Machine Learning with Python

Machine Learning with Python ʙ1ZUIPOͰػցֶशͷσϞΞϓϦΛ࡞Δʙ Yohei Munesada

फఆ༸ฏ ΉͶͩ͞Α͏΁͍ (`T"$"%&.:50,:0ϝϯλʔ ϒϩάIUUQXXXZPIFJNOFU about me

corse catalog 㾎 1ZUIPOͱ͸ 㾎 1ZUIPOجຊฤ 㾎 Ϟδϡʔϧͱύοέʔδ 㾎
8FCεΫϨΠϐϯά 㾎 8FCαʔόʔ 㾎 1ZUIPOͱػցֶश Basic Advanced

a concept of the day ػցֶशͱ͍͑͹1ZUIPOͱݴΘΕΔͱ͜ΖΛମݧ͢Δ ػցֶशͷ࡞ۀεςοϓΛ1ZUIPOͰ࣮૷͢Δ ԋशΛ௨࣮ͯ͠૷ͯ͠ମݧ͢Δ

today’s demo https://goo.gl/KMgK3e

the contents ػցֶशͷ֓࿦ ػցֶशΛऔΓר͘1ZUIPOϥΠϒϥϦ 47.Λ༻͍ͨखॻ͖൑ఆΞϓϦΛ࡞Δ

Machine Leaning Basic

agenda ػցֶशͱ͸Կ͔ʁ ճؼͱ෼ྨ ڭࢣ͋Γͱڭࢣͳ͠ ػցֶशͷओͳͭͷछྨ ػցֶशͷεςοϓ

machine learning is … ? ?

machine learning is … ? େྔσʔλ͔ΒύλʔϯΛೝࣝ͢Δ ໌೔͕੖ΕΔ৚݅ͱ͸ʁ Πέϝϯʹೝఆ͞ΕΔύλʔϯͱ͸ʁ ໌೔ͷσΟζχʔϥϯυ͕ࠞΉ৚݅͸ʁ
ύλʔϯΛ࢖ͬͯɺະ஌ͷσʔλΛ༧ଌ͢Δ ͋ͷಈ͖ɺ΋͠΍ෆ৹ऀͰ͸ɾɾɾʁ ࠓि຤ͷ͓ళ΁ͷདྷ٬਺͸ʁ དྷ݄ͷిྗ࢖༻ྔ͸ʁ

ػցֶशͷϞσϧ͸ɺճؼ 3FHSFTTJPO ͱ෼ྨ $MBTTJpDBUJPO ʹେผ͞Ε·͢ ճؼ(Regression) ෼ྨ(Classiﬁcation) ೖྗ͞Εͨσʔλ͔Β਺஋Λ༧ଌ͢ΔϞσϧɻ ྫɿϢʔβʔͷߪೖֹ༧ଌɺϢʔβʔͷεϚϗར༻࣌ؒ༧ଌ
ೖྗ͞Εͨσʔλ͔Β෼ྨΛ༧ଌ͢ΔϞσϧɻ෼ྨث $MBTTJpFS ͱ΋ݺ͹ΕΔɻ ྫɿϢʔβʔ͕ߪೖ͢Δ͔൱͔ɺը૾ʹೣؚ͕·ΕΔ͔൱͔ɺखॻ͖਺஋ͷ஋͸Կ͔ʁ regression and classiﬁcation

ػցֶशͷϞσϧ͸ɺڭࢣ͋Γ 4VQFSWJTFE ͱڭࢣͳ͠ 6OTVQFSWJTFE ʹେผ͞Ε·͢ ڭࢣ͋Γ(Supervised) ڭࢣͳ͠(Unsupervised) ࣄલʹ༩͑ΒΕͨσʔλ τϨʔχϯάσʔλ
Λ࢖ֶͬͯशΛߦ͍ɺͦΕΛ΋ͱʹ༧ଌ͢Δɻ ྫɿઢܗճؼɺϩδεςΟοΫճؼɺ47.ɺχϡʔϥϧωοτɺܾఆ໦ɺFUD ࣄલσʔλͳ͠ʹɺ༩͑ΒΕͨະ஌ͳσʔλ͔ΒԿΒ͔ͷຊ࣭తͳߏ଄Λಋ͖ग़͢ɻ ྫɿΫϥελϦϯάɺओ੒෼෼ੳɺFUD supervised and unsupervised

ڭࢣ͋Γ Supervised ڭࢣͳ͠ Unsupervised ճؼ Regression A B ෼ྨ
Classiﬁcation C D 4 types of machine learning ଞʹ΋ڧԽֶश΍Ҩ఻తΞϧΰϦζϜ΋

ref: ڧԽֶश(Reinforcement Learning) Ҩ఻తΞϧΰϦζϜ(Genetic Algorithm) http://yoheim.net/work/01_maze/ https://goo.gl/3gS3fq https://goo.gl/POq5el

ػձֶश͸ҎԼͷεςοϓͰ࣮ߦ͠·͢ steps of machine learning ?

