Overview Ensemble Model Director of TEAMLAB Sungchul Choi
|
|
|
- 진근 강
- 4 years ago
- Views:
Transcription
1 Overview Ensemble Model Director of TEAMLAB Sungchul Choi
2
3
4
5 Ensemble Model - 하나의모델이아니라여러개모델의투표로 Y값예측 - Regression 문제에서는평균값을예측함 - meta-classifier - stacking (meta-ensemble) 등으로발전 - 학습은오래걸리나성능이매우좋음 - Kaggle 의대세기법 (structed dataset)
6
7
8 Keywords - Vallila ensemble - Boosting - Bagging - Adaptive boosting (AdaBoost) - XGBoost - Light GBM
9 Human knowledge belongs to the world.
10 Voting classifier Ensemble Model Director of TEAMLAB Sungchul Choi
11 Voting classifier - 가장기본적인 Ensemble classifier - 여러개의 Model 의투표를통해최종선택을실시 - Majority voting or Vallila Ensemble 모델이라고부름
12 Test Instance Classifier 1 Classifier 2 Classifier 3 Classifier 4 True True True True True
13
14 sklearn.ensemble.votingclassifier Hard voting 의합 Sotf 확률의합
15
16
17 Template clf1 = LogisticRegression(random_state=1) clf2 = DecisionTreeClassifier(random_state=1) clf3 = GaussianNB() eclf = VotingClassifier( estimators=[('lr', clf1), ('rf', clf2), ('gnb', clf3)], voting='hard')
18 Human knowledge belongs to the world.
19 Bagging Ensemble Model Director of TEAMLAB Sungchul Choi
20 Sampling? - 단순히같은데이터셋으로만드는 Classifier? - 같은 dataset으로만든여러개의 tree? - Dataset 은모데이터의 Sampling의불과! - 다양한 sampling dataset으로다양한 classifier 를만들자 - Sampling à 다양한데이터셋에강한 robust classifier
21
22
23 Bootstrapping - 학습데이터에서임의의복원추출 à Subset 학습데이터 n 개를추출하는것
24 Bootstrap? - 신발에서혼자신을수있게끔돕는끈 - 일반적으로외부 Input 없이처음시작하는일을지칭 - 컴퓨터를처음동작할때 memory - 데이터를외부추가없이추출하는것
25 <latexit sha1_base64="ls/bhagtftw9d55b+nxhbzwb2vk=">aaaci3icbvblswmxgmzwv62vvy9egkwoh5anciogfl14rgaf0n2wbdbbhmyfjfmxlptfvphxvhhqihcp/hftdg/aohayzmxh8o0bcyavzx0zhzxvtfwn4mzpa3tnd8/cp2jjkbgenknei9fxsaschbspmok0ewuka5fttju6nfrtryoki8ihny6pe+bbyhxgsnjs37xcvtswkvfrty8wsvgwetnpnaoukd3pxrfopkfupb20irk+wbzq1gxwmacclegort+c2f5ekocginaszrdzsxjslbqjngylo5e0xmseb7sraygdkp10tmmgt7tiqt8s+oqkzttfeykopbwhrk4gwa3lojcv//o6ifivnzsfcajosoyp+qmhkoltwqdhbcwkjzxbrdd9v0igwbejdk0lxqjaxhmztm5qykqh+/ny/savowiowdgoaaquqb3cgqzoagkewst4bx/gi/fmtizpebrg5doh4a+m7x/zmkkg</latexit> <latexit sha1_base64="ls/bhagtftw9d55b+nxhbzwb2vk=">aaaci3icbvblswmxgmzwv62vvy9egkwoh5anciogfl14rgaf0n2wbdbbhmyfjfmxlptfvphxvhhqihcp/hftdg/aohayzmxh8o0bcyavzx0zhzxvtfwn4mzpa3tnd8/cp2jjkbgenknei9fxsaschbspmok0ewuka5fttju6nfrtryoki8ihny6pe+bbyhxgsnjs37xcvtswkvfrty8wsvgwetnpnaoukd3pxrfopkfupb20irk+wbzq1gxwmacclegort+c2f5ekocginaszrdzsxjslbqjngylo5e0xmseb7sraygdkp10tmmgt7tiqt8s+oqkzttfeykopbwhrk4gwa3lojcv//o6ifivnzsfcajosoyp+qmhkoltwqdhbcwkjzxbrdd9v0igwbejdk0lxqjaxhmztm5qykqh+/ny/savowiowdgoaaquqb3cgqzoagkewst4bx/gi/fmtizpebrg5doh4a+m7x/zmkkg</latexit> <latexit sha1_base64="ls/bhagtftw9d55b+nxhbzwb2vk=">aaaci3icbvblswmxgmzwv62vvy9egkwoh5anciogfl14rgaf0n2wbdbbhmyfjfmxlptfvphxvhhqihcp/hftdg/aohayzmxh8o0bcyavzx0zhzxvtfwn4mzpa3tnd8/cp2jjkbgenknei9fxsaschbspmok0ewuka5fttju6nfrtryoki8ihny6pe+bbyhxgsnjs37xcvtswkvfrty8wsvgwetnpnaoukd3pxrfopkfupb20irk+wbzq1gxwmacclegort+c2f5ekocginaszrdzsxjslbqjngylo5e0xmseb7sraygdkp10tmmgt7tiqt8s+oqkzttfeykopbwhrk4gwa3lojcv//o6ifivnzsfcajosoyp+qmhkoltwqdhbcwkjzxbrdd9v0igwbejdk0lxqjaxhmztm5qykqh+/ny/savowiowdgoaaquqb3cgqzoagkewst4bx/gi/fmtizpebrg5doh4a+m7x/zmkkg</latexit> <latexit sha1_base64="ls/bhagtftw9d55b+nxhbzwb2vk=">aaaci3icbvblswmxgmzwv62vvy9egkwoh5anciogfl14rgaf0n2wbdbbhmyfjfmxlptfvphxvhhqihcp/hftdg/aohayzmxh8o0bcyavzx0zhzxvtfwn4mzpa3tnd8/cp2jjkbgenknei9fxsaschbspmok0ewuka5fttju6nfrtryoki8ihny6pe+bbyhxgsnjs37xcvtswkvfrty8wsvgwetnpnaoukd3pxrfopkfupb20irk+wbzq1gxwmacclegort+c2f5ekocginaszrdzsxjslbqjngylo5e0xmseb7sraygdkp10tmmgt7tiqt8s+oqkzttfeykopbwhrk4gwa3lojcv//o6ifivnzsfcajosoyp+qmhkoltwqdhbcwkjzxbrdd9v0igwbejdk0lxqjaxhmztm5qykqh+/ny/savowiowdgoaaquqb3cgqzoagkewst4bx/gi/fmtizpebrg5doh4a+m7x/zmkkg</latexit> <latexit sha1_base64="6wryusouxvndbt7ffpn4dra30pi=">aaab8nicbvbns8naej3ur1q/qh69bivgqsqi6lhoxwmfwwttkjvnpl262q27e6ge/awvhhtx6q/x5r9x2+agrq8ghu/nmdmvtau36hnftmvtfwnzq7pd29nd2z+ohx51jco0zr2qhnk9kbgmugqd5chyl9wmjkfgj+hkduy/pjftujipoe1zkjcr5dgnbk3uh8sa0nwv8qgy1hte05vdxsv+srpqoj2sfw0irboesascgnp3vrsdngjkvlcinsgmswmdkbhrwypjwkyqz08u3dorrg6stc2j7lz9pzgtxjhpetrohodylhsz8t+vn2f8hercphkysrel4ky4qnzz/27enamoppyqqrm91avjylnam1lnhuavv7xkuhdn32v695en1k0zrxvo4btowycramedtkedfbq8wyu8oei8oo/ox6k14pqzx/ahzucphtqrzq==</latexit> <latexit sha1_base64="6wryusouxvndbt7ffpn4dra30pi=">aaab8nicbvbns8naej3ur1q/qh69bivgqsqi6lhoxwmfwwttkjvnpl262q27e6ge/awvhhtx6q/x5r9x2+agrq8ghu/nmdmvtau36hnftmvtfwnzq7pd29nd2z+ohx51jco0zr2qhnk9kbgmugqd5chyl9wmjkfgj+hkduy/pjftujipoe1zkjcr5dgnbk3uh8sa0nwv8qgy1hte05vdxsv+srpqoj2sfw0irboesascgnp3vrsdngjkvlcinsgmswmdkbhrwypjwkyqz08u3dorrg6stc2j7lz9pzgtxjhpetrohodylhsz8t+vn2f8hercphkysrel4ky4qnzz/27enamoppyqqrm91avjylnam1lnhuavv7xkuhdn32v695en1k0zrxvo4btowycramedtkedfbq8wyu8oei8oo/ox6k14pqzx/ahzucphtqrzq==</latexit> <latexit