンスと IT 技術進歩に伴う変化 1 データサイエンスの構成技術 2 コンピューターの進歩によるデータサイエンスの変化 3 データサイエンスから 化学データサイエンス へ 3. 機械学習型人工知能適用上での解決すべき点 4. 次世代型データ解析手法開発の必要性 1KY 法の開発 2 テーラーメードモデリングの開発 7. 化学ビッグデータの構築上での問題点 8. まとめ

16 2. データサイエンスと IT 技術進歩に伴う変化 1 データサイエンスの構成技術 データ解析手法 データベース 解析目的 多種多様の適用分野

17 2. データサイエンスと IT 技術進歩に伴う変化 2 コンピューターの進歩によるデータサイエンスの変化 ビッグデータベース ビッグデータ対応データ解析手法 人工知能 解析目的 多種多様の適用分野

18 2. データサイエンスと IT 技術進歩に伴う変化 3 データサイエンスから 化学データサイエンス へ 化合物関連データベース ケモメトリックス 化学関連人工知能 化学関連の適用分野 化学関連解析目的

19 講演討論内容 1. 計算毒性学における人工知能の歴史 2. データサイエンスと IT 技術進歩に伴う変化 1 データサイエンスの構成技術 2 コンピューターの進歩によるデータサイエンスの変化 3 データサイエンスから 化学データサイエンス へ 3. 機械学習型人工知能適用上での解決すべき点 4. 次世代型データ解析手法開発の必要性 1KY 法の開発 2 テーラーメードモデリングの開発 7. 化学ビッグデータの構築上での問題点 8. まとめ

20 AI を正しく適用し 信頼性の高い結果を得るために 機械学習 ( 深層学習 ) 実施上での問題点 1 サンプル数と量ニューラルネットワークである限り 膨大な数のサンプルが必要 2 サンプルの品質と内容サンプルは解析目的に従った情報を有する必要があるサンプルはデータの偏りのないことが望ましい * ネガデータサンプルだけの情報で ポジを検討できない * 間違った情報を学習させない ヘイト発言をするロボット 3 学習した事や獲得情報以外への適用困難一秒後の状態認識できない 動く自動車の写真解析で一秒後を予測できない 動くものと動かないものを認識する学習必要 4 ネットワーク構造が複雑なので 要因解析ができない * 新たな研究や基本原理の解明が出来ない * 理由がわからないと 結果の保証や適用限界が出来ない

21 3. 機械学習型人工知能適用上での解決すべき点 1 サンプル数と量 2 サンプルの品質と内容 Tox21 DATA Challenge で利用されたデータ

22 3. 機械学習型人工知能適用上での解決すべき点 1 サンプル数と量 : 約 10,000 サンプルデータ 2 サンプルの品質と内容 : 品質と内容は保証済み 上記条件下で DNN を実施し 従来のデータ解析手法と比較 Applied Methods: DNNs (Deep Neural Network), SVMs (Tanimoto kernel), random forests (RF), elastic net (ElNet) Tox21 DATA Challenge で総合優勝したグループが行ったデータ解析手法間の比較結果 (cross-validation)

23 3. 機械学習型人工知能適用上での解決すべき点 3 学習した事や獲得情報以外への適用困難一秒後の状態認識できない 動く自動車の写真解析で一秒後を予測できない 動くものと動かないものを認識する学習必要 * サンプル数が十分であっても 学習で獲得できないものがある

24 3. 機械学習型人工知能適用上での解決すべき点 4 ネットワーク構造が極めて複雑なので 要因解析ができない 新たな研究や基本原理の解明が出来ない 理由がわからないと 結果の保証や適用限界が出来ない * ニューラルネットワーク発表当時 構造 - 活性相関分野ではこの説明困難な事実が 大きな問題となった できるだけわかりやすく説明してみるという実験 %EF%BC%9A 深層学習 %EF%BC%88 ディープラーニング %EF%BC%89 の基礎 ニューラルネットワーク

25 講演討論内容 1. 計算毒性学における人工知能の歴史 2. データサイエンスと IT 技術進歩に伴う変化 1 データサイエンスの構成技術 2 コンピューターの進歩によるデータサイエンスの変化 3 データサイエンスから 化学データサイエンス へ 3. 機械学習型人工知能適用上での解決すべき点 4. 次世代型データ解析手法開発の必要性 1KY 法の開発 2 テーラーメードモデリングの開発 7. 化学ビッグデータの構築上での問題点 8. まとめ

26 4. 次世代型データ解析手法開発の必要性 ビッグデータベース ビッグデータ対応データ解析手法 KY 法の展開テーラーメードモデリング ICT, IoT,CPU の急速な進歩 従来のデータ解析手法では大量データの扱いが困難 データベース 少数データ対応データ解析手法 従来手法

27 Challenge for classification and prediction K-step Yard sampling methods KY-methods The most powerful and advanced data analysis method The most difficult classification problem 6,965 sample of Ames test samples were, Classified perfectly

28 Application test of K-step Yard sampling Samples 1. Ames test data 2. Sample population total :6,965 Mutagen; 2,932 Non-mutagen; 4,033 Result of KY-method 1. Number of steps : 23 steps ; 22 (2 models) + 1 (1 model) 2.Classification ratio : 100 % Used system ADMEWORKS / ModelBuilder V Used parameters (Initial condition) Number of generated parameters : 838 Number of parameters for step 1 : 98 Confidency index (Samples(6965) / Parameters(98)) : 71.1 > 4.0

29 Application test by normal and various D.A. methods 1. Linear discriminant analysis with linear least-squares method Classification ratio : total; 73.50(6965), Mutagen;73.02(2932), Non mutagen;73.84(4033) Number of mis-classified : (1846), ( 791) (1055) Prediction ratio (L100 out) 72.58% deviance(0.92%) (L500 out) 73.32% deviance(0.18%) 2. SVM (Support Vector Machine with Kernel) Classification ratio : total; 90.87(6965), Mutagen;86.83(2932) Non mutagen; 93.80(4033) Number of mis-classified : ( 636), ( 386) ( 250) Prediction ratio (L500 out) 80.99% deviance(9.88%) 3. AdaBoost Classification ratio : total; 77.24(6965), Mutagen;66.13(2932) Non-mutagen; 85.32(4033) Number of mis-classified : (1585) ( 993) ( 592) Prediction ratio (L500 out) 75.16% deviance(2.08%)

30 Classification result by the AdaBoost Classification ratio of total 6,965 compounds is 77.24% Mis-classified Correctryclassified Correctry-classified Mis-classified Figure of the ADMEWORKS

31 Classification result by the AdaBoost Mis-classified Correctryclassified Correctry-classified Mis-classified Figure, in the case of a successful complete (100%) classification of all 6,965 Ames test compounds Figure of the ADMEWORKS

32 KY(K-step Yard sampling) 法 二クラス分類 KY 法 :Binary classification KY methods 二モデル KY 法 ; Two models BC KY method 一モデル KY 法 ; One model BC KY method モデルフリー KY 法 ; Model free BC KY method 重回帰 ( フィッティング )KY 法 :Regression KY methods 判別関数付き重回帰 KY 法 ;BC regression KY method 三ゾーン重回帰 KY 法 ;Three zone regression KY method モデルフリー重回帰 KY 法 ;Model free regression KY method クラスタリング KY 法 ; Clustering KY methods 主成分 KY 法 ; Principal component KY methods 現在開発中 リバーシ KY 法 : 二クラス分類の簡略法 ポピュレーションフリー KY 法 : ポピュレーション比率の悪い時に適用 テーラーメードモデリング (Tailor made modeling) は特許出願済み は検証済み なお テーラーメードモデリング特許出願済

