AI 클라우드로 완전 정복하기 - 데이터 분석부터 딥러닝까지 (윤석찬, AWS테크에반젤리스트)

1. © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
,

2. S 1
• S
• S 1
S 1 2
• M 2
• . 2 1 A

3. .

5. Accelerated

6. 데이터베이스 관Dd 부담이 많습니다.
관계형 DB 는 확장성이 T지 않아요.
Had,,p 배o 및 관Dp기. 힘?니다.
기존 D)는 L잡p고 비싸고 느립니다.
상a DB는 고비a에 관D, 확장이 어B워요.
실W/ 데이터는 수집p고 분석p기 힘?니다.
데이터 클E징(E(L)을 좀더 T게 할 수 없을까요?
딥러닝 데이터 H델/배o를 좀 더 T게 p고 싶어요.
ü Amazon RDS
ü Amazon DynamoDB
ü Amazon EMR
ü Amazon Redshift
ü Amazon Aurora
ü Amazon Kinesis
ü AWS Glue
ü Amazon SageMaker

7. ,
A

9. Transactions
ERP
Data analysts
1 4
0 9
5
Amazon
Quicksight

10. 2 45 ) B e M
01) 3 W D cKa cG CS
: ) R D (
:4 ) : E D
https://aws.amazon.com/ko/solutions/case-studies/supercell/

11. Transactions
ERP
Data
Lake
expdp
Data Data analysts
Data Warehouse
Amazon Redshift
Direct Query
Amazon Athena
Data Storage
Amazon S3

12. Data
Lake
Business users
Transactions
ERP
Social media
Data
Stream
Capture
Amazon
Kinesis
Events
Amazon
QuickSight
Data Warehouse
Amazon Redshift
Stream Data
Amazon
ElasticSearch
Data Storage
Amazon S3

13. Raw Data
Amazon S3
ETL (Hadoop)
Amazon EMR
Triggered Code
Amazon Lambda
Staged Data
(Data Lake)
Amazon S3
ETL & Catalog Management
AWS Glue
Data Warehouse
Amazon Redshift
Triggered Code
Amazon Lambda

14. Transactions
Data scientists
Business users
Connected
devices
Data
Event
Insights
Data
Lake
ML / Analytics / DLWeb logs /
clickstream
Amazon SageMaker

15. A QUIET OFFICE
Amazon SageMaker
Image Classification
Amazon Rekognition
Image
CHAIR
LAPTOP
LAMP
DESK
97%
95%
88%
82%
Object Identification
WORKING!
<HISTORY>

16. Modern data architecture
Real-time engagement and interactive customer experiences
Transactions
ERP
Data analysts
Data scientists
Business users
Engagement platformsConnected
devices
Automation / events
Data
Event Action
Insights
Data
Lake
ML / Analytics
Predict / Recommend
AI Services
Social media
Web logs /
clickstream

17. .

18. AWS DEEP LEARNING AMI
Apache MXNet TensorFlowCaffe2 Torch KerasCNTK PyTorch GluonTheano
VISION
AWS DeepLensAmazon SageMaker
LANGUAGE
Amazon
Rekognition
Amazon
Polly Amazon Lex
Amazon Rekognition
Video
Amazon
Transcribe
Amazon
Comprehend
Alexa for
Business
VR/AR
Amazon Sumerian
Amazon Machine Learning Amazon EMR & SparkMechanical Turk
INSTANCES
GPU(G2/P2/P3) CPU (C5) FPGA (F1)
Amazon
Translate
( )

19. !
(
)

20. -
J N
)
( ((

21. -
J N
)
( ((
K-Means Clustering
Principal Component Analysis
Neural Topic Modelling
Factorization Machines
Linear Learner - Regression
XGBoost
Latent Dirichlet Allocation
Image Classification
Seq2Seq
Linear Learner - Classification
ALGORITHMS
Apache MXNet
TensorFlow
Caffe2, CNTK,
PyTorch, Torch
FRAMEWORKS

22. -
J N
)
( ((

23. -
H
J
N

24. K-Means Clustering
Principal Component Analysis
Neural Topic Modelling
Factorization Machines
Linear Learner - Regression
XGBoost
Latent Dirichlet Allocation
Image Classification
Seq2Seq
Linear Learner - Classification
BUILT
ALGORITHMS
Caffe2, CNTK, PyTorch,
Torch
IM Estimators in Spark
DEEP LEARNING
FRAMEWORKS
Bring Your Own Script
(IM builds the Container)
BRING YOUR OWN
MODEL
ML
Training
code
Fetch Training data
Save Model
Artifacts
Amazon ECR
Save Inference
Image
Amazon S3

25. https://nucleusresearch.com/research/single/guidebook-tensorflow-aws/
In analyzing the experiences of researchers supporting
more than 388unique projects, Nucleus found that 88
percent of cloud-based TensorFlow projects are
running on Amazon Web Services (AWS).
“