sha1_base64="6wryusouxvndbt7ffpn4dra30pi=">aaab8nicbvbns8naej3ur1q/qh69bivgqsqi6lhoxwmfwwttkjvnpl262q27e6ge/awvhhtx6q/x5r9x2+agrq8ghu/nmdmvtau36hnftmvtfwnzq7pd29nd2z+ohx51jco0zr2qhnk9kbgmugqd5chyl9wmjkfgj+hkduy/pjftujipoe1zkjcr5dgnbk3uh8sa0nwv8qgy1hte05vdxsv+srpqoj2sfw0irboesascgnp3vrsdngjkvlcinsgmswmdkbhrwypjwkyqz08u3dorrg6stc2j7lz9pzgtxjhpetrohodylhsz8t+vn2f8hercphkysrel4ky4qnzz/27enamoppyqqrm91avjylnam1lnhuavv7xkuhdn32v695en1k0zrxvo4btowycramedtkedfbq8wyu8oei8oo/ox6k14pqzx/ahzucphtqrzq==</latexit> <latexit sha1_base64="6wryusouxvndbt7ffpn4dra30pi=">aaab8nicbvbns8naej3ur1q/qh69bivgqsqi6lhoxwmfwwttkjvnpl262q27e6ge/awvhhtx6q/x5r9x2+agrq8ghu/nmdmvtau36hnftmvtfwnzq7pd29nd2z+ohx51jco0zr2qhnk9kbgmugqd5chyl9wmjkfgj+hkduy/pjftujipoe1zkjcr5dgnbk3uh8sa0nwv8qgy1hte05vdxsv+srpqoj2sfw0irboesascgnp3vrsdngjkvlcinsgmswmdkbhrwypjwkyqz08u3dorrg6stc2j7lz9pzgtxjhpetrohodylhsz8t+vn2f8hercphkysrel4ky4qnzz/27enamoppyqqrm91avjylnam1lnhuavv7xkuhdn32v695en1k0zrxvo4btowycramedtkedfbq8wyu8oei8oo/ox6k14pqzx/ahzucphtqrzq==</latexit> <latexit sha1_base64="gli0+x/t6jrjfjrwtueyoxsynzk=">aaab9hicbvbns8naej3ur1q/qh69lbbbiyurgh6lxjxwsb/qhrlzbnqlm03c3rrkyo/w4kerr/4yb/4bt20o2vpg4pheddpz/erwbvz32ymtrw9sbpw3kzu7e/sh1cojto5trvmlxijwxz9ojrhklconyn1emrl5gnx88d3m70yy0jywj2aamc8iq8ldtomxkocv+qeinmn5fusdas2tu3ogvyiluomczuh1qx/eni2ynfqqrxvytyyxewu4fsyv9fpneklhzmh6lkosme1l86nzdgavaiwxsiunmqu/jzisat2nfnszetpsy95m/m/rpsa88tiuk9qwsrelwlqge6nzaijgilejppyqqri9fdersskym1pfhocxx14l7cs6duv44arwuc3ikmmjnmi5yligbtxde1pa4qme4rxeninz4rw7h4vwklpmhmmfoj8/ye6r1w==</latexit> <latexit sha1_base64="gli0+x/t6jrjfjrwtueyoxsynzk=">aaab9hicbvbns8naej3ur1q/qh69lbbbiyurgh6lxjxwsb/qhrlzbnqlm03c3rrkyo/w4kerr/4yb/4bt20o2vpg4pheddpz/erwbvz32ymtrw9sbpw3kzu7e/sh1cojto5trvmlxijwxz9ojrhklconyn1emrl5gnx88d3m70yy0jywj2aamc8iq8ldtomxkocv+qeinmn5fusdas2tu3ogvyiluomczuh1qx/eni2ynfqqrxvytyyxewu4fsyv9fpneklhzmh6lkosme1l86nzdgavaiwxsiunmqu/jzisat2nfnszetpsy95m/m/rpsa88tiuk9qwsrelwlqge6nzaijgilejppyqqri9fdersskym1pfhocxx14l7cs6duv44arwuc3ikmmjnmi5yligbtxde1pa4qme4rxeninz4rw7h4vwklpmhmmfoj8/ye6r1w==</latexit> <latexit sha1_base64="gli0+x/t6jrjfjrwtueyoxsynzk=">aaab9hicbvbns8naej3ur1q/qh69lbbbiyurgh6lxjxwsb/qhrlzbnqlm03c3rrkyo/w4kerr/4yb/4bt20o2vpg4pheddpz/erwbvz32ymtrw9sbpw3kzu7e/sh1cojto5trvmlxijwxz9ojrhklconyn1emrl5gnx88d3m70yy0jywj2aamc8iq8ldtomxkocv+qeinmn5fusdas2tu3ogvyiluomczuh1qx/eni2ynfqqrxvytyyxewu4fsyv9fpneklhzmh6lkosme1l86nzdgavaiwxsiunmqu/jzisat2nfnszetpsy95m/m/rpsa88tiuk9qwsrelwlqge6nzaijgilejppyqqri9fdersskym1pfhocxx14l7cs6duv44arwuc3ikmmjnmi5yligbtxde1pa4qme4rxeninz4rw7h4vwklpmhmmfoj8/ye6r1w==</latexit> <latexit sha1_base64="gli0+x/t6jrjfjrwtueyoxsynzk=">aaab9hicbvbns8naej3ur1q/qh69lbbbiyurgh6lxjxwsb/qhrlzbnqlm03c3rrkyo/w4kerr/4yb/4bt20o2vpg4pheddpz/erwbvz32ymtrw9sbpw3kzu7e/sh1cojto5trvmlxijwxz9ojrhklconyn1emrl5gnx88d3m70yy0jywj2aamc8iq8ldtomxkocv+qeinmn5fusdas2tu3ogvyiluomczuh1qx/eni2ynfqqrxvytyyxewu4fsyv9fpneklhzmh6lkosme1l86nzdgavaiwxsiunmqu/jzisat2nfnszetpsy95m/m/rpsa88tiuk9qwsrelwlqge6nzaijgilejppyqqri9fdersskym1pfhocxx14l7cs6duv44arwuc3ikmmjnmi5yligbtxde1pa4qme4rxeninz4rw7h4vwklpmhmmfoj8/ye6r1w==</latexit>.632 bootstrap - 일반적으로전체데이터 S에서 n번데이터를추출할때 - 각데이터가나타날확률이 라는것에서유래 1 d 1 1 d 1 Y (1 1 d )=1 (1 1 d )d 1 e 1
26 Bagging - Bootstrap Aggregation - Bootstrap 의 subset sample 로모델 n개를학습 à 앙상블 - 기존앙상블과달리하나의모델에다양한데이터대입 - High variance(overfitting 이심함 ) 모델이적합 - Regressor( 평균 or median), Classifier 모두존재
27
28 Out of bag error - OOB error estimation - Bagging 실행시, bag에미포함데이터로성능측정 - Validation set 을처리하는방법과유사 - Bagging 성능측적을위한좋은지표
29 sklearn.ensemble.baggingclassifier base_estimator : object or None, optional (default=none) n_estimators : int, optional (default=10) max_samples : int or float, optional (default=1.0) max_features : int or float, optional (default=1.0) bootstrap : boolean, optional (default=true) bootstrap_features : boolean, optional (default=false) oob_score : bool warm_start : bool, optional (default=false) n_jobs : int, optional (default=1)
30 sklearn.ensemble.baggingregressor base_estimator : object or None, optional (default=none) n_estimators : int, optional (default=10) max_samples : int or float, optional (default=1.0) max_features : int or float, optional (default=1.0) bootstrap : boolean, optional (default=true) bootstrap_features : boolean, optional (default=false) oob_score : bool warm_start : bool, optional (default=false) n_jobs : int, optional (default=1)
31 Template clf = DecisionTreeClassifier(random_state=1) eclf = BaggingClassifier(clf, oob_score=true) params ={ "n_estimators" : [10,20,30,40,50,55], "max_samples" : [0.5,0.6,0.7,0.8,0.9,1] } grid = GridSearchCV(estimator=eclf, param_grid=params, cv=5)
32 Human knowledge belongs to the world.
33 Random Forest Ensemble Model Director of TEAMLAB Sungchul Choi
34 Random Forest - Bagging + Randomized decision tree - Variance 가높은 decision tree들의 ensemble - 여러개의나무 à Forest - 가장간단하면서높은성능을자랑하는대표적인모델 - Regressor 와 Classifier 모두지원
35 Random Forest
36 Random Forest - correlation 낮은 m 개의 subset data 로학습 - Tree 의구성은 binary 로구성 - Split 시검토대상 feature 를 random 하게 n 개선정 - 전체 feature 를 p 라할때, n = p 이면 bagging tree - Feature 의재사용이가능, n은 p 또는 p 3 - Variance 가높은트리 à last node 1 ~ 5