33 4. 次世代型データ解析手法開発の必要性 テーラーメード モデリングの特徴 ビッグデータ対応で 予測率を高める手法 開発目標 (1) 高い予測率の実現 開発目標 (2) サンプル数がどんなに大きくなっても 同じ操作性で 予測率がより安定する手法 開発目標 (3) 分類率と予測率の乖離が少ない手法を目指す All Rights Reserved, Copyright (C)

34 従来手法による予測アプローチ (Prediction approach by traditional method) 特徴 : 総てのサンプルを対象とした予測モデルの構築 Features:Generate a prediction model which can handle all samples サンプル 1(Sample 1) サンプル 2(Sample 2) 予測結果 1 (Result 1) 予測結果 2 (Result 2) サンプル3(Sample 3) サンプル (Sample ) サンプル (Sample ) サンプル (N-1)(Sample (N-1)) サンプルN(Sample N) 予測モデル (Prediction Model) 高い汎用性 予測結果 3 (Result 3) 予測結果 (Result ) 予測結果 (Result ) 予測結果 (N-1) (Result(N-1)) 予測結果 N (Result N) 利点 (Merit) : 少ない数の予測モデル作成で済む (Small number of prediction models are generated ) 難点 (Weakness) : 予測率の向上が困難である (Difficult to achieve high prediction ratio ) All Rights Reserved, Copyright (C)

35 予測率向上に関する現時点での限界 予測の一般的な実施形態 一つの判別関数で多様性の高い複数サンプルの予測実施 メタン エタンレベルの予測から ステロイドやマクロライド等までの予測が要求される予測に無理がある 判別関数は高い汎用性を持つ このために予測の切れが悪くなる 複数サンプルの予測を保証 この目的のために 余分な情報を含み オーバースペック All Rights Reserved, Copyright (C)

36 予測率向上のための発想の転換 現状でのアプローチ 少ない判別関数で多数のサンプルを予測予測対象サンプル特異性の無い判別関数による予測 発想転換今回の提案によるアプローチ サンプル特異性の高い判別関数を構築提案 テーラーメード モデリング All Rights Reserved, Copyright (C)

37 テーラーメード モデリング の究極の形 特徴 : サンプル単位での予測モデルの構築 Features:Generate a prediction model which is designed for only 1 samples サンプル 1(Sample 1) サンプル 2(Sample 2) 予測モデル 1(Prediction model 1) 予測モデル 2(Prediction model 2) 予測結果 1 (Result 1) 予測結果 2 (Result 2) サンプル 3(Sample 3) 予測モデル 3(Prediction model 3) 予測結果 3 (Result 3) サンプル (Sample ) 予測モデル (Prediction model ) 予測結果 (Result ) サンプル (Sample ) サンプル (N-1)(Sample (N-1)) サンプル N(Sample N) 予測モデル (Prediction model ) 予測モデル (N-1)(Prediction model(n-1)) 予測モデル (N-1)(Prediction model(n-1)) 予測結果 (Result ) 予測結果 (N-1) (Result(N-1)) 予測結果 N (Result N) 利点 (Merit) : 予測率が大幅に向上する (High prediction ratio will be achieved) 難点 (Weakness): 計算時間がかかる (Need large calculation time) All Rights Reserved, Copyright (C)

38 サンプル母集団からの予測用サンプル取り出し 予測対象サンプル類似サンプル群サンプル母集団類似サンプル空間 似た化合物は似た活性を示す All Rights Reserved, Copyright (C)

39 予測用サンプルの取り出し テーラーメード予測 取り出されたサブセットのサンプル空間再構成と テーラーメード予測の実施予測対象サンプルを中心とし サンプル母集団からの類似サンプル群の取り出しサンプル類似空間サンプル分類空間 予測対象サンプル All Rights Reserved, Copyright (C)

40 テーラーメード モデリング の特徴 長所 : 1. 予測率基本原理から予測率の向上が期待される ( 要検証 ) 2. サンプル数の問題個々のサンプル単位で予測モデルを構築する 従って サンプル数の多少にかかわらず限界に近く 高い予測率を得る 3.KY 法との連携でより高い予測率が得られる分類率で 100% を実現する KY 法との連携により テーラーメード モデリング単体での実施よりも更に高い予測率を達成する可能性がある 欠点 : 1. 分類 / 予測の実施に計算時間がかかる ( スパコン主体?) All Rights Reserved, Copyright (C)

41 講演討論内容 1. 計算毒性学における人工知能の歴史 2. データサイエンスと IT 技術進歩に伴う変化 1 データサイエンスの構成技術 2 コンピューターの進歩によるデータサイエンスの変化 3 データサイエンスから 化学データサイエンス へ 3. 機械学習型人工知能適用上での解決すべき点 4. 次世代型データ解析手法開発の必要性 1KY 法の開発 2 テーラーメードモデリングの開発 5. 化学ビッグデータ構築上での問題点 6. まとめ

42 5. 化学ビッグデータ構築上での問題点 データベース中のサンプルについて 7.1 データベースに求められる一般的な事項 1 実験プロトコルが統一されているデータが必要 2 クラスサンプルポピュレーションに大きな偏りがないこと 3 偏ったデータだけではデータ解析も人工知能学習も意味無し 4 解析目的に必要な情報が含まれていること 7.2 データ解析手法の違いによる一般的な事項 1 比較的少数のサンプル群通常の多変量解析 / パターン認識手法を適用する時 2 大量のサンプル群深層学習を実施する時

43 データベース中のサンプルについて 1 実験プロトコルが統一されているデータが必要 * 実験プロトコルが異なるデータが多数存在し データ解析や機械学習の要求条件を満たしても実際の解析は意味が無いか収束しなくなる 例 : 魚毒性関連データ 文献数は多いが 大半は研究結果発表のため 魚の種類が文献単位で異なっていることが多い 文献からデータを集めてビッグデータ化しても使えない DB となる

44 データベース中のサンプルについて 2 クラスサンプルポピュレーションに大きな偏りがないこと 3 偏ったデータは データ解析も人工知能学習も意味無し片一方のクラスデータだけで 対立クラスの予測はできない * ネガティブデータの学習でポジティブデータを予測できない Tox21 で利用されたデータ

45 データベース中のサンプルについて 4 解析目的に必要な情報が含まれていること * 適用分野と関係のないサンプルデータは使えないサンプル数を多くすれば良いというわけではない * 解析目的に関係するようで関係のない情報がある場合一般的にはノイズデータ ノイズサンプルとなる サンプルデータそのものの場合は データ解析実施前に取り出し必要 解析目的に関係するが解析上ノイズとなるサンプル テーラーメードモデリング ( 後述 ) の実施 パラメーターに含まれている場合 特徴抽出の実施

46 6. まとめ データサイエンスの時代は確実にやってくる情報やコンピューターハードウエアの進歩は止まない 新しいデータサイエンス時代に向けた対応が必要新しいデータサイエンス時代では データベースのスケールアップ 新規のデータ解析手法の展開に加えて人工知能等が重要なコア技術となる 化学分野では 化学分野の様々な対応を解析目的とする 化学データサイエンス の展開が必要創薬 機能性化合物デザイン 化合物安全性評価その他の様々な分野で化合物とデジタルの接点が重要となっている 化学データサイエンス を基本とした基本技術の構成 1 化学ビッグデータベース 2 ケモメトリックス 3 化学人工知能 個別の展開と同時に 化学分野全体を見つめた開発が必要データ量が大きいので 再構築や再解析等は簡単にできない