26. from sagemaker.tensorflow import TensorFlow
tf_estimator = TensorFlow(
entry_point="tf-train.py", role='SageMakerRole',
training_steps=10000, evaluation_steps=100,
train_instance_count=1, train_instance_type='ml.p2.xlarge’)
tf_estimator.fit('s3://bucket/path/to/training/data’)
from sagemaker.pytorch import Pytorch
pytorch_estimator = Pytorch(entry_point="pt-train.py",
framework_version=”0.4.0”, role='SageMakerRole',
train_instance_type="ml.p2.xlarge", train_instance_count=2,
hyperparameters={ 'epochs': 6, 'backend': 'gloo’ })
pytorch_estimator.fit("s3://my_bucket/my_training_data/")

27. -
H
J
N

28. -
H
J
N

29. predictor = tf_estimator.deploy(
initial_instance_count=1,
instance_type='ml.c4.xlarge')
predictor = mxnet_estimator.deploy(
deploy_instance_type="ml.p2.xlarge",
min_instances=1,
https://runtime.sagemaker.us-east-1.amazonaws.com/
endpoints/model-name/invocations
• BK A ID A
• A I

30. SageMaker
Notebooks
Training
Algorithm
SageMaker
Training
Amazon ECR
Code Commit
Code Pipeline
SageMaker
Hosting
Coco dataset
AWS
Lambda
API
Gateway
Build Train
Deploy
static website hosted on S3
Inference requests
Amazon S3
Amazon
Cloudfront
Web assets on
Cloudfront

31. -
•
• w I n
l 9 l NF
l
l T
• C x N S
• , C e ,
• 03 2 x
oMs r C
• N S
• C 8

32. sagemaker = boto3.client(service_name='sagemaker')
sagemaker.create_training_job(**training_params)
create_model_response = sage.create_model(
ModelName = model_name,
ExecutionRoleArn = role,
PrimaryContainer = primary_container)
endpoint_config_response = sage.create_endpoint_config(
EndpointConfigName = endpoint_config_name,
ProductionVariants=[{
'InstanceType':'ml.m4.xlarge',
'InitialInstanceCount':1,
'ModelName':model_name,
'VariantName':'AllTraffic'}])
endpoint_response = sagemaker.create_endpoint(
'EndpointName': endpoint_name,
'EndpointConfigName': endpoint_config_name
2
.
1 3
.

33. •
•
•
•

34. 1
4.75
8.5
12.25
16
1 4.75 8.5 12.25 16
Speedup(x)
# GPUs
Resnet 152
Inceptin V3
Alexnet
Ideal
P2.16xlarge (8 Nvidia Tesla K80 - 16 GPUs)
Synchronous SGD (Stochastic Gradient Descent)
91%
Efficiency
88%
Efficiency
16x P2.16xlarge by AWS CloudFormation
Mounted on Amazon EFS
# GPUs

35. ## train data
num_gpus = 4
gpus = [mx.gpu(i) for i in range(num_gpus)]
model = mx.model.FeedForward(
ctx = gpus,
symbol = softmax,
num_round = 20,
learning_rate = 0.01,
momentum = 0.9,
wd = 0.00001)
model.fit(X = train, eval_data = val,
batch_end_callback =
mx.callback.Speedometer(batch_size=batch_size))

36. 기반 예제
B : A I A AA
• ( A B
• . DD A DD A B A
• A A IBD A AD D AD
• -A D AD D : D
• BB A
• -. D A: D :
• /D BD D A
• - AD C D :
• A D
• D A D )A B D A A
)..
http://mxnet.io/
https://github.com/dmlc/mxnet
http://incubator.apache.org/projects/mxnet.html

37. http://gluon.mxnet.io
-
H
• ,X P b fd S
• ( C X g NT
MI ce
• ) A ) A A
A K a W
• A ,C C a

38. We plan to use Amazon SageMaker to train models
against petabytes of Earth observation imagery datasets
using hosted Jupyter notebooks, so DigitalGlobe's
Geospatial Big Data Platform (GBDX) users can just push a
button, create a model, and deploy it all within one
scalable distributed environment at scale.
- Dr. Walter Scott, CTO of Maxar Technologies and founder of DigitalGlobe

39. EC
: A C
“With Amazon SageMaker, we can accelerate our Artificial Intelligence
initiatives at scale by building and deploying our algorithms on the
platform. We will create novel large-scale machine learning and AI
algorithms and deploy them on this platform to solve complex
problems that can power prosperity for our customers."
- Ashok Srivastava, Chief Data Officer, Intuit

40. $$$$
$$$
$$
$
Minutes Hours Days Weeks Months
Single
Machine
Distributed, with
Strong Machines

41. $$$$
$$$
$$
$
Minutes Hours Days Weeks Months
EC2 + AMI
Amazon SageMaker
On-premise

42. !
AWS
Only
http://bit.ly/awskr-ml-credits
-
e - o n S Ug m M
m R . 21, 31 :
r L A : k a) .( m :