37 sklearn.ensemble.randomforestclassifier
38 sklearn.ensemble.randomforestregressor
39 Human knowledge belongs to the world.
40 Adaptive Boosting Ensemble Model Director of TEAMLAB Sungchul Choi
41 Boosting - 간단하면서도성능이매우높은앙상블기법 - 학습 Round 를진행하면서모델을생성 à 모델에의해각 Instance의 weight 를업데이트 - Instance weight가높은 instance 를중심으로모델을생성 - 해당모델들로앙상블모델을만듦 (meta-classifier) - 잘못분류된데이터를더잘분류해보자
42
43
44 Adaboost - Adapative Boosting - 매라운드마다 instance 의 weight 값을계산 - 틀리는 instance 의 weight up à weight 기준 resampling - Instance weight 합이클수록, model 의 weight 를줄임 - 기본분류기에입력값을변화시켜새로운분류기를만듦 - high-depth tree, NN에는적합치않음
45 <latexit sha1_base64="0soxrtgxz+enpkhqt9gqu2gtimu=">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</latexit> <latexit sha1_base64="0soxrtgxz+enpkhqt9gqu2gtimu=">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</latexit> <latexit sha1_base64="0soxrtgxz+enpkhqt9gqu2gtimu=">aaacz3icbvhrshwxfm1mbwthw7e1incxq0vlcrlmfkf9ear9ql4pucrslemme2cnzjlt5i64doth+uz7x/oxza4j6g4vbe7opecmoulljs2f4b3nv1h6+er18ptgzfxtu7xw+w9ntqimwj4ovgeuum5rsy09kqtwojti81thexr1c9o/v0zjzafpavziiocjltmpodkqad0edcaj3ivfsd65setoduxdabcnojk6fm6ynbg9qahfica8ovpmmafncsphf8txtbqtimknv+d0mephmzpptcnuoctybfed2qyp46r1fw8luewosshubt8ksxru3jaucidbxfksubjii+w7qhmodldpcpraz8cmisumw5pgxj511dy3dpyntplzurtzvsn5v16/ouz7oja6rai1edgoqxrqadpqysgncljjb7gw0t0vxcu3xjd7mscfem0/ergcfe1gytc62wsf/gjiwgaf2dbrsih9ywfskb2zhhpsjxd4695h76+/5m/4mw9s32s86+xz+vv/abqusoq=</latexit> <latexit sha1_base64="0soxrtgxz+enpkhqt9gqu2gtimu=">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</latexit> Adaboost 값샘플의 weight 를 1/N 로초기화 M 은모델의갯수 N 은데이터의갯수 Weight 값을기준으로분류기생성 (resampling) 해당분류기의에러계산 I(y i,g m (x i )) = ( 0 if y i = G m (x i ) 1 if y i 6= G m (x i ) 해당분류기의분류기가중치생성 Instance 의 weight 업데이트
46 Adaboost 값샘플의 weight 를 1/N 로초기화 M 은모델의갯수 N 은데이터의갯수 Weight 값을기준으로분류기생성 (resampling) 해당분류기의에러계산 분류기의 weight
47 값샘플의 weight를 1/d로초기화 k는모델의갯수 D는데이터의갯수데이터샘플링 Error 가 0.5 이상은해당모델을버림 Update 방법이다름 Alpha 값이 weight
48 Adaboost
49 Adaboost with stump - Adaboost의 classifier를 1-depth tree를 사용하는 기법
50 Bagging vs. Boosting
51 Bagging vs. Boosting - 병렬화여부의차이 - bagging은 variance 가높은 base estimator 를 - boosting 은 bias가높은 base estimator 를사용 - boosting 은비용이매우높은알고리즘 - bagging은기본적으로데이터의 subset à boosting 보다좋은성능을내기는어려움
52 sklearn.ensemble.adaboostclassifier
53 sklearn.ensemble.adaboostregressor
54 Adaboost - 가장인기있는 boosting 알고리즘 + 높은성능 - 비슷한결과를내는 base-estimator à bagging 가유사함 - Weak classifier 의성능을극적으로향상 à 계속틀리는 instance 를맞추는순간 model 중요도 up - 특정문제에대해서는데이터와기본학습자가중요
55 Human knowledge belongs to the world.
56 Gradient boosting Ensemble Model Director of TEAMLAB Sungchul Choi
57 Gradient Boosting - Adaboost 와같은 boosting 기법의일종 - Regression, Classification 등모두사용 - Sequential + Additive Model - 이전모델의 Residual 를가지고 weak learner 를강화함 - residual 을예측하는형태의모델 - classification은 K-L Divergence 를사용
58
59
60
61
62 예측모델의값으로 0 으로초기화, r 은 y 의원래값 Subsampling 또는 Constraints 를가진트리모델로 r 과 X 를 fitting 새로운모델 = 이전모델 + learning_rate * 위에서적합된모델 현재의 residual - learning_rate * 적합된모델
63 from sklearn.tree import DecisionTreeRegressor tree_reg1 = DecisionTreeRegressor(max_depth=2) tree_reg1.fit(x, y) r1 = y - tree_reg1.predict(x) tree_reg2 = DecisionTreeRegressor(max_depth=2) tree_reg2.fit(x, r1) r2 = r1 - tree_reg2.predict(x) tree_reg3 = DecisionTreeRegressor(max_depth=2) tree_reg3.fit(x, r2) y_pred = sum(tree.predict(x_new) for tree in (tree_reg1, tree_reg2, tree_reg3))
64 <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> ŷ (0) i =0 ŷ (1) i = f 1 (x i )=ŷ (0) i + f 1 (x i ) ŷ (2) i = f 1 (x i )+f 2 (x i )=ŷ (1) i + f 2 (x i )... tx ŷ (t) i = k=1 (t 1) f k (x i )=ŷ i + f t (x) M(x, y from sklearn.tree import DecisionTreeRegressor tree_reg1 = DecisionTreeRegressor(max_depth=2) tree_reg1.fit(x, y) r1 = y - tree_reg1.predict(x) tree_reg2 = DecisionTreeRegressor(max_depth=2) tree_reg2.fit(x, r1) tx k=1 f k 1 (x)) = f t (x) r2 = r1 - tree_reg2.predict(x) tree_reg3 = DecisionTreeRegressor(max_depth=2) tree_reg3.fit(x, r2) y_pred = sum(tree.predict(x_new) for tree in (tree_reg1, tree_reg2, tree_reg3))
65 <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> (x, y tx k=1 f k 1 (x)) = f t (x) ŷ (t) i = tx k=1 (t 1) f k (x i )=ŷ i + f t (x)
66 <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="mohea0psyihayoqooao9sn+xe7i=">aaacvxichvfds8mwfe3rd/2a+uhlccgtdlrduarrfesgdwpob+ucazzqne1lkqql9e+kl/4b03vo3qqvhbzouefemxsvylqq2/4wzlhxicmp6rlrdm5+ybg0thwtw1hg0sahc0xtq5iwyklducvimxiebr4jn97tsa7fpbmhacivvbkrdoduofuprkptndl7bjyvjfvwtpk6uqkpooslhnnkarcl6rawpbumuq61aeu5lupgp6mnfcy3qkjrxga9+8z5vjaoo6chyvxhyg89rfj0+jphc9v/dbostqlsv+1ewfhg9eez9ooiu3pzuygoa8ivzkjklmnhqp3mvtejmexgkkqip6f70tkqo4didtrbfgbxndoffij04qr22j+ofavsjognmwokhuswlpn/aa1y+xvtlpiovotjopefm6hcmh8l7fjbsgkjbgglqmef+ahp5sj94fksnoenj4lrwtwxq87ltvnoul+oabak1kafogaxhiezcaeaabv7xp1bjufz0cqmm3mrahp9zwr4febljy8vzhq=</latexit> <latexit sha1_base64="mohea0psyihayoqooao9sn+xe7i=">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</latexit> <latexit sha1_base64="mohea0psyihayoqooao9sn+xe7i=">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</latexit> <latexit sha1_base64="mohea0psyihayoqooao9sn+xe7i=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">aaac53icbvlpa9swfjbdbuu8x2l33eu0tcsmbssmtkugbjcdo1jaqpwawzytevk20vnomp4heumhy+y6f2m3/tolcukw1u0dwcf33ve99yrfurqgfp+/425spnn6bou59+llq9dvots7ryyrnonjlslmn0tucckuh4mayu9yzwkasx4clb7w+eofxburqr+wzpk0ptmleseowcrsxaurnwlvgtsmqmboovxwotgte36/wvsj7aebd5cmazojse8sfh08wm3r+9tkszpssevc4qrenygnyfdgzd+imzatn1i7baziw3ixitupwm3iet/4comivbn+wfxcrot5yafrd/xv4ieankclmjgmo//slkxiuqimqtet4ucwlakgwssvvkawpkdsqwd8yqgiktftcvvofd6ztiyttnujak/yu4qspsys08hwphtmpp2rycdykwksz9nsqlwarti6uvjidbmuhx3hqnmgcmkbzvrywtgbu00z2k9rxwjpr/wqha0hxb+q7x+7b1+a69hc79au6igcpqed9a0dojfituyco5fol1e4f+5v98+61huazvt0l9y/1+qn4zc=</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> M(x, y tx k=1 f k 1 (x)) = f t (x) ŷ (t) i = tx k=1 (t 1) f k (x i )=ŷ i + f t (x) L(y, F(x)) = (y F (x))2 2 J = X (x) P L(y, y F (x) (x) = F (x) y