47 Thank you for your attention 株式会社インシリコデータ 湯田浩太郎

スライド 1

スライド 1 計算毒性学における 人工知能の歴史と現状 株式会社インシリコデータ 湯田浩太郎 http://www.insilicodata.com 人工知能に注目させたトピックス 成功事例 AlphaGo( アルファ碁 ) が人間に打ち勝って世界一になった人間がコンピュータに勝てる最後の分野の神話が崩れた 碁の学習アルゴリズムに深層学習を適用していた 失敗事例 人工知能のチャットボット Tay がヘイト発言を乱発した

More information

スライド 1

スライド 1 ICT IoT やビッグデータ時代の ケモメトリックス / 人工知能を知って 新たなチャレンジを 株式会社インシリコデータ 湯田浩太郎 http://www.insilicodata.com 時代の新しい三大潮流 ICT : Information and Communication Technology ( 情報通信技術 ) 情報技術に通信コミュニケーションの重要性を加味した言葉 IoT : Internet

More information

Rの基本操作

Rの基本操作 Microsoft Azure 高校生のための Azure Machine Learning By M. Takezawa 機械学習 (Machine Learning) とは 機械学習とは 機械にデータを学習させ データに潜むパターンや特性を発見し予測させることです Microsoft Azure Machine Learning とは Microsoft 社が提供する Azure の機能の一つであり

More information

ハピタス のコピー.pages

ハピタス のコピー.pages Copyright (C) All Rights Reserved. 10 12,500 () ( ) ()() 1 : 2 : 3 : 2 4 : 5 : Copyright (C) All Rights Reserved. Copyright (C) All Rights Reserved. Copyright (C) All Rights Reserved. Copyright (C) All

More information

Copyright 2008 All Rights Reserved 2

Copyright 2008 All Rights Reserved 2 Copyright 2008 All Rights Reserved 1 Copyright 2008 All Rights Reserved 2 Copyright 2008 All Rights Reserved 3 Copyright 2008 All Rights Reserved 4 Copyright 2008 All Rights Reserved 5 Copyright 2008 All

More information

udc-2.dvi

udc-2.dvi 13 0.5 2 0.5 2 1 15 2001 16 2009 12 18 14 No.39, 2010 8 2009b 2009a Web Web Q&A 2006 2007a20082009 2007b200720082009 20072008 2009 2009 15 1 2 2 2.1 18 21 1 4 2 3 1(a) 1(b) 1(c) 1(d) 1) 18 16 17 21 10

More information

AI AI Artificial Intelligence AI Strategy& Foresight AI AI AI AI 1 AI AI AI AI AI AI AI AI AI AI AI AI AI 2 AI 1 AI AI 3 AI 3 20 AI AI AI AI AI

AI AI Artificial Intelligence AI Strategy& Foresight AI AI AI AI 1 AI AI AI AI AI AI AI AI AI AI AI AI AI 2 AI 1 AI AI 3 AI 3 20 AI AI AI AI AI AI AIArtificial Intelligence AI Strategy& Foresight AI AIAI AI 1 AI AI AI AI AI AI AI AI AI AI AI AI AI 2 AI 1 AI 50 80 AI 3 AI 3 20 AI AI AI AI AI IoT AI AI 4 Strategy& Foresight Vol.15 2018 Spring masahiro.m.ozaki@pwc.com

More information

ソフト活用事例③自動Rawデータ管理システム

ソフト活用事例③自動Rawデータ管理システム ソフト活用事例 3 自動 Raw データ管理システム ACD/Labs NMR 無料講習会 & セミナー 2014 於 )2014.7.29 東京 /2014.7.31 大阪 富士通株式会社テクニカルコンピューティング ソリューション事業本部 HPC アプリケーション統括部 ACD/Spectrus をご選択頂いた理由 (NMR 領域 ) パワフルな解 析機能 ベンダーニュートラルな解析環境 直感的なインターフェース

More information

kut-paper-template.dvi

kut-paper-template.dvi 26 Discrimination of abnormal breath sound by using the features of breath sound 1150313 ,,,,,,,,,,,,, i Abstract Discrimination of abnormal breath sound by using the features of breath sound SATO Ryo

More information

..,,,, , ( ) 3.,., 3.,., 500, 233.,, 3,,.,, i

..,,,, , ( ) 3.,., 3.,., 500, 233.,, 3,,.,, i 25 Feature Selection for Prediction of Stock Price Time Series 1140357 2014 2 28 ..,,,,. 2013 1 1 12 31, ( ) 3.,., 3.,., 500, 233.,, 3,,.,, i Abstract Feature Selection for Prediction of Stock Price Time

More information

24 SPAM Performance Comparison of Machine Learning Algorithms for SPAM Discrimination

24 SPAM Performance Comparison of Machine Learning Algorithms for SPAM Discrimination 24 SPAM Performance Comparison of Machine Learning Algorithms for SPAM Discrimination 1130378 2013 3 9 SPAM SPAM SPAM SPAM SVM AdaBoost RandomForest SPAM SPAM UCI Machine Learning Repository Spambase 4601

More information

Ⅱ 方法と対象 1. 所得段階別保険料に関する情報の収集 ~3 1, 分析手法

Ⅱ 方法と対象 1. 所得段階別保険料に関する情報の収集 ~3 1, 分析手法 67 論文 要旨 : 1507 キーワード : Ⅰ 緒言 65 2014 89 5 2 3 5 6 2 5 6 3 4 3 4 2006 2015 9 9 68 2018. 3 2015 Ⅱ 方法と対象 1. 所得段階別保険料に関する情報の収集 6 2015 7 2016 2 ~3 1,507 2. 分析手法 69 3 3 3 A B C 3 A 4 1 1 2 2 2014 2014 2014 5

More information

弥生会計/やよいの青色申告

弥生会計/やよいの青色申告 c c c c c 1 c c c c c c c c c c c 2 3 c c 4 a a a a a a a a a

More information

AI技術の紹介とセンサーデータ解析への応用

AI技術の紹介とセンサーデータ解析への応用 AI を活用したセンサーデータ解析 MathWorks Japan アプリケーションエンジニアリンググループアプリケーションエンジニア吉田剛士 2018 The MathWorks, Inc. 1 AI を活用したセンサーデータ解析 11:20-11:50 MATLAB による AI 作成 アプリを使った簡易的な解析 学習モデルのパラメータ自動調整 学習モデルのスタンドアロン化 2 課題 : ターボファンエンジンの予知保全

More information

初心者にもできるアメブロカスタマイズ新2016.pages

初心者にもできるアメブロカスタマイズ新2016.pages Copyright All Rights Reserved. 41 Copyright All Rights Reserved. 60 68 70 6 78 80 Copyright All Rights Reserved. FC2 97 Copyright All Rights Reserved. Copyright All Rights Reserved. Copyright All Rights

More information

- 2 Copyright (C) 2006. All Rights Reserved.