ػձֶश͸ҎԼͷεςοϓͰ࣮ߦ͠·͢ σʔλΛೖखͯ͠ཧղ͢Δ ΞϧΰϦζϜΛܾΊΔ ಛ௃ྔΛܾΊΔ σʔλͷΫϨϯδϯά੔ܗΛ͢Δ ֶश͢Δ ϞσϧΛධՁ͢Δ ϞσϧΛվળ͢Δ steps
of machine learning

for your information ˞Ҿ༻ɿӬా͞ΜߨٛࢿྉΑΓ

re:agenda ػցֶशͱ͸Կ͔ʁ ճؼͱ෼ྨ ڭࢣ͋Γͱڭࢣͳ͠ ػցֶशͷओͳͭͷछྨ ػցֶशͷεςοϓ

python libraries for ML

main python libraries

NumPy ߦྻ΍ϕΫτϧΛѻ͏͜ͱ͕Ͱ͖ɺ$ݴޠϨϕϧͰߴ଎ʹॲཧ͢Δ import numpy as np X = np.array([
[5,0,1], [1,6,0], [0,0,1] ]) X.shape X.T E = np.eye(X.shape[0]) X.dot(E)

Matplotlib %ϓϩοτΛߦ͏ϥΠϒϥϦͰ͢ import numpy as np import matplotlib.pylab as
plt x = np.linspace(-np.pi, np.pi, 200) plt.plot(x, np.sin(x)) plt.show()

http://scikit-learn.org/

hand-written digits recognition

agenda σʔλΛೖखͯ͠ཧղ͢Δ ΞϧΰϦζϜΛܾΊΔ ಛ௃ྔΛܾΊΔ σʔλͷΫϨϯδϯά੔ܗΛ͢Δ ֶश͢Δ ϞσϧΛධՁ͢Δ ϞσϧΛվળ͢Δ ֶश݁ՌΛอଘ͢Δ
8FCαʔόʔ͔Βར༻͢Δ

http://yann.lecun.com/exdb/mnist/

samples

download MNIST data ./*45σʔλ͸8FC͔Βऔಘ͢Δ͜ͱ͕Ͱ͖·͢ http://yann.lecun.com/exdb/mnist/ data_mnist train-images-idx3-… train-labels-idx1-… …

original format

ڭࢣ͋Γ Supervised ڭࢣͳ͠ Unsupervised ճؼ Regression A B ෼ྨ
Classiﬁcation C D re: 4 types of machine learning

Support Vector Machine ڭࢣ͋Γɺ෼ྨ໰୊Λղ͘ΞϧΰϦζϜͰ͢ ˚ ˚ ˚ ˚ ˚
˚ ˚ ˚ ˚ ˚ ˚ ˚ ˚ ˚ ˚ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋

Support Vector Machine ڭࢣ͋Γɺ෼ྨ໰୊Λղ͘ΞϧΰϦζϜͰ͢ ˚ ˚ ˚ ˚ ˚
˚ ˚ ˚ ˚ ˚ ˚ ˚ ˚ ˚ ˚ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ ̋ Ϛʔδϯͷ࠷େԽ

Wanna dive into deep ? https://goo.gl/JE8Vvx ػցֶशͷجૅʢ਺ࣜฤʣ https://goo.gl/ATGxjM αϙʔτϕΫλʔϚγϯ

feature is ػցֶशͷΞϧΰϦζϜʹೖྗ͢Δಛ௃ྔ GFBUVSF Ͱ͢ ?