67 <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="xeocsord3gqj4fxztn59yh5zdxk=">aaace3icbzdlsgmxfiyz9vbhw9wlm2arwrflrgtdfipu3agv7axawjjppg3nxejosmsw7+dgv3hjqhg3btz5nqaxhbb+epjyn3nizu+egiuwrg8jtbs8srqwxjc3nre2dzk7ezuvrjkykg1eibsouuxwn1wbg2cnudlioylvnchvuf5/yflxwl+ducjahun53owugly6mwpzjjc8wancs0vejx6u7oqeskupyce5yt6ps/ogmsxojmsvrynwitgzykkzkp3mv6sb0mhjplbblgravgjtmejgvldebewkhyqosi81nfrey6odt3zk8jf2utgnpd4+4in7eyimnlijz9gdhog+mq+nzf9qzqjci3bm/tac5tppq24kmar4hbducskoijegqixxf8w0tyshogmch2dpr7witdoibrxt27ns+xiwrxodoeouqzy6r2v0jsqoiih6rm/ofb0zt8al8w58tfttxmxmh/2r8fkd92abda==</latexit> <latexit sha1_base64="v/ck/lsb3zovnf3nie5nxj7tldg=">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</latexit> <latexit sha1_base64="v/ck/lsb3zovnf3nie5nxj7tldg=">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</latexit> <latexit sha1_base64="v/ck/lsb3zovnf3nie5nxj7tldg=">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</latexit> <latexit sha1_base64="v/ck/lsb3zovnf3nie5nxj7tldg=">aaackhicjvfbs8mwge3rbu6ncz76ehzkhjhafdydw6kg4pocm4o1ldrnnts9kktikp09/h/f/dem22dqfdwqodnfltmfmzaqpgf8aprc4tlysmm1vlze2disbtx6ik45jj0cs5gpxcqioxhpssozgssconbl5nenror44wvhgsbrgxwlxa7ru0r9ipfukln9u3acxmst7nfgtzmfh2ze3a6gxvqpd0hfcablleeef2u1rvdws6t5nxll5whnvok4pijb23zgzwv5u3uq4ftrrssya84tc0rqyio7p/puetfoqxjjzjaqq9nipj0vezajedlkbukqdtatgsoaozaioxsbmsm9pxjqj7k6kyrj9wtfhkihrqgrmkmkn8xpwch+fhum0m/bgy2svjiitwb5kymyhsv2oec5wzknfegyu/vwij+rmkqqhrymmd+/pe/6ry3tajn3j/xu5dsoetgbu6abthakuuag3ieewfpfo9boti5e09v6ux4xsdw1ac02+ab99hosa7mp</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> <latexit sha1_base64="htgqdnhuyxei49dxfjup/3fn8is=">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</latexit> M(x, y tx k=1 f k 1 (x)) = f t (x) ŷ (t) i = tx k=1 (t 1) f k (x i )=ŷ i + f t (x) F k (x) =F k 1 (x)+ f k = F k 1 (x)+ (y F k 1 (x)) = F k 1 k 1 (x) i := i functional gradient descent gradient descent with functions
68 Turning parameters - number of tree (estimators) - depth of tree - subsampling - shrinkage parameter λ - Fitting to low variance
69 loss : { ls, lad, huber, quantile }, optional (default= ls ) learning_rate : float, optional (default=0.1) n_estimators : int (default=100) max_depth : integer, optional (default=3) subsample : float, optional (default=1.0)
70 loss : { deviance, exponential }, optional (default= deviance ) learning_rate : float, optional (default=0.1) n_estimators : int (default=100) max_depth : integer, optional (default=3) subsample : float, optional (default=1.0)
71 Gradient Boosting - greedy algorithm - scale에강건한모델 - 다양한 loss function을지원 (huber) - Overfitting problem - Slow model & High computation resource
72 Human knowledge belongs to the world.
73 XGBoost & LightGBM Ensemble Model Director of TEAMLAB Sungchul Choi
74 Gradient boosting packages - GBM 연산량과병렬처리를위한패키지가존재 - 대표적인패키지는 XGBoost 와 LightGBM - XGBoost - extreme Gradient Boosting - LightGBM Light Gradient Boosting Machhine - 구현알고리즘은일부상이하나목표등은유사한패키지
75 XGBOOST Light GBM
76 XGBoost Light GBM
77 Instllation 1. Git 설치 ( 2. Cmake 설치 ( win) 3. Visual studio 2015 (win) 4. Brew (mac)
78 Parameters
79 Human knowledge belongs to the world.
80 Stacking Ensemble Model Director of TEAMLAB Sungchul Choi
81 Stacking - Stacked Generalization - meta ensemble à 여러모델의결과를묶어서예측 - 컴퓨터성능향상과함께최근대중화되고있음 - 키워드 : stacking kaggle stacknet - 복잡성으로인해완벽한파이썬구현체는아직없음
82 여러개의모델을생성 Stacking Subset 1 로만학습을함 데이터셋을나눔
83 Meta model 각모델의결과값이 Meta 모델의 feature 로학습됨
84 Stacking X_train, X_test, y_train, y_test = train_test_split( X, y, test_size=self.test_ratio) for estimator in self.base_estimators: estimator.fit(x_train, y_train) meta_train_set = np.array([estimator.predict(x_test) for estimator in self.base_estimators]).t self.meta_estimator.fit(meta_train_set, y_test)
85 Stacking def predict(self, X, y=none): meta_x = [] for estimator in self.base_estimators: meta_x.append(estimator.predict(x)) meta_x = np.array(meta_x).t return self.meta_estimator.predict(meta_x)
86 StackNet - Stacking 지원을위한 JAVA 기반구현체 - XGBoost, LightGBM, Scikit-learn등다양한구현체사용 - 복잡한설치, 그러나 Kaggle 등에서높은성능을보여줌 - Neural net과대비하여더좋은성능, 쉬운학습이가능?
87
88 Human knowledge belongs to the world.
Overview Decision Tree Director of TEAMLAB Sungchul Choi
Overview Decision Tree Director of TEAMLAB Sungchul Choi 머신러닝의학습방법들 - Gradient descent based learning - Probability theory based learning - Information theory based learning - Distance similarity based
Probability Overview Naive Bayes Classifier Director of TEAMLAB Sungchul Choi
Probability Overview Naive Bayes Classifier Director of TEAMLAB Sungchul Choi 머신러닝의학습방법들 - Gradient descent based learning - Probability theory based learning - Information theory based learning - Distance
Tree 기반의 방법
Tree 기반의방법 박창이 서울시립대학교통계학과 박창이 ( 서울시립대학교통계학과 ) Tree 기반의방법 1 / 25 학습내용 의사결정나무 (decision tree) 회귀나무 (regresion tree) 분류나무 (classification tree) 비교앙상블알고리즘 (ensemble algorithm) 배깅 (bagging) 랜덤포레스트 (random
김기남_ATDC2016_160620_[키노트].key
![김기남_ATDC2016_160620_[키노트].key](/thumbs/85/91488855.jpg "김기남_ATDC2016_160620_[키노트].key")
metatron Enterprise Big Data SKT Metatron/Big Data Big Data Big Data... metatron Ready to Enterprise Big Data Big Data Big Data Big Data?? Data Raw. CRM SCM MES TCO Data & Store & Processing Computational
adfasdfasfdasfasfadf
C 4.5 Source code Pt.3 ISL / 강한솔 2019-04-10 Index Tree structure Build.h Tree.h St-thresh.h 2 Tree structure *Concpets : Node, Branch, Leaf, Subtree, Attribute, Attribute Value, Class Play, Don't Play.
슬라이드 1
빅데이터분석을위한데이터마이닝방법론 SAS Enterprise Miner 활용사례를중심으로 9 주차 예측모형에대한평가 Assessment of Predictive Model 최종후, 강현철 차례 6. 모형평가의기본개념 6.2 모델비교 (Model Comparison) 노드 6.3 임계치 (Cutoff) 노드 6.4 의사결정 (Decisions) 노드 6.5 기타모형화노드들
Manufacturing6
σ6 Six Sigma, it makes Better & Competitive - - 200138 : KOREA SiGMA MANAGEMENT C G Page 2 Function Method Measurement ( / Input Input : Man / Machine Man Machine Machine Man / Measurement Man Measurement
PowerPoint 프레젠테이션
I. 문서표준 1. 문서일반 (HY중고딕 11pt) 1-1. 파일명명체계 1-2. 문서등록정보 2. 표지표준 3. 개정이력표준 4. 목차표준 4-1. 목차슬라이드구성 4-2. 간지슬라이드구성 5. 일반표준 5-1. 번호매기기구성 5-2. 텍스트박스구성 5-3. 테이블구성 5-4. 칼라테이블구성 6. 적용예제 Machine Learning Credit Scoring
(JBE Vol. 23, No. 2, March 2018) (Special Paper) 23 2, (JBE Vol. 23, No. 2, March 2018) ISSN
(Special Paper) 23 2, 2018 3 (JBE Vol. 23, No. 2, March 2018) https://doi.org/10.5909/jbe.2018.23.2.186 ISSN 2287-9137 (Online) ISSN 1226-7953 (Print) a), a) Robust Online Object Tracking via Convolutional
23_Time-Series-Prediction
TensorFlow #23 (Time-Series Prediction) Magnus Erik Hvass Pedersen (http://www.hvass-labs.org/) / GitHub (https://github.com/hvass- Labs/TensorFlow-Tutorials) / Videos on YouTube (https://www.youtube.com/playlist?