- 2 Copyright (C) 2006. All Rights Reserved. - 2 Copyright (C) 2006. All Rights Reserved. 2-3 Copyright (C) 2006. All Rights Reserved. 70-4 Copyright (C) 2006. All Rights Reserved. ...1...3...7...8 1...9...14...16 2...18...20...21 3...22...23...23...24

More information

1. 多変量解析の基本的な概念 1. 多変量解析の基本的な概念 1.1 多変量解析の目的 人間のデータは多変量データが多いので多変量解析が有用 特性概括評価特性概括評価 症 例 主 治 医 の 主 観 症 例 主 治 医 の 主 観 単変量解析 客観的規準のある要約多変量解析 要約値 客観的規準のな

1. 多変量解析の基本的な概念 1. 多変量解析の基本的な概念 1.1 多変量解析の目的 人間のデータは多変量データが多いので多変量解析が有用 特性概括評価特性概括評価 症 例 主 治 医 の 主 観 症 例 主 治 医 の 主 観 単変量解析 客観的規準のある要約多変量解析 要約値 客観的規準のな 1.1 多変量解析の目的 人間のデータは多変量データが多いので多変量解析が有用 特性概括評価特性概括評価 症 例 治 医 の 観 症 例 治 医 の 観 単変量解析 客観的規準のある要約多変量解析 要約値 客観的規準のない要約知識 直感 知識 直感 総合的評価 考察 総合的評価 考察 単変量解析の場合 多変量解析の場合 < 表 1.1 脂質異常症患者の TC と TG と重症度 > 症例 No. TC

More information

Microsoft PowerPoint - 資料04 重回帰分析.ppt

Microsoft PowerPoint - 資料04 重回帰分析.ppt 04. 重回帰分析 京都大学 加納学 Division of Process Control & Process Sstems Engineering Department of Chemical Engineering, Koto Universit manabu@cheme.koto-u.ac.jp http://www-pse.cheme.koto-u.ac.jp/~kano/ Outline

More information

untitled

untitled http://www.riskdatabank.co.jp The of Japan, Ltd. All rights reserved. 2 The of Japan, Ltd. All rights reserved. 3 The of Japan, Ltd. All rights reserved. 4 The of Japan, Ltd. All rights reserved. 5 The

More information

WHITE PAPER RNN

WHITE PAPER RNN WHITE PAPER RNN ii 1... 1 2 RNN?... 1 2.1 ARIMA... 1 2.2... 2 2.3 RNN Recurrent Neural Network... 3 3 RNN... 5 3.1 RNN... 6 3.2 RNN... 6 3.3 RNN... 7 4 SAS Viya RNN... 8 4.1... 9 4.2... 11 4.3... 15 5...

More information

25 fmri A study of discrimination of musical harmony using brain activity obtained by fmri

25 fmri A study of discrimination of musical harmony using brain activity obtained by fmri 25 fmri A study of discrimination of musical harmony using brain activity obtained by fmri 1140359 2014 2 28 fmri fmri BCI(Brain Computer Interface) 6 (C C# D D# E F) 6 (Cm C#m Dm D#m Em Fm) 12 fmri fmri

More information

Copyright All Rights Reserved. -2 -!

Copyright All Rights Reserved. -2 -! http://ameblo.jp/admarketing/ Copyright All Rights Reserved. -2 -! Copyright All Rights Reserved. -3- Copyright All Rights Reserved. -4- Copyright All Rights Reserved. -5 - Copyright All Rights Reserved.

More information

PowerPoint プレゼンテーション

PowerPoint プレゼンテーション GSN を応用したナレッジマネジメントシステムの提案 2017 年 10 月 27 日 D-Case 研究会 国立研究開発法人宇宙航空研究開発機構 研究開発部門第三研究ユニット 梅田浩貴 2017/3/27 C Copyright 2017 JAXA All rights reserved 1 目次 1 課題説明 SECI モデル 2 GSN を応用したナレッジマネジメントシステム概要 3 ツリー型チェックリスト分析

More information

1. 薬物や機能性化合物デザインの基本と環境の変化 2. 多変量解析 / パターン認識に基づく インシリコスクリーニングとインシリコデザインの基本原理 3. 並列創薬 ( 複数特性の同時最適化 ) の提案 4. 並列創薬 と 逐次創薬 の開発成功確率に基づく比較シミュレーション 5. 分類 / 予測

1. 薬物や機能性化合物デザインの基本と環境の変化 2. 多変量解析 / パターン認識に基づく インシリコスクリーニングとインシリコデザインの基本原理 3. 並列創薬 ( 複数特性の同時最適化 ) の提案 4. 並列創薬 と 逐次創薬 の開発成功確率に基づく比較シミュレーション 5. 分類 / 予測 2009.2.27 早期 ADMET の基本と 21 世紀の創薬手法 並列創薬 のご提案 富士通 ( 株 ) バイオ IT 事業開発本部大阪大学臨床医工学融合研究教育センター湯田浩太郎 1. 薬物や機能性化合物デザインの基本と環境の変化 2. 多変量解析 / パターン認識に基づく インシリコスクリーニングとインシリコデザインの基本原理 3. 並列創薬 ( 複数特性の同時最適化 ) の提案 4. 並列創薬

More information

IPA:セキュアなインターネットサーバー構築に関する調査

IPA:セキュアなインターネットサーバー構築に関する調査 Copyright 2003 IPA, All Rights Reserved. Copyright 2003 IPA, All Rights Reserved. Copyright 2003 IPA, All Rights Reserved. Copyright 2003 IPA, All Rights Reserved. Copyright 2003 IPA, All Rights Reserved.

More information

ワトソンで体感する人工知能 フォローアップ情報 株式会社リックテレコム / 書籍出版部 ( 最終情報更新日 :2018 年 4 月 5 日 ) [INDEX] 2018 年 4 月 1 日時点の IBM Watson 仕様変更について ( 著者 : 井上研一氏からのフォロー情報 ) [ 変更点 -1

ワトソンで体感する人工知能 フォローアップ情報 株式会社リックテレコム / 書籍出版部 ( 最終情報更新日 :2018 年 4 月 5 日 ) [INDEX] 2018 年 4 月 1 日時点の IBM Watson 仕様変更について ( 著者 : 井上研一氏からのフォロー情報 ) [ 変更点 -1 ワトソンで体感する人工知能 フォローアップ情報 株式会社リックテレコム / 書籍出版部 ( 最終情報更新日 :2018 年 4 月 5 日 ) [INDEX] 2018 年 4 月 1 日時点の IBM Watson 仕様変更について ( 著者 : 井上研一氏からのフォロー情報 ) [ 変更点 -1] IBM Cloud の登録とライトアカウントについて [ 変更点 -2] IBM Cloud における

More information

Microsoft Word - 最終版 バックせどりismマニュアル .docx

Microsoft Word - 最終版 バックせどりismマニュアル .docx ism ISM ISM ISM ISM ISM ISM Copyright (c) 2010 All Rights Reserved. Copyright (c) 2010 All Rights Reserved. Copyright (c) 2010 All Rights Reserved. ISM Copyright (c) 2010 All Rights Reserved. Copyright

More information

情報処理学会研究報告 IPSJ SIG Technical Report Vol.2015-GI-34 No /7/ % Selections of Discarding Mahjong Piece Using Neural Network Matsui

情報処理学会研究報告 IPSJ SIG Technical Report Vol.2015-GI-34 No /7/ % Selections of Discarding Mahjong Piece Using Neural Network Matsui 2 3 2000 3.3% Selections of Discarding Mahjong Piece Using Neural Network Matsui Kazuaki Matoba Ryuichi 2 Abstract: Mahjong is one of games with imperfect information, and its rule is very complicated

More information

kubostat2018d p.2 :? bod size x and fertilization f change seed number? : a statistical model for this example? i response variable seed number : { i

kubostat2018d p.2 :? bod size x and fertilization f change seed number? : a statistical model for this example? i response variable seed number : { i kubostat2018d p.1 I 2018 (d) model selection and kubo@ees.hokudai.ac.jp http://goo.gl/76c4i 2018 06 25 : 2018 06 21 17:45 1 2 3 4 :? AIC : deviance model selection misunderstanding kubostat2018d (http://goo.gl/76c4i)