feature is ػցֶशͷΞϧΰϦζϜʹೖྗ͢Δಛ௃ྔ GFBUVSF Ͱ͢ Ϟς཰ ϕʔε஋ ਎௕ ϧοΫε
ੑ֨ + = + + ൑ఆ݁Ռ ಛ௃ ಛ௃ ಛ௃

the features of MNIST ./*45ͷը૾ͷಛ௃ྔͱ͸ԿͰ͠ΐ͏͔ʁ
ʜ ϐΫηϧ͕ͦΕͧΕಛ௃ྔʢίʣ

data pre-processing ./*45σʔλ͔ΒɺϞσϧʹಡΈࠐ·ͤΔσʔλΛ࡞Γ·͢ $47ܗࣜʹม׵͢Δ αϯϓϧը૾Λग़ྗͯ͠ΈΔʢՃ޻݁Ռͷ֬ೝʣ

convert to csv format ./*45ͷόΠφϦʔσʔλΛ$47ܗࣜʹม׵͠·͢ ͕ɺͦͷલʹʜ

binary data with Python όΠφϦσʔλΛTUSVDUϞδϡʔϧ ඪ४ϥΠϒϥϦ Ͱѻ͍·͢ https://docs.python.org/ja/3/library/struct.html

binary data with Python όΠφϦσʔλΛTUSVDUϞδϡʔϧ ඪ४ϥΠϒϥϦ Ͱѻ͍·͢ import struct
f = open(“file.binary", "rb") num1, num2 = struct.unpack(">II", f.read(8))

big-endian and little-endian (ࢀߟ) ΤσΟΞϯʹ͍ͭͯɿhttp://www.ertl.jp/~takayuki/readings/info/no05.html 100

convert to csv format ./*45ͷόΠφϦʔσʔλΛ$47ܗࣜʹม׵͠·͢

˞ϥϕϧσʔλ

import struct import gzip # Read MNIST `label`. fpath
= "./data_mnist/train-labels-idx1-ubyte.gz" with gzip.open(fpath, "rb") as f: magic_number, img_count = struct.unpack(">II", f.read(8)) labels = [] for i in range(img_count): label = str(struct.unpack("B", f.read(1))[0]) labels.append(label) # Write as csv. outpath = './csv/train-labels.csv' with open(outpath, "w") as f: f.write("\n".join(labels)) ˞ϥϕϧσʔλ

˞ը૾σʔλ

˞ը૾σʔλ import struct import gzip # Read MNIST `images`.
fpath = "./data_mnist/train-images-idx3-ubyte.gz" with gzip.open(fpath, "rb") as f: _, img_count = struct.unpack(">II", f.read(8)) rows, cols = struct.unpack(">II", f.read(8)) images = [] for i in range(img_count): binary = f.read(rows * cols) images.append(",".join([str(b) for b in binary])) # Write as csv. outpath = './csv/train-images.csv' with open(outpath, "w") as f: f.write("\n".join(images))

output as images Ұ෦Λը૾ʹग़ྗͯ͠$47݁Ռ͕ਖ਼͍͜͠ͱΛ֬ೝ͠·͢ with open("./csv/train-images.csv") as f: images
= f.read().split("\n") for i, image in enumerate(images[:10]): with open("./image/%d.pgm" % i, "w") as f: s = "P2 28 28 255\n" s += " ".join(image.split(",")) f.write(s)

output as images Ұ෦Λը૾ʹग़ྗͯ͠$47݁Ռ͕ਖ਼͍͜͠ͱΛ֬ೝ͠·͢

use Support Vector Machine ͦΕͰ͸47.ͰύλʔϯೝࣝΛߦ͍͍ͨͱࢥ͍·͢ from sklearn import svm
# Load training data. with open("./csv/train-images.csv") as f: images = f.read().split("\n")[:500] with open("./csv/train-labels.csv") as f: labels = f.read().split("\n")[:500] # Convert data. images = [[int(i)/256 for i in image.split(",")] for image in images] labels = [int(l) for l in labels] # Use SVM. clf = svm.SVC() clf.fit(images, labels)

evaluate the model ςετσʔλΛ༻ֶ͍ͯशͨ݁͠Ռ͔Β༧ଌΛߦ͍ɺਫ਼౓ΛධՁ͠·͢ from sklearn import metrics test_images
= #ಡΈࠐΈॲཧ͸লུʢࠓճ͸500݅ಡΈࠐΉʣ# test_labels = #ಡΈࠐΈॲཧ͸লུʢࠓճ͸500݅ಡΈࠐΉʣ# # Predict. predict = clf.predict(test_images) # Show results. ac_score = metrics.accuracy_score(test_labels, predict) print("Accuracy:", ac_score) cl_report = metrics.classification_report(test_labels, predict) print(cl_report)