홍익3월웹진PDF
C o n t e n t s 04 20 28 35 44 48 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 Human Resource Trends 50 Human Resource
홍익노사5월웹진용
C o n t e n t s 04 30 32 13 47 22 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 Human Resource Trends 49 50 Human Resource
Lab - Gradient descent Copyright 2018 by Introduction [PDF 파일다운로드 ]() 이번랩은우리가강의를통해들은 Gradient descent 을활용하여 LinearRegression
![Lab - Gradient descent Copyright 2018 by Introduction [PDF 파일다운로드 ]() 이번랩은우리가강의를통해들은 Gradient descent 을활용하여 LinearRegression](/thumbs/100/143645600.jpg "Lab - Gradient descent Copyright 2018 by Introduction [PDF 파일다운로드 ]() 이번랩은우리가강의를통해들은 Gradient descent 을활용하여 LinearRegression")
Lab - Gradient descent Copyright 2018 by teamlab.gachon@gmail.com Introduction [PDF 파일다운로드 ]() 이번랩은우리가강의를통해들은 Gradient descent 을활용하여 LinearRegression 모듈을구현하는것을목표로합니다. 앞서 우리가 Normal equation lab 을수행하였듯이,
methods.hwp
1. 교과목 개요 심리학 연구에 기저하는 기본 원리들을 이해하고, 다양한 심리학 연구설계(실험 및 비실험 설계)를 학습하여, 독립된 연구자로서의 기본적인 연구 설계 및 통계 분석능력을 함양한다. 2. 강의 목표 심리학 연구자로서 갖추어야 할 기본적인 지식들을 익힘을 목적으로 한다. 3. 강의 방법 강의, 토론, 조별 발표 4. 평가방법 중간고사 35%, 기말고사
An Effective Sentence-Extraction Technique Using Contextual Information and Statistical Approaches for Text Summarization
한국 BI 데이터마이닝학회 2010 추계학술대회 Random Forests 기법을사용한 저수율반도체웨이퍼검출및혐의설비탐색 고태훈, 김동일, 박은정, 조성준 * Data Mining Lab., Seoul National University, hooni915@snu.ac.kr Introduction 반도체웨이퍼의수율 반도체공정과웨이퍼의수율 반도체공정은수백개의프로세스로이루어져있음
PowerPoint 프레젠테이션
@ Lesson 2... ( ). ( ). @ vs. logic data method variable behavior attribute method field Flow (Type), ( ) member @ () : C program Method A ( ) Method B ( ) Method C () program : Java, C++, C# data @ Program
조사연구 권 호 연구논문 한국노동패널조사자료의분석을위한패널가중치산출및사용방안사례연구 A Case Study on Construction and Use of Longitudinal Weights for Korea Labor Income Panel Survey 2)3) a
조사연구 권 호 연구논문 한국노동패널조사자료의분석을위한패널가중치산출및사용방안사례연구 A Case Study on Construction and Use of Longitudinal Weights for Korea Labor Income Panel Survey 2)3) a) b) 조사연구 주제어 패널조사 횡단면가중치 종단면가중치 선형혼합모형 일반화선형혼 합모형
사회통계포럼
wcjang@snu.ac.kr Acknowledgements Dr. Roger Peng Coursera course. https://github.com/rdpeng/courses Creative Commons by Attribution /. 10 : SNS (twitter, facebook), (functional data) : (, ),, /Data Science
Secure Programming Lecture1 : Introduction
Malware and Vulnerability Analysis Lecture3-2 Malware Analysis #3-2 Agenda 안드로이드악성코드분석 악성코드분석 안드로이드악성코드정적분석 APK 추출 #1 adb 명령 안드로이드에설치된패키지리스트추출 adb shell pm list packages v0nui-macbook-pro-2:lecture3 v0n$
OPCTalk for Hitachi Ethernet 1 2. Path. DCOMwindow NT/2000 network server. Winsock update win95. . . 3 Excel CSV. Update Background Thread Client Command Queue Size Client Dynamic Scan Block Block
슬라이드 1
Pairwise Tool & Pairwise Test NuSRS 200511305 김성규 200511306 김성훈 200614164 김효석 200611124 유성배 200518036 곡진화 2 PICT Pairwise Tool - PICT Microsoft 의 Command-line 기반의 Free Software www.pairwise.org 에서다운로드후설치
Artificial Intelligence: Assignment 6 Seung-Hoon Na December 15, Sarsa와 Q-learning Windy Gridworld Windy Gridworld의 원문은 다음 Sutton 교재의 연습문제
Artificial Intelligence: Assignment 6 Seung-Hoon Na December 15, 2018 1 1.1 Sarsa와 Q-learning Windy Gridworld Windy Gridworld의 원문은 다음 Sutton 교재의 연습문제 6.5에서 찾아볼 수 있다. http://incompleteideas.net/book/bookdraft2017nov5.pdf
유니티 변수-함수.key
C# 1 or 16 (Binary or Hex) 1:1 C# C# (Java, Python, Go ) (0101010 ). (Variable) : (Value) (Variable) : (Value) ( ) (Variable) : (Value) ( ) ; (Variable) : (Value) ( ) ; = ; (Variable) : (Value) (Variable)
3 Gas Champion : MBB : IBM BCS PO : 2 BBc : : /45
3 Gas Champion : MBB : IBM BCS PO : 2 BBc : : 20049 0/45 Define ~ Analyze Define VOB KBI R 250 O 2 2.2% CBR Gas Dome 1290 CTQ KCI VOC Measure Process Data USL Target LSL Mean Sample N StDev (Within) StDev
김경재 안현철 지능정보연구제 17 권제 4 호 2011 년 12 월
지능정보연구제 17 권제 4 호 2011 년 12 월 (pp.241~254) Support vector machines(svm),, CRM. SVM,,., SVM,,.,,. SVM, SVM. SVM.. * 2009() (NRF-2009-327- B00212). 지능정보연구제 17 권제 4 호 2011 년 12 월 김경재 안현철 지능정보연구제 17 권제 4 호
ETL_project_best_practice1.ppt
ETL ETL Data,., Data Warehouse DataData Warehouse ETL tool/system: ETL, ETL Process Data Warehouse Platform Database, Access Method Data Source Data Operational Data Near Real-Time Data Modeling Refresh/Replication
15인플레이션01-목차1~9
ISSN 87-381 15. 1 15. 1 13 1 1.3 1. 1.8 1.5 1. 1.1 () 1.5 1..1 1.8 1.7 1.3 () 1..7.6...3 (). 1.5 3.6 3.3.9. 6.3 5.5 5.5 5.3.9.9 ().6.3.. 1.6 1. i 6 5 6 5 5 5 3 3 3 3 1 1 1 1-1 -1 13 1 1).6..3.1.3.
chap 5: Trees
5. Threaded Binary Tree 기본개념 n 개의노드를갖는이진트리에는 2n 개의링크가존재 2n 개의링크중에 n + 1 개의링크값은 null Null 링크를다른노드에대한포인터로대체 Threads Thread 의이용 ptr left_child = NULL 일경우, ptr left_child 를 ptr 의 inorder predecessor 를가리키도록변경
Artificial Intelligence: Assignment 5 Seung-Hoon Na December 15, Numpy: Tutorial 다음 자료를 참조하여 numpy기본을 공부하시오.