More information

Slide 1

Slide 1 ハンズオン受講の為の準備講座 これから始める人の為の ディープラーニング基礎講座 村上真奈 NVIDIA CUDA & Deep Learning Solution Architect NVIDIA Corporation 1 機械学習とディープラーニングの関係 AGENDA ディープラーニングとは? ニューラルネットワークの構造 ディープラーニングの学習とは 畳み込みニューラルネットワーク 午後に予定しているハンズオンの為の基礎講座ディープラーニングをこれから始める方を対象に基礎概念や用語の解説を行います

More information

IPSJ SIG Technical Report Vol.2014-IOT-27 No.14 Vol.2014-SPT-11 No /10/10 1,a) 2 zabbix Consideration of a system to support understanding of f

IPSJ SIG Technical Report Vol.2014-IOT-27 No.14 Vol.2014-SPT-11 No /10/10 1,a) 2 zabbix Consideration of a system to support understanding of f 1,a) 2 zabbix Consideration of a system to support understanding of fault occurrences based on the similarity of the time series Miyaza Nao 1,a) Masuda Hideo 2 Abstract: With the development of network

More information

パワーポイントの品質と生産性を向上させるデザイン・テンプレート

パワーポイントの品質と生産性を向上させるデザイン・テンプレート 著作権法改正が AI 開発に与える 衝撃 2019.3.06 STORIA 法律事務所弁護士柿沼太一 自己紹介 2000 年 4 月に弁護士登録 2015 年 3 月に神戸三宮に STORIA 法律事務所設立 AI IT 知的財産 ベンチャーを主として取り扱う 2016 年 10 月からAIに関して積極的な情報発信を始め 現在自動車系 医療系 工場系 WEB 系など多様なAI 企業からの相談 顧問契約を締結

More information

PowerPoint プレゼンテーション

PowerPoint プレゼンテーション 復習 ) 時系列のモデリング ~a. 離散時間モデル ~ y k + a 1 z 1 y k + + a na z n ay k = b 0 u k + b 1 z 1 u k + + b nb z n bu k y k = G z 1 u k = B(z 1 ) A(z 1 u k ) ARMA モデル A z 1 B z 1 = 1 + a 1 z 1 + + a na z n a = b 0

More information

浜松医科大学紀要

浜松医科大学紀要 On the Statistical Bias Found in the Horse Racing Data (1) Akio NODA Mathematics Abstract: The purpose of the present paper is to report what type of statistical bias the author has found in the horse

More information

Numerical Analysis II, Exam End Term Spring 2017

Numerical Analysis II, Exam End Term Spring 2017 H. Ammari W. Wu S. Yu Spring Term 2017 Numerical Analysis II ETH Zürich D-MATH End Term Spring 2017 Problem 1 Consider dx = f(t, x), t [0, T ] dt x(0) = x 0 R [28 Marks] with f C subject to the Lipschitz

More information

Study on Throw Accuracy for Baseball Pitching Machine with Roller (Study of Seam of Ball and Roller) Shinobu SAKAI*5, Juhachi ODA, Kengo KAWATA and Yu

Study on Throw Accuracy for Baseball Pitching Machine with Roller (Study of Seam of Ball and Roller) Shinobu SAKAI*5, Juhachi ODA, Kengo KAWATA and Yu Study on Throw Accuracy for Baseball Pitching Machine with Roller (Study of Seam of Ball and Roller) Shinobu SAKAI*5, Juhachi ODA, Kengo KAWATA and Yuichiro KITAGAWA Department of Human and Mechanical

More information

自己紹介 名前 : 竹田卓也 年齢 : 20 歳 ( 大学生 ) 経歴 : 人工知能歴 1ヶ月プログラミング歴 5 年くらい 言語 : PythonとかJavaとかGoとか 趣味 : オンライン オフラインゲーム 2

自己紹介 名前 : 竹田卓也 年齢 : 20 歳 ( 大学生 ) 経歴 : 人工知能歴 1ヶ月プログラミング歴 5 年くらい 言語 : PythonとかJavaとかGoとか 趣味 : オンライン オフラインゲーム 2 リカレントニューラルネットワークの概要と動作原理 竹田卓也 後援 : ドワンゴ 1 自己紹介 名前 : 竹田卓也 年齢 : 20 歳 ( 大学生 ) 経歴 : 人工知能歴 1ヶ月プログラミング歴 5 年くらい 言語 : PythonとかJavaとかGoとか 趣味 : オンライン オフラインゲーム 2 アウトライン Feed forward neural network Recurrent neural

More information

Microsoft Word - ??? ????????? ????? 2013.docx

Microsoft Word - ??? ????????? ????? 2013.docx @ィーィェィケィャi@@ @@pbィ 050605a05@07ィ 050605a@070200 pbィ 050605a05@07ィ 050605a@070200@ィーィィu05@0208 1215181418 12 1216121419 171210 1918181811 19181719101411 1513 191815181611 19181319101411 18121819191418 1919151811

More information

untitled

untitled mitsuya Copyright (C) 2007. All Rights Reserved. 1/1 mitsuya Copyright (C) 2007. All Rights Reserved. 2/2 mitsuya Copyright (C) 2007. All Rights Reserved. 3/3 mitsuya Copyright (C) 2007. All Rights Reserved.

More information

! " # $ % & ' ( ) +, -. / 0 1 2 3 4 5 6 7 8 9 : ; < = >? @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z [ ] ^ _ ` a b c d e f h i j k l m n o p q r s t u v w x y z { } ~ This product is

More information

it-ken_open.key

it-ken_open.key 深層学習技術の進展 ImageNet Classification 画像認識 音声認識 自然言語処理 機械翻訳 深層学習技術は これらの分野において 特に圧倒的な強みを見せている Figure (Left) Eight ILSVRC-2010 test Deep images and the cited4: from: ``ImageNet Classification with Networks et

More information

29 Short-time prediction of time series data for binary option trade

29 Short-time prediction of time series data for binary option trade 29 Short-time prediction of time series data for binary option trade 1180365 2018 2 28 RSI(Relative Strength Index) 3 USD/JPY 1 2001 1 2 4 10 2017 12 29 17 00 1 high low i Abstract Short-time prediction

More information

健康保険組合のあゆみ_top

健康保険組合のあゆみ_top (1912) (1951) 2,00024,000 (1954) (1958) (1962) (1965) (1968) (1969) (1971) (1972) (1973) (1974) (1976) (1978) (1980) (1982) (1983) (1984) (1985) (1987) (1988) (1989) (1990) (1991) (1992) (1994) (1995)

More information

2017 (413812)

2017 (413812) 2017 (413812) Deep Learning ( NN) 2012 Google ASIC(Application Specific Integrated Circuit: IC) 10 ASIC Deep Learning TPU(Tensor Processing Unit) NN 12 20 30 Abstract Multi-layered neural network(nn) has

More information

Solibri Model Checker 9.5 スタードガイド

Solibri Model Checker 9.5 スタードガイド SOLIBRI MODEL CHECKER V9.5 Copyright 2014 Solibri Oy All Rights Reserved 1 Copyright 2014 Solibri Oy All Rights Reserved 2 Copyright 2014 Solibri Oy All Rights Reserved 3 Copyright 2014 Solibri Oy All

More information

PPTテンプレート集 ver.1.0

PPTテンプレート集 ver.1.0 Copyright 2012 Radishbo-ya co., Ltd All Rights Reserved. 1 !!!!!!!!!! Copyright 2012 Radishbo-ya co., Ltd All Rights Reserved. 2 Copyright 2012 Radishbo-ya co., Ltd All Rights Reserved. Copyright 2012