ςετσʔλΛ༻ֶ͍ͯशͨ݁͠Ռ͔Β༧ଌΛߦ͍ɺਫ਼౓ΛධՁ͠·͢ evaluate the model ద߹཰ͱ࠶ݱ཰ɿIUUQEIBUFOBOFKQ;FMMJKQ

Ͳ͏΍ͬͯվળ͠·͔͢ʁ how to improve the model

τϨʔχϯάσʔλΛ૿΍͢ ಛ௃Λ૿΍͢ ಛ௃ΛݮΒ͢ ਖ਼ଇԽ߲ͷӨڹ౓ʢЕʣΛ૿΍͢ ਖ਼ଇԽ߲ͷӨڹ౓ʢЕʣΛݮΒ͢ σʔλͷਖ਼نԽ how to improve
the model

ֶशσʔλΛˠ ݅ʹ૿΍͠·͢ from sklearn import svm # Load training
data. with open("./csv/train-images.csv") as f: images = f.read().split("\n")[:5000] with open("./csv/train-labels.csv") as f: labels = f.read().split("\n")[:5000] # Convert data. images = [[int(i)/256 for i in image.split(",")] for image in images] labels = [int(l) for l in labels] # Use SVM. clf = svm.SVC() clf.fit(images, labels) how to improve the model

evaluate the training result ֶशσʔλΛˠ ݅ʹ૿΍͢ͱɺਫ਼౓͕ˠʹվળ͠·ͨ͠

ֶश݁ՌΛอଘ͠ɺͷͪʹ8FCαʔόʔ͔Βར༻Ͱ͖ΔΑ͏ʹ͠·͢ save the training result # Save the training
result. from sklearn.externals import joblib joblib.dump(clf, "./result/svm.pkl") result svm.pkl

ઌ΄Ͳอଘֶͨ͠श݁ՌΛɺผͷϓϩάϥϜ͔Βར༻͠·͢ use the model from sklearn.externals import joblib #
Load the training result. pklfile = path.join("result", "svm.pkl") clf = joblib.load(pklfile) # Predict. predict = clf.predict([data]) number = str(predict.tolist()[0])

σʔλΛೖखͯ͠ཧղ͢Δ ΞϧΰϦζϜΛܾΊΔ ಛ௃ྔΛܾΊΔ σʔλͷΫϨϯδϯά੔ܗΛ͢Δ ֶश͢Δ ϞσϧΛධՁ͢Δ ϞσϧΛվળ͢Δ ֶश݁ՌΛอଘ͢Δ 8FCαʔόʔ͔Βར༻͢Δ
re:agenda

2nd demo app खॻ͖਺஋൑ఆΞϓϦΛ࡞Ζ͏

today’s demo https://goo.gl/KMgK3e

how it works 'MBTLΞϓϦέʔγϣϯ͔Βอଘֶͨ͠श݁ՌΛར༻͠·͢ http://localhost:5000 index.html /api/judge ?data=0,0,255,86,0,0,255,10,…
svm.pkl judge!! “3”

how to create ϨϙδτϦΛΫϩʔϯ͢Δ ϥΠϒϥϦҰཡΛಡΈࠐΉ ىಈͯ͠ΈΔ جૅ՝୊Λ࣮૷͢Δ ಈ࡞ςετΛ͢Δ https://github.com/yoheimune-python-lecture/hand-written-digit-recognition

Finish !

summary

summary 1ZUIPOºػցֶशͷମݧΛ͠·ͨ͠ 1ZUIPOͷʮ਺ߦͰ΍Γ͍ͨ͜ͱ͕Ͱ͖Δʯ΋ମݧͰ͖·͔ͨ͠ʁ ίʔυ͸ॻ͔ͳ͖Ό਎ʹ͔ͭͳ͍ʂ ָ͍͠1ZUIPOϥΠϑΛʂ

https://www.coursera.org/learn/machine-learning

next step https://goo.gl/ccYxtS https://goo.gl/POq5el

next step https://goo.gl/EXcphV https://goo.gl/51viVt

https://goo.gl/JE8Vvx next step

next step https://goo.gl/J6jXhp

Thank you http://www.yoheim.net @yoheiMune IUUQTqJDLSQN[N2,

Machine Learning with Python

Machine Learning with Python

More Decks by Yohei Munesada

Other Decks in Technology

Featured

Transcript