Artificial Intelligence: Assignment 5 Seung-Hoon Na December 15, 2018 1 Numpy: Tutorial 다음 자료를 참조하여 numpy기본을 공부하시오. https://docs.scipy.org/doc/numpy-1.15.0/user/quickstart.html https://www.machinelearningplus.com/python/
정보기술응용학회 발표
, hsh@bhknuackr, trademark21@koreacom 1370, +82-53-950-5440 - 476 - :,, VOC,, CBML - Abstract -,, VOC VOC VOC - 477 - - 478 - Cost- Center [2] VOC VOC, ( ) VOC - 479 - IT [7] Knowledge / Information Management
PowerPoint Presentation
컴퓨터비전 및 패턴인식 연구회 2009.2.12 Support Vector Machines http://cespc1.kumoh.ac.kr/~nonezero/svm ws cvpr.pdf 금오공과대학교 컴퓨터공학부 고재필 1 Contents Introduction Optimal Hyperplane Soft-Margin SVM Nonlinear SVM with Kernel
untitled
Math. Statistics: Statistics? 1 What is Statistics? 1. (collection), (summarization), (analyzing), (presentation) (information) (statistics).., Survey, :, : : QC, 6-sigma, Data Mining(CRM) (Econometrics)
Lecture12_Bayesian_Decision_Thoery
Bayesian Decision Theory Jeonghun Yoon Terms Random variable Bayes rule Classification Decision Theory Bayes classifier Conditional independence Naive Bayes Classifier Laplacian smoothing MLE / Likehood
Microsoft PowerPoint - PL_03-04.pptx
Copyright, 2011 H. Y. Kwak, Jeju National University. Kwak, Ho-Young http://cybertec.cheju.ac.kr Contents 1 프로그래밍 언어 소개 2 언어의 변천 3 프로그래밍 언어 설계 4 프로그래밍 언어의 구문과 구현 기법 5 6 7 컴파일러 개요 변수, 바인딩, 식 및 제어문 자료형 8
2002년 2학기 자료구조
자료구조 (Data Structures) Chapter 1 Basic Concepts Overview : Data (1) Data vs Information (2) Data Linear list( 선형리스트 ) - Sequential list : - Linked list : Nonlinear list( 비선형리스트 ) - Tree : - Graph : (3)
example code are examined in this stage The low pressure pressurizer reactor trip module of the Plant Protection System was programmed as subject for
2003 Development of the Software Generation Method using Model Driven Software Engineering Tool,,,,, Hoon-Seon Chang, Jae-Cheon Jung, Jae-Hack Kim Hee-Hwan Han, Do-Yeon Kim, Young-Woo Chang Wang Sik, Moon
Microsoft Word - [2017SMA][T8]OOPT_Stage_1000 ver2.docx
![Microsoft Word - [2017SMA][T8]OOPT_Stage_1000 ver2.docx](/thumbs/95/122981801.jpg "Microsoft Word - [2017SMA][T8]OOPT_Stage_1000 ver2.docx")
OOPT Stage 1000 - Plan & Elaboration Feesual CPT Tool Project Team T8 Date 2017-04-13 T8 Team Information 201211347 박성근 201211376 임제현 201411270 김태홍 2017 Team 8 1 Table of Contents 1 Activity 1001. Define
thesis
( Design and Implementation of a Generalized Management Information Repository Service for Network and System Management ) ssp@nile nile.postech.ac..ac.kr DPE Lab. 1997 12 16 GMIRS GMIRS GMIRS prototype
C# Programming Guide - Types
C# Programming Guide - Types 최도경 lifeisforu@wemade.com 이문서는 MSDN 의 Types 를요약하고보충한것입니다. http://msdn.microsoft.com/enus/library/ms173104(v=vs.100).aspx Types, Variables, and Values C# 은 type 에민감한언어이다. 모든
다운로드된 lab_normal_equation.zip 파일을작업폴더로이동한후압축해제후작업하시길바랍니다. 압축해제하면폴더가 linux_mac 과 windows 로나눠져있습니다. 자신의 OS에맞는폴더로이동해서코드를수정해주시기바랍니다. linear_model.py 코드 구조
Lab-Normalequation Copyright 2018 by teamlab.gachon@gmail.com Introduction [PDF 파일다운로드 ]() 이번랩은우리가강의를통해들은 Normal equation을활용하여 LinearRegression 모듈을구현하는것을목표로한다. LinearRegression 모듈의구현을위해서는 numpy와 Python
목 차
Oracle 9i Admim 1. Oracle RDBMS 1.1 (System Global Area:SGA) 1.1.1 (Shared Pool) 1.1.2 (Database Buffer Cache) 1.1.3 (Redo Log Buffer) 1.1.4 Java Pool Large Pool 1.2 Program Global Area (PGA) 1.3 Oracle
장기계획-내지4차
2011~2020 KOREA FOREST SERVICE 2011~2020 2011~2020 KOREA FOREST SERVICE 2011~2020 2011~2020 6 7 2011~2020 8 9 2011~2020 10 11 2011~2020 12 2011~2020 KOREA FOREST SERVICE 2011~2020 14 15 2011~2020 16 17
Mobile Service > IAP > Android SDK [ ] IAP SDK TOAST SDK. IAP SDK. Android Studio IDE Android SDK Version (API Level 10). Name Reference V
![Mobile Service > IAP > Android SDK [ ] IAP SDK TOAST SDK. IAP SDK. Android Studio IDE Android SDK Version (API Level 10). Name Reference V](/thumbs/103/160143957.jpg "Mobile Service > IAP > Android SDK [ ] IAP SDK TOAST SDK. IAP SDK. Android Studio IDE Android SDK Version (API Level 10). Name Reference V")
Mobile Service > IAP > Android SDK IAP SDK TOAST SDK. IAP SDK. Android Studio IDE 2.3.3 Android SDK Version 2.3.3 (API Level 10). Name Reference Version License okhttp http://square.github.io/okhttp/ 1.5.4
public key private key Encryption Algorithm Decryption Algorithm 1
public key private key Encryption Algorithm Decryption Algorithm 1 One-Way Function ( ) A function which is easy to compute in one direction, but difficult to invert - given x, y = f(x) is easy - given
Microsoft PowerPoint - 알고리즘_5주차_1차시.pptx
Basic Idea of External Sorting run 1 run 2 run 3 run 4 run 5 run 6 750 records 750 records 750 records 750 records 750 records 750 records run 1 run 2 run 3 1500 records 1500 records 1500 records run 1
확률과통계 강의자료-1.hwp
1. 통계학이란? 1.1 수학적 모형 실험 또는 증명을 통하여 자연현상을 분석하기 위한 수학적인 모형 1 결정모형 (deterministic model) - 뉴톤의 운동방정식 : - 보일-샤를의 법칙 : 일정량의 기체의 부피( )는 절대 온도()에 정비례하고, 압력( )에 반비례한다. 2 확률모형 (probabilistic model) - 주사위를 던질 때
Algorithm_Trading_Simple
Algorithm Trading Introduction DeepNumbers, 안명호 james@deepnumbers.com 1 3 4 5 적절한종목을선택하고매도와매수시기를알아내수익을실현하는것!!! 6 미국은 2012 년알고리즘트레이딩거래량이 85% 에달할만큼알고리즘트레이딩은가파르게증가 7 Goove WM, Zald DH등이작성한 Clinical versus
SchoolNet튜토리얼.PDF
Interoperability :,, Reusability: : Manageability : Accessibility :, LMS Durability : (Specifications), AICC (Aviation Industry CBT Committee) : 1988, /, LMS IMS : 1997EduCom NLII,,,,, ARIADNE (Alliance
DIY 챗봇 - LangCon
without Chatbot Builder & Deep Learning bage79@gmail.com Chatbot Builder (=Dialogue Manager),. We need different chatbot builders for various chatbot services. Chatbot builders can t call some external
Robust Segmentation for Sensor Data in Semiconductor Manufacturing 한국 BI 데이터마이닝학회 2012 춘계학술대회 박은정, 박주성, 양지원, 조성준 Seoul National University Industrial
Robust Segmentation for Sensor Data in Semiconductor Manufacturing 한국 BI 데이터마이닝학회 2012 춘계학술대회 박은정, 박주성, 양지원, 조성준 Seoul National University Industrial Engineering Department 1. Introduction 반도체공정에서는다양한분석및공정상태를모니터링에활용하기위한
Microsoft Word - [2017SMA][T8]OOPT_Stage_1000_ docx
![Microsoft Word - [2017SMA][T8]OOPT_Stage_1000_ docx](/thumbs/95/122981820.jpg "Microsoft Word - [2017SMA][T8]OOPT_Stage_1000_ docx")
OOPT Stage 1000 - Plan & Elaboration Feesual CPT Tool Project Team T8 Date 2017-03-30 T8 Team Information 201211347 박성근 201211376 임제현 201411270 김태홍 2017 Team 8 1 Table of Contents 1 Activity 1001. Define
02 C h a p t e r Java
02 C h a p t e r Java Bioinformatics in J a va,, 2 1,,,, C++, Python, (Java),,, (http://wwwbiojavaorg),, 13, 3D GUI,,, (Java programming language) (Sun Microsystems) 1995 1990 (green project) TV 22 CHAPTER
JUNIT 실습및발표
JUNIT 실습및발표 JUNIT 접속 www.junit.org DownLoad JUnit JavaDoc API Document 를참조 JUNIT 4.8.1 다운로드 설치파일 (jar 파일 ) 을다운로드 CLASSPATH 를설정 환경변수에서설정 실행할클래스에서 import JUnit 설치하기 테스트실행주석 @Test Test 를실행할 method 앞에붙임 expected
딥러닝 첫걸음
딥러닝첫걸음 4. 신경망과분류 (MultiClass) 다범주분류신경망 Categorization( 분류 ): 예측대상 = 범주 이진분류 : 예측대상범주가 2 가지인경우 출력층 node 1 개다층신경망분석 (3 장의내용 ) 다범주분류 : 예측대상범주가 3 가지이상인경우 출력층 node 2 개이상다층신경망분석 비용함수 : Softmax 함수사용 다범주분류신경망
SOSCON-MXNET_1014
딥러닝계의블루오션, Apache MXNet 공헌하기 윤석찬, Amazon Web Services 오규삼, 삼성 SDS SAMSUNG OPEN SOURCE CONFERENCE 018 목차 11 1 : 4 2 031 1 1 21 51: 1 L 1 S D A 1 S D N M Deep Learning 101 SAMSUNG OPEN SOURCE CONFERENCE
금오공대 컴퓨터공학전공 강의자료
C 프로그래밍프로젝트 Chap 14. 포인터와함수에대한이해 2013.10.09. 오병우 컴퓨터공학과 14-1 함수의인자로배열전달 기본적인인자의전달방식 값의복사에의한전달 val 10 a 10 11 Department of Computer Engineering 2 14-1 함수의인자로배열전달 배열의함수인자전달방식 배열이름 ( 배열주소, 포인터 ) 에의한전달 #include
UML
Introduction to UML Team. 5 2014/03/14 원스타 200611494 김성원 200810047 허태경 200811466 - Index - 1. UML이란? - 3 2. UML Diagram - 4 3. UML 표기법 - 17 4. GRAPPLE에 따른 UML 작성 과정 - 21 5. UML Tool Star UML - 32 6. 참조문헌
<C7A5C1F620BEE7BDC4>
연세대학교 상경대학 경제연구소 Economic Research Institute Yonsei Universit 서울시 서대문구 연세로 50 50 Yonsei-ro, Seodaemun-gS gu, Seoul, Korea TEL: (+82-2) 2123-4065 FAX: (+82- -2) 364-9149 E-mail: yeri4065@yonsei.ac. kr http://yeri.yonsei.ac.kr/new
EA0015: 컴파일러
5 Context-Free Grammar 무엇을공부하나? 앞에서배운 " 정규식 " 은언어의 " 어휘 (lexeme)" 를표현하는도구로사용되었다. 언어의 " 구문 (syntax)" 은 " 정규언어 " 의범위를벗어나기때문에 " 정규식 " 으로표현이불가능하다. 본장에서배우는 " 문맥자유문법 " 은언어의 " 구문 (syntax)" 을표현할수있는도구이다. 어떤 " 문맥자유문법
Evolutionary Optimization of a Collection of Variable-Length Subpatterns for Pattern Classification ( ) ( ) Robert Ian McKay ( )
Evolutionary Optimization of a Collection of Variable-Length Subpatterns for Pattern Classification 2008 8 Evolutionary Optimization of a Collection of Variable-Length Subpatterns for Pattern Classification
PowerPoint 프레젠테이션
Visual Search At SK-Planet sk-planet Machine Intelligence Lab. 나상일 1. 개발배경 2. 첫접근방법 3. 개선된방법 A. Visual recognition technology B. Guided search C. Retrieval system 개발배경 개발배경 상품검색을좀더쉽게 Key-word 트렌치코트버튺벨트
HW5 Exercise 1 (60pts) M interpreter with a simple type system M. M. M.., M (simple type system). M, M. M., M.