More information

10 2000 11 11 48 ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) CU-SeeMe NetMeeting Phoenix mini SeeMe Integrated Services Digital Network 64kbps 16kbps 128kbps 384kbps

More information

Chapter 1 Epidemiological Terminology

Chapter 1 Epidemiological Terminology Appendix Real examples of statistical analysis 検定 偶然を超えた差なら有意差という P

More information

A Study on Throw Simulation for Baseball Pitching Machine with Rollers and Its Optimization Shinobu SAKAI*5, Yuichiro KITAGAWA, Ryo KANAI and Juhachi

A Study on Throw Simulation for Baseball Pitching Machine with Rollers and Its Optimization Shinobu SAKAI*5, Yuichiro KITAGAWA, Ryo KANAI and Juhachi A Study on Throw Simulation for Baseball Pitching Machine with Rollers and Its Optimization Shinobu SAKAI*5, Yuichiro KITAGAWA, Ryo KANAI and Juhachi ODA Department of Human and Mechanical Systems Engineering,

More information

コンピュータ応用・演習 情報処理システム

コンピュータ応用・演習 情報処理システム 2010 年 12 月 15 日 データエンジニアリング 演習 情報処理システム データマイニング ~ データからの自動知識獲得手法 ~ 1. 演習の目的 (1) 多種多様な膨大な量のデータを解析し, 企業の経営活動などに活用することが望まれている. 大規模データベースを有効に活用する, データマイニング技術の研究が脚光を浴びている 1 1. 演習の目的 (2) POS データを用いて顧客の購買パターンを分析する.

More information

人工知能学会研究会資料 SIG-KBS-B Analysis of Voting Behavior in One Night Werewolf 1 2 Ema Nishizaki 1 Tomonobu Ozaki Graduate School of Integrated B

人工知能学会研究会資料 SIG-KBS-B Analysis of Voting Behavior in One Night Werewolf 1 2 Ema Nishizaki 1 Tomonobu Ozaki Graduate School of Integrated B 人工知能学会研究会資料 SIG-KBS-B508-09 Analysis of Voting Behavior in One Night Werewolf 1 2 Ema Nishizaki 1 Tomonobu Ozaki 2 1 1 Graduate School of Integrated Basic Sciences, Nihon University 2 2 College of Humanities

More information

Microsoft PowerPoint - H17-5時限(パターン認識).ppt

Microsoft PowerPoint - H17-5時限(パターン認識).ppt パターン認識早稲田大学講義 平成 7 年度 独 産業技術総合研究所栗田多喜夫 赤穂昭太郎 統計的特徴抽出 パターン認識過程 特徴抽出 認識対象から何らかの特徴量を計測 抽出 する必要がある 認識に有効な情報 特徴 を抽出し 次元を縮小した効率の良い空間を構成する過程 文字認識 : スキャナ等で取り込んだ画像から文字の識別に必要な本質的な特徴のみを抽出 例 文字線の傾き 曲率 面積など 識別 与えられた未知の対象を

More information

Abstract This paper concerns with a method of dynamic image cognition. Our image cognition method has two distinguished features. One is that the imag

Abstract This paper concerns with a method of dynamic image cognition. Our image cognition method has two distinguished features. One is that the imag 2004 RGB A STUDY OF RGB COLOR INFORMATION AND ITS APPLICATION 03R3237 Abstract This paper concerns with a method of dynamic image cognition. Our image cognition method has two distinguished features. One

More information

CPP46 UFO Image Analysis File on yucatan091206a By Tree man (on) BLACK MOON (Kinohito KULOTSUKI) CPP46 UFO 画像解析ファイル yucatan091206a / 黒月樹人 Fig.02 Targe

CPP46 UFO Image Analysis File on yucatan091206a By Tree man (on) BLACK MOON (Kinohito KULOTSUKI) CPP46 UFO 画像解析ファイル yucatan091206a / 黒月樹人 Fig.02 Targe CPP46 UFO Image Analysis File on yucatan091206a By Tree man (on) BLACK MOON (Kinohito KULOTSUKI) CPP46 UFO 画像解析ファイル yucatan091206a / 黒月樹人 Fig.02 Target (T) of Fig.01 Original Image of yucatan091206a yucatan091206a

More information

Studies of Foot Form for Footwear Design (Part 9) : Characteristics of the Foot Form of Young and Elder Women Based on their Sizes of Ball Joint Girth

Studies of Foot Form for Footwear Design (Part 9) : Characteristics of the Foot Form of Young and Elder Women Based on their Sizes of Ball Joint Girth Studies of Foot Form for Footwear Design (Part 9) : Characteristics of the Foot Form of Young and Elder Women Based on their Sizes of Ball Joint Girth and Foot Breadth Akiko Yamamoto Fukuoka Women's University,

More information

1 Table 1: Identification by color of voxel Voxel Mode of expression Nothing Other 1 Orange 2 Blue 3 Yellow 4 SSL Humanoid SSL-Vision 3 3 [, 21] 8 325

1 Table 1: Identification by color of voxel Voxel Mode of expression Nothing Other 1 Orange 2 Blue 3 Yellow 4 SSL Humanoid SSL-Vision 3 3 [, 21] 8 325 社団法人人工知能学会 Japanese Society for Artificial Intelligence 人工知能学会研究会資料 JSAI Technical Report SIG-Challenge-B3 (5/5) RoboCup SSL Humanoid A Proposal and its Application of Color Voxel Server for RoboCup SSL

More information

Copyright 2017 JAPAN POST BANK CO., LTD. All Rights Reserved. 1

Copyright 2017 JAPAN POST BANK CO., LTD. All Rights Reserved. 1 Copyright 2017 JAPAN POST BANK CO., LTD. All Rights Reserved. 1 Copyright 2017 JAPAN POST BANK CO., LTD. All Rights Reserved. 2 60 50 40 30 20 10 0 20173 20183 Copyright 2017 JAPAN POST BANK CO., LTD.

More information

PowerPoint Presentation

PowerPoint Presentation . カーネル法への招待 正定値カーネルによるデータ解析 - カーネル法の基礎と展開 - 福水健次統計数理研究所 / 総合研究大学院大学 統計数理研究所公開講座 0 年 月 34 日 概要 カーネル法の基本 線形データ解析と非線形データ解析 カーネル法の原理 カーネル法の つの例 カーネル主成分分析 : PCA の非線形拡張 リッジ回帰とそのカーネル化 概要 カーネル法の基本 線形データ解析と非線形データ解析

More information

3. ( 1 ) Linear Congruential Generator:LCG 6) (Mersenne Twister:MT ), L 1 ( 2 ) 4 4 G (i,j) < G > < G 2 > < G > 2 g (ij) i= L j= N

3. ( 1 ) Linear Congruential Generator:LCG 6) (Mersenne Twister:MT ), L 1 ( 2 ) 4 4 G (i,j) < G > < G 2 > < G > 2 g (ij) i= L j= N RMT 1 1 1 N L Q=L/N (RMT), RMT,,,., Box-Muller, 3.,. Testing Randomness by Means of RMT Formula Xin Yang, 1 Ryota Itoi 1 and Mieko Tanaka-Yamawaki 1 Random matrix theory derives, at the limit of both dimension

More information

Microsoft Word - 1 color Normalization Document _Agilent version_ .doc

Microsoft Word - 1 color Normalization Document _Agilent version_ .doc color 実験の Normalization color 実験で得られた複数のアレイデータを相互比較するためには Normalization( 正規化 ) が必要です 2 つのサンプルを異なる色素でラベル化し 競合ハイブリダイゼーションさせる 2color 実験では 基本的に Dye Normalization( 色素補正 ) が適用されますが color 実験では データの特徴と実験の目的 (