오늘할것 5 6 HW5 Exercise 1 (60pts) M interpreter with a simple type system M. M. M.., M (simple type system). M, M. M., M. Review: 5-2 7 7 17 5 4 3 4 OR 0 2 1 2 ~20 ~40 ~60 ~80 ~100 M 언어 e ::= const constant
슬라이드 1
Auto ML 과 XAI 를위한 H20 Driverless AI 소개 애자일소다컨설팅사업본부이동훈전무 2018.04 1. Intro 2. DAI 주요기능소개 & Demo 3. 결언 What are Auto ML & XAI? 데이타분석관련한기업의고민은 CIO 마케팅팀장 분석가 / 팀장 AI 나 ML 관련해서우리직원들의역량을어떻게끌어올려야할지? 이번에구축한시스템은우리직원들만의역량으로안정화와운영이가능할까?
Multi-pass Sieve를 이용한 한국어 상호참조해결 반-자동 태깅 도구
Siamese Neural Network 박천음 강원대학교 Intelligent Software Lab. Intelligent Software Lab. Intro. S2Net Siamese Neural Network(S2Net) 입력 text 들을 concept vector 로표현하기위함에기반 즉, similarity 를위해가중치가부여된 vector 로표현
1 : HEVC (Jeonghwan Heo et al.: Fast Partition Decision Using Rotation Forest for Intra-Frame Coding in HEVC Screen Content Coding Extension) (Regular
(Regular Paper) 23 1, 2018 1 (JBE Vol. 23, No. 1, January 2018) https://doi.org/10.5909/jbe.2018.23.1.115 ISSN 2287-9137 (Online) ISSN 1226-7953 (Print) HEVC a), a) Fast Partition Decision Using Rotation
SLA QoS
SLA QoS 2002. 12. 13 Email: really97@postech.ac.kr QoS QoS SLA POS-SLMS (-Service Level Monitoring System) SLA (Service Level Agreement) SLA SLA TM Forum SLA QoS QoS SLA SLA QoS QoS SLA POS-SLMS ( Service
C 프로그래밍 언어 입문 C 프로그래밍 언어 입문 김명호저 숭실대학교 출판국 머리말..... C, C++, Java, Fortran, Python, Ruby,.. C. C 1972. 40 C.. C. 1999 C99. C99. C. C. C., kmh ssu.ac.kr.. ,. 2013 12 Contents 1장 프로그래밍 시작 1.1 C 10 1.2 12
PowerPoint 프레젠테이션
@ Lesson 3 if, if else, if else if, switch case for, while, do while break, continue : System.in, args, JOptionPane for (,, ) @ vs. logic data method variable Data Data Flow (Type), ( ) @ Member field
1-1-basic-43p
A Basic Introduction to Artificial Neural Network (ANN) 도대체인공신경망이란무엇인가? INDEX. Introduction to Artificial neural networks 2. Perceptron 3. Backpropagation Neural Network 4. Hopfield memory 5. Self Organizing
4. #include <stdio.h> #include <stdlib.h> int main() { functiona(); } void functiona() { printf("hihi\n"); } warning: conflicting types for functiona
이름 : 학번 : A. True or False: 각각항목마다 True 인지 False 인지적으세요. 1. (Python:) randint 함수를사용하려면, random 모듈을 import 해야한다. 2. (Python:) '' (single quote) 는한글자를표현할때, (double quote) 는문자열을표현할때사용한다. B. 다음에러를수정하는방법을적으세요.
chap 5: Trees
Chapter 5. TREES 목차 1. Introduction 2. 이진트리 (Binary Trees) 3. 이진트리의순회 (Binary Tree Traversals) 4. 이진트리의추가연산 5. 스레드이진트리 (Threaded Binary Trees) 6. 히프 (Heaps) 7. 이진탐색트리 (Binary Search Trees) 8. 선택트리 (Selection
초보자를 위한 C# 21일 완성
C# 21., 21 C#., 2 ~ 3 21. 2 ~ 3 21.,. 1~ 2 (, ), C#.,,.,., 21..,.,,, 3. A..,,.,.. Q&A.. 24 C#,.NET.,.,.,. Visual C# Visual Studio.NET,..,. CD., www. TeachYour sel f CSharp. com., ( )., C#.. C# 1, 1. WEEK
탐색적데이터분석 (Exploratory Data Analysis) 데이터가지닌주요특성 / 개괄을 ( 우선적으로 ) 탐구함으로써 데이터분석을시도하려는형태 모델링이나가설을세우고이를검증하기보다데이터자체 가우리에게말하려고하는것을알아내는것의중요성을강 조하며시각화플롯을많이활용 J
탐색적데이터분석 Supervised Learning 탐색적데이터분석 (Exploratory Data Analysis) 데이터가지닌주요특성 / 개괄을 ( 우선적으로 ) 탐구함으로써 데이터분석을시도하려는형태 모델링이나가설을세우고이를검증하기보다데이터자체 가우리에게말하려고하는것을알아내는것의중요성을강 조하며시각화플롯을많이활용 John Tukey 가그중요성을강조 S 와
16
Vocation & Human Resource Development 15 16 Vocation & Human Resource Development 17 18 한 서비스와 품질향상을 위한 숙련의 심화가 요청되고 더욱 중요성을 갖게 되기 때문이다. 주5일 근무시대에 있다. 는 몇 시간 근로했는가 아니라 시간당 부가가치가 더욱 우리사회에는 근로자 1인당 연간 200시간의
°Ÿ»4º¨Ö
나무모형 이재용 서울대학교통계학과 2015 년 2 월 16 일 의사결정나무 I Years < 4.5 238 R 3 Hits R 1 117.5 5.11 Hits < 117.5 R 2 6.00 6.74 1 4.5 24 Years 1 Hitters 자료 1. ISLR 패키지에있는자료 2. 1986 년 322 명의메이저리그선수들에대한관측치. 20 개의변수 3. log(salary)
Microsoft Word - KSR2014S042
2014 년도 한국철도학회 춘계학술대회 논문집 KSR2014S042 안전소통을 위한 모바일 앱 서비스 개발 Development of Mobile APP Service for Safety Communication 김범승 *, 이규찬 *, 심재호 *, 김주희 *, 윤상식 **, 정경우 * Beom-Seung Kim *, Kyu-Chan Lee *, Jae-Ho
Ç¥Áö
Vocation & Human Resource Development 74 Vocation & Human Resource Development 75 76 Vocation & Human Resource Development 77 78 Vocation & Human Resource Development 79 80 Vocation & Human Resource Development
산선생의 집입니다. 환영해요
Biped Walking Robot Biped Walking Robot Simulation Program Down(Visual Studio 6.0 ) ). Version.,. Biped Walking Robot - Project Degree of Freedom : 12(,,, 12) :,, : Link. Kinematics. 1. Z (~ Diablo Set
비트와바이트 비트와바이트 비트 (Bit) : 2진수값하나 (0 또는 1) 를저장할수있는최소메모리공간 1비트 2비트 3비트... n비트 2^1 = 2개 2^2 = 4개 2^3 = 8개... 2^n 개 1 바이트는 8 비트 2 2
비트연산자 1 1 비트와바이트 비트와바이트 비트 (Bit) : 2진수값하나 (0 또는 1) 를저장할수있는최소메모리공간 1비트 2비트 3비트... n비트 2^1 = 2개 2^2 = 4개 2^3 = 8개... 2^n 개 1 바이트는 8 비트 2 2 진수법! 2, 10, 16, 8! 2 : 0~1 ( )! 10 : 0~9 ( )! 16 : 0~9, 9 a, b,
05-class.key
5 : 2 (method) (public) (private) (interface) 5.1 (Method), (public method) (private method) (constructor), 3 4 5.2 (client). (receiver)., System.out.println("Hello"); (client object) (receiver object)
2 : (Juhyeok Mun et al.: Visual Object Tracking by Using Multiple Random Walkers) (Special Paper) 21 6, (JBE Vol. 21, No. 6, November 2016) ht
(Special Paper) 21 6, 2016 11 (JBE Vol. 21, No. 6, November 2016) http://dx.doi.org/10.5909/jbe.2016.21.6.913 ISSN 2287-9137 (Online) ISSN 1226-7953 (Print) a), a), a) Visual Object Tracking by Using Multiple
2 : (EunJu Lee et al.: Speed-limit Sign Recognition Using Convolutional Neural Network Based on Random Forest). (Advanced Driver Assistant System, ADA
(JBE Vol. 20, No. 6, November 2015) (Regular Paper) 20 6, 2015 11 (JBE Vol. 20, No. 6, November 2015) http://dx.doi.org/10.5909/jbe.2015.20.6.938 ISSN 2287-9137 (Online) ISSN 1226-7953 (Print) a), a),
Microsoft PowerPoint - 3ÀÏ°_º¯¼ö¿Í »ó¼ö.ppt
변수와상수 1 변수란무엇인가? 변수 : 정보 (data) 를저장하는컴퓨터내의특정위치 ( 임시저장공간 ) 메모리, register 메모리주소 101 번지 102 번지 변수의크기에따라 주로 byte 단위 메모리 2 기본적인변수형및변수의크기 변수의크기 해당컴퓨터에서는항상일정 컴퓨터마다다를수있음 short
CATIA V5 SimDesigner KUBRIX Smarteam Satisfaction Static on demand Business Business process decision-making Rigid organizational structure Slow and steady economic growth Long-term product lifecycles
DBPIA-NURIMEDIA
ISSN 2383-6318(Print) / ISSN 2383-6326(Online) KIISE Transactions on Computing Practices, Vol 20, No 10, pp 549-554, 2014 10 http //dx doi org/10 5626/KTCP 2014 20 10 549 불균형데이터처리를위한과표본화기반앙상블학습기법 (Oversampling-Based
07 자바의 다양한 클래스.key
[ 07 ] . java.lang Object, Math, String, StringBuffer Byte, Short, Integer, Long, Float, Double, Boolean, Character. java.util Random, StringTokenizer Calendar, GregorianCalendar, Date. Collection, List,
02(848-853) SAV12-19.hwp
848 정보과학회논문지 : 소프트웨어 및 응용 제 39 권 제 11 호(2012.11) 3차원 객체인식을 위한 보완적 특징점 기반 기술자 (Complementary Feature-point-based Descriptors for 3D Object Recognition) 장영균 김 주 환 문 승 건 (Youngkyoon Jang) (Ju-Whan Kim) (Seung
기본자료형만으로이루어진인자를받아서함수를결과값으로반환하는고차함수 기본자료형과함수를인자와결과값에모두이용하는고차함수 다음절에서는여러가지예를통해서고차함수가어떤경우에유용한지를설명한다. 2 고차함수의 예??장에서대상체만바뀌고중간과정은동일한계산이반복될때함수를이용하면전체연산식을간 단
EECS-101 전자계산입문 고차함수 박성우 2008년5월 29일 지금까지정수나부동소수와같은기본적인자료형의조합을인자로받고결과값으로반환하는 함수에대해서배웠다. 이번강의에서는함수자체를다른함수의인자로이용하거나결과값으로 이용하는 방법을 배운다. 1 고차함수의 의미 계산은무엇을어떻게처리하여결과값을얻는지설명하는것으로이루어진다. 여기서 무엇 과 결 과값 은계산의대상체로서정수나부동소수와같은기본자료형의조합으로표현하며,
融合先验信息到三维重建 组会报 告[2]
![融合先验信息到三维重建 组会报 告[2]](/thumbs/99/140143252.jpg "融合先验信息到三维重建 组会报 告[2]")
[1] Crandall D, Owens A, Snavely N, et al. "Discrete-continuous optimization for large-scale structure from motion." (CVPR), 2011 [2] Crandall D, Owens A, Snavely N, et al. SfM with MRFs: Discrete-Continuous
歯4차학술대회원고(장지연).PDF
* 1)., Heckman Selection. 50.,. 1990 40, -. I.,., (the young old) (active aging). 1/3. 55 60 70.,. 2001 55 64 55%, 60%,,. 65 75%. 55 64 25%, 32% , 65 55%, 53% (, 2001)... 1998, 8% 41.5% ( 1998). 2002 7.8%
<4D6963726F736F667420576F7264202D20B1E2C8B9BDC3B8AEC1EE2DC0E5C7F5>
주간기술동향 2016. 5.18. 컴퓨터 비전과 인공지능 장혁 한국전자통신연구원 선임연구원 최근 많은 관심을 받고 있는 인공지능(Artificial Intelligence: AI)의 성과는 뇌의 작동 방식과 유사한 딥 러닝의 등장에 기인한 바가 크다. 이미 미국과 유럽 등 AI 선도국에서는 인공지능 연구에서 인간 뇌 이해의 중요성을 인식하고 관련 대형 프로젝트들을
_KrlGF발표자료_AI
AI 의과거와현재그리고내일 AI is the New Electricity 2017.09.15 AI! 2 Near Future of Super Intelligence? *source l http://www.motherjones.com/media/2013/05/robots-artificial-intelligence-jobs-automation 3 4 I think
DBPIA-NURIMEDIA
The e-business Studies Volume 17, Number 6, December, 30, 2016:3~20 Received: 2016/12/04, Accepted: 2016/12/27 Revised: 2016/12/27, Published: 2016/12/30 [ABSTRACT] This study aims to comprehensively analyze
Microsoft PowerPoint - additional01.ppt [호환 모드]
![Microsoft PowerPoint - additional01.ppt [호환 모드]](/thumbs/90/102753799.jpg "Microsoft PowerPoint - additional01.ppt [호환 모드]")
1.C 기반의 C++ part 1 함수 오버로딩 (overloading) 디폴트매개변수 (default parameter) 인-라인함수 (in-line function) 이름공간 (namespace) Jong Hyuk Park 함수 Jong Hyuk Park 함수오버로딩 (overloading) 함수오버로딩 (function overloading) C++ 언어에서는같은이름을가진여러개의함수를정의가능
01-OOPConcepts(2).PDF
Object-Oriented Programming Concepts Tel: 02-824-5768 E-mail: hhcho@selabsoongsilackr? OOP (Object) (Encapsulation) (Message) (Class) (Inheritance) (Polymorphism) (Abstract Class) (Interface) 2 1 + = (Dependency)
Delving Deeper into Convolutional Networks for Learning Video Representations - Nicolas Ballas, Li Yao, Chris Pal, Aaron Courville arXiv:
Delving Deeper into Convolutional Networks for Learning Video Representations Nicolas Ballas, Li Yao, Chris Pal, Aaron Courville arxiv: 1511.06432 Il Gu Yi DeepLAB in Modu Labs. June 13, 2016 Il Gu Yi
(, sta*s*cal disclosure control) - (Risk) and (U*lity) (Synthe*c Data) 4. 5.
1 (, ), ( ) 2 1. 2. (, sta*s*cal disclosure control) - (Risk) and (U*lity) - - 3. (Synthe*c Data) 4. 5. 3 1. + 4 1. 2.,. 3. K + [ ] 5 ' ', " ", " ". (SNS), '. K KT,, KG (PG), 'CSS'(Credit Scoring System)....,,,.
Microsoft PowerPoint - 07-Data Manipulation.pptx
Digital 3D Anthropometry 7. Data Analysis Sungmin Kim SEOUL NATIONAL UNIVERSITY Body 기본정보표시 Introduction 스케일조절하기 단면형상추출 단면정보관리 3D 단면형상표시 2 기본정보표시및스케일조절 UI 및핸들러구성 void fastcall TMainForm::BeginNewProject1Click(TObject
Problem New Case RETRIEVE Learned Case Retrieved Cases New Case RETAIN Tested/ Repaired Case Case-Base REVISE Solved Case REUSE Aamodt, A. and Plaza, E. (1994). Case-based reasoning; Foundational