More information

概要 単語の分散表現に基づく統計的機械翻訳の素性を提案 既存手法の FFNNLM に CNN と Gate を追加 dependency- to- string デコーダにおいて既存手法を上回る翻訳精度を達成

概要 単語の分散表現に基づく統計的機械翻訳の素性を提案 既存手法の FFNNLM に CNN と Gate を追加 dependency- to- string デコーダにおいて既存手法を上回る翻訳精度を達成 Encoding Source Language with Convolu5onal Neural Network for Machine Transla5on Fandong Meng, Zhengdong Lu, Mingxuan Wang, Hang Li, Wenbin Jiang, Qun Liu, ACL- IJCNLP 2015 すずかけ読み会奥村 高村研究室博士二年上垣外英剛 概要

More information

22 2016 3 82 1 1

22 2016 3 82 1 1 : 81 1 2 3 4 1990 2015 22 2016 3 82 1 1 83 : 2 5 84 22 2016 3 6 3 7 8 2 : 85 1 S 12 S S S S S S S S S 86 22 2016 3 S S S S S S S S 2 S S : 87 S 9 3 2 1 10 S 11 22 2016 3 88 1 : 89 1 2 3 4 90 22 2016 3

More information

スライド 1

スライド 1 膨大なデータからの価値創出 ~ 画像 映像処理技術の最先端 ~ 長谷山美紀 第 9 回インフラ イノベーション研究会 IST Information Science and Technology Hokkaido University 本日の講演内容 1. はじめに大量ディジタルデータ時代の到来 2. 画像 映像処理の最先端 いかにして望む情報を獲得するか? 3. 実データ適用の試み 4. まとめディジタルデータ時代のこれから

More information

新製品開発プロジェクトの評価手法

新製品開発プロジェクトの評価手法 CIRJE-J-60 2001 8 A note on new product project selection model: Empirical analysis in chemical industry Kenichi KuwashimaUniversity of Tokyo Junichi TomitaUniversity of Tokyo August, 2001 Abstract By

More information

企業倒産とマクロ経済要因

企業倒産とマクロ経済要因 No.28 2003 11 Abstract The purpose of this paper is to clarify the relationship between business bankruptcy and macro-economic indices. The method is a multiple linear regression model. As dependent variables,

More information

/ , ,908 4,196 2, ,842 38, / / 2 33 /

/ , ,908 4,196 2, ,842 38, / / 2 33 / MathWorks Automotive Conference 2014 ( ) ECU 0.1. 1 /30 1949 12 16 1,874 4 959 2 4,908 4,196 2,993 139,842 38,581 62 26 35 56 / 6 185 13 4 3 11 / 2 33 / 2014 3 31 0.1. 2 /30 ETC 0.2. 3 /30 1. 1. 2. 2.

More information

P. 2 P. 4 P. 5 P. 6 P. 7 P. 9 P P.11 P.14 P.15 P.16 P.16 P.17 P.19 P.20 P.22 P P P P P P P P P

P. 2 P. 4 P. 5 P. 6 P. 7 P. 9 P P.11 P.14 P.15 P.16 P.16 P.17 P.19 P.20 P.22 P P P P P P P P P 201628 3 2016 5 13 P. 2 P. 4 P. 5 P. 6 P. 7 P. 9 P.10 2016 P.11 P.14 P.15 P.16 P.16 P.17 P.19 P.20 P.22 P.23 10 P.24 11 P.26 12 P.27 13 P.28 14 P.28 15 P.29 16 P.30 17 P.31 P.33 P.34 Copyright 2016 JAPAN

More information

P. 2 P. 4 P. 5 P. 6 P. 7 P. 9 P.10 P.12 P.13 P.14 P.14 P.15 P.17 P.18 P.20 P P P P P.25 P.27 P.28 Copyright 2016 JAPAN POST BA

P. 2 P. 4 P. 5 P. 6 P. 7 P. 9 P.10 P.12 P.13 P.14 P.14 P.15 P.17 P.18 P.20 P P P P P.25 P.27 P.28 Copyright 2016 JAPAN POST BA 201729 3 1 2016 8 12 P. 2 P. 4 P. 5 P. 6 P. 7 P. 9 P.10 P.12 P.13 P.14 P.14 P.15 P.17 P.18 P.20 P.21 10 P.22 11 P.23 12 P.24 13 P.25 P.27 P.28 Copyright 2016 JAPAN POST BANK CO., LTD. All Rights Reserved.

More information

Fig. 3 Coordinate system and notation Fig. 1 The hydrodynamic force and wave measured system Fig. 2 Apparatus of model testing

Fig. 3 Coordinate system and notation Fig. 1 The hydrodynamic force and wave measured system Fig. 2 Apparatus of model testing The Hydrodynamic Force Acting on the Ship in a Following Sea (1 St Report) Summary by Yutaka Terao, Member Broaching phenomena are most likely to occur in a following sea to relative small and fast craft

More information

25 Removal of the fricative sounds that occur in the electronic stethoscope

25 Removal of the fricative sounds that occur in the electronic stethoscope 25 Removal of the fricative sounds that occur in the electronic stethoscope 1140311 2014 3 7 ,.,.,.,.,.,.,.,,.,.,.,.,,. i Abstract Removal of the fricative sounds that occur in the electronic stethoscope

More information

日本機械学会 生産システム部門研究発表講演会 2015 資料

日本機械学会 生産システム部門研究発表講演会 2015 資料 ( 社 ) 日本機械学会生産システム部門研究発表講演会 2015 製造オペレーションマネジメント入門 ~ISA-95 が製造業を変える ~ 事例による説明 2015-3-16 Ver.1 IEC/SC65E/JWG5 国内委員アズビル株式会社村手恒夫 目次 事例によるケーススタディの目的 事例 : 果汁入り飲料水製造工場 情報システム構築の流れ 1. 対象問題のドメインと階層の確認 2. 生産現場での課題の調査と整理

More information

Microsoft, Windows Microsoft Corporation

Microsoft, Windows Microsoft Corporation Copyright 2000-2002 T&D Corporation. All rights reserved. 2002.07 16004194030 Microsoft, Windows Microsoft Corporation This Modem Logger is designed for use in Japan only and can not be used in any other

More information

Influence of Material and Thickness of the Specimen to Stress Separation of an Infrared Stress Image Kenji MACHIDA The thickness dependency of the temperature image obtained by an infrared thermography

More information

① Copyright 2005 Impex.,inc. All Rights Reserved 1 Copyright 2005 Impex.,inc. All Rights Reserved 2 Copyright 2005 Impex.,inc. All Rights Reserved 3 Copyright 2005 Impex.,inc. All Rights Reserved 4 Copyright

More information

終身雇用制 : 再考 - ニューラルネットワークによる財務業績の検証 Revising Japanese Lifetime Employment System: Financial Performance Analysis using Artificial Neural Networks 岡本大輔

終身雇用制 : 再考 - ニューラルネットワークによる財務業績の検証 Revising Japanese Lifetime Employment System: Financial Performance Analysis using Artificial Neural Networks 岡本大輔 Powered by TCPDF (www.tcpdf.org) Title 終身雇用制 : 再考 : ニューラルネットワークによる財務業績の検証 Sub Title Author 岡本, 大輔 (Okamoto, Daisuke) Publisher 慶應義塾大学出版会 Publication year 2010 Jtitle 三田商学研究 (Mita business review). Vol.53,

More information

Microsoft PowerPoint - #07 Quiz Are you still with me .pptx

Microsoft PowerPoint - #07 Quiz Are you still with me .pptx Quiz: Are You Still With Me? Takumi Information Technology Maiko Akutagawa Copyright Takumi Information Techonology CORPORATION, All rights reserved. What s Direct Identifier? -level1 Is it compulsory

More information

Copyright 2008 NIFTY Corporation All rights reserved. 2

Copyright 2008 NIFTY Corporation All rights reserved. 2 Copyright 2008 NIFTY Corporation All rights reserved. 2 Copyright 2008 NIFTY Corporation All rights reserved. 3 Copyright 2008 NIFTY Corporation All rights reserved. 4 Copyright 2008 NIFTY Corporation

More information

各学科 課程 専攻別開設授業科目 ( 教職関係 ) 総合情報学科 ( 昼間コース ) 中学校教諭 1 種免許状 ( 数学 ) 高等学校教諭 1 種免許状 ( 数学 ) 代数学 線形代数学第一 2 線形代数学第二 2 離散数学 2 応用代数学 2 オペレーションズ リサーチ基礎 2 数論アルゴリズム

各学科 課程 専攻別開設授業科目 ( 教職関係 ) 総合情報学科 ( 昼間コース ) 中学校教諭 1 種免許状 ( 数学 ) 高等学校教諭 1 種免許状 ( 数学 ) 代数学 線形代数学第一 2 線形代数学第二 2 離散数学 2 応用代数学 2 オペレーションズ リサーチ基礎 2 数論アルゴリズム 免許状取得に必要な履修科目 教育職員免許法施行規則に 左に該当する本学の 履修 高等学校教諭 高等学校教諭 中学校教諭 定める修得を要する科目 開設科目及び単位数 年次 専修免許状 1 種免許状 1 種免許状 教職の意義等に関する科目教職論 2 1 年 2 単位 2 単位 2 単位 教 教育原理 2 1 年 職 に教育の基礎理論に関する科教育心理学 2 1 年 6 単位 6 単位 6 単位 関目 す

More information

0801391,繊維学会ファイバ12月号/報文-01-西川

0801391,繊維学会ファイバ12月号/報文-01-西川 Pattern Making Method and Evaluation by Dots of Monochrome Shigekazu Nishikawa 1,MarikoYoshizumi 1,andHajime Miyake 2 1 Miyagi University of Education, 149, Aramaki-aza-Aoba, Aoba-ku, Sendai-shi, Miyagi

More information

Deep Learningでの地図タイル活用の検討

Deep Learningでの地図タイル活用の検討 第 7 回地理院地図パートナーネットワーク会議 2017/6/8 Deep Learning での 地図タイル活用の検討 OSGeo 財団日本支部 岩崎亘典 和山亮介 1 はじめに 発表内容 2 /36 汎用的フォーマットとしての地図タイル 地図タイルと Deep Learning CNN を用いた旧版地形図の分類 地形図から土地利用分類 Conditional GAN を用いたタイル画像変換 空中写真

More information

<8ED089EF8B D312D30914F95742E696E6464>

<8ED089EF8B D312D30914F95742E696E6464> * 1 The problem of privacy in the news of newspapers and weekly magazines The analysis of the articles on the child serial killing incidents in Kobe Akihiko SHIMAZAKI Kenzo SHIDA Toshihiko KATANO *2 1

More information

23 The Study of support narrowing down goods on electronic commerce sites

23 The Study of support narrowing down goods on electronic commerce sites 23 The Study of support narrowing down goods on electronic commerce sites 1120256 2012 3 15 i Abstract The Study of support narrowing down goods on electronic commerce sites Masaki HASHIMURA Recently,

More information

0 部分的最小二乗回帰 Partial Least Squares Regression PLS 明治大学理 学部応用化学科 データ化学 学研究室 弘昌

0 部分的最小二乗回帰 Partial Least Squares Regression PLS 明治大学理 学部応用化学科 データ化学 学研究室 弘昌 0 部分的最小二乗回帰 Parial Leas Squares Regressio PLS 明治大学理 学部応用化学科 データ化学 学研究室 弘昌 部分的最小二乗回帰 (PLS) とは? 部分的最小二乗回帰 (Parial Leas Squares Regressio, PLS) 線形の回帰分析手法の つ 説明変数 ( 記述 ) の数がサンプルの数より多くても計算可能 回帰式を作るときにノイズの影響を受けにくい

More information

FAX-760CLT

FAX-760CLT FAX-760CLT ;; yy 1 f a n l p w s m t v y k u c j 09,. i 09 V X Q ( < N > O P Z R Q: W Y M S T U V 1 2 3 4 2 1 1 2 1 2 j 11 dd e i j i 1 ; 3 oo c o 1 2 3 4 5 6 j12 00 9 i 0 9 i 0 9 i 0 9 i oo

More information

要旨 1. 始めに PCA 2. 不偏分散, 分散, 共分散 N N 49

要旨 1. 始めに PCA 2. 不偏分散, 分散, 共分散 N N 49 要旨 1. 始めに PCA 2. 不偏分散, 分散, 共分散 N N 49 N N Web x x y x x x y x y x y N 三井信宏 : 統計の落とし穴と蜘蛛の糸,https://www.yodosha.co.jp/jikkenigaku/statistics_pitfall/pitfall_.html 50 標本分散 不偏分散 図 1: 不偏分散のほうが母集団の分散に近付くことを示すシミュレーション

More information

kubostat2017b p.1 agenda I 2017 (b) probability distribution and maximum likelihood estimation :

kubostat2017b p.1 agenda I 2017 (b) probability distribution and maximum likelihood estimation : kubostat2017b p.1 agenda I 2017 (b) probabilit distribution and maimum likelihood estimation kubo@ees.hokudai.ac.jp http://goo.gl/76c4i 2017 11 14 : 2017 11 07 15:43 1 : 2 3? 4 kubostat2017b (http://goo.gl/76c4i)

More information

Clustering in Time and Periodicity of Strong Earthquakes in Tokyo Masami OKADA Kobe Marine Observatory (Received on March 30, 1977) The clustering in time and periodicity of earthquake occurrence are investigated

More information

206“ƒŁ\”ƒ-fl_“H„¤‰ZŁñ

206“ƒŁ\”ƒ-fl_“H„¤‰ZŁñ 51 206 51 63 2007 GIS 51 1 60 52 2 60 1 52 3 61 2 52 61 3 58 61 4 58 Summary 63 60 20022005 2004 40km 7,10025 2002 2005 19 3 19 GIS 2005GIS 2006 2002 2004 GIS 52 2062007 1 2004 GIS Fig.1 GIS ESRIArcView

More information

VE-GD21DL_DW_ZB

VE-GD21DL_DW_ZB V E-G D21D L V E-G D21D W 1 2 3 4 1 2 1 2 1 2 2 1 2 3 1 2 3 1 2 3 1 4 4 2 3 5 5 1 2 3 4 1 2 3 1 2 3 4 1 2 3 2006 Copyrights VisionInc. @. _ & $ % + = ^ 2011

More information

x T = (x 1,, x M ) x T x M K C 1,, C K 22 x w y 1: 2 2

x T = (x 1,, x M ) x T x M K C 1,, C K 22 x w y 1: 2 2 Takio Kurita Neurosceince Research Institute, National Institute of Advanced Indastrial Science and Technology takio-kurita@aistgojp (Support Vector Machine, SVM) 1 (Support Vector Machine, SVM) ( ) 2

More information