data science, realizing the hype cycle. · § models are built using “training” computation...
TRANSCRIPT
Luigi Di Rito, Director Data Science Team, SAP Center of Excellence
Data Science, realizing the Hype Cycle.
2
Data Science, Machine Learning and Artificial Intelligence
AREAS OF AI
ArtificialIntelligence
Rule-based Reasoning
Machine Learning
Deep Learning
Natural Language Processing
Translation Machine Vision
Speech to Text
Speech
Text to Speech
Robotics
Autonomics Vehicles
Artificial Intelligence§ Intelligence exhibited by machines§ Broadly defined to include any simulation of human
intelligence§ Expanding and branching areas of research,
development and investment§ Includes robotics, rule-based reasoning, natural
language processing (NLP), knowledge representationtechniques (knowledge graphs), …
`̀
Data Science – akaPredictive/ AdvancedAnalytics§ Algorithmic and computational
techniques and tools forhandling large data sets
§ Increasingly focused onpreparing and modeling data forML and DL tasks
Machine Learning§ A subfield of AI which aims to teach computers the ability to do tasks with data,
without explicit programming§ Uses numerical and statistical approaches, including artificial neural network
techniques to encode learning§ Models are built using “training” computation runs, can also train through usage
Deep Learning§ A subfield of ML that uses specialized computational techniques,
typically multi-layer (2+) artificial neural networks§ Layering allows cascaded learning and abstraction levels (e.g. line
recognition -> shape -> object -> scene)§ Computationally intensive enabled by clouds, GPUs, and increasingly
more specialized HW such as FPGA and new custom hardware
3
Is Machine Learning something new?
§ Big Data (for example, cloud applications, the Internet of Things, Data Lakes)
§ Massive improvements in hardware (graphics processing unit [GPU] and multicore)
§ Deep learning and other new algorithms performing complex operations
§ Computers learn from data without being explicitly programmed.
§ Machine learning is not a new concept but is now becomingmainstream
What is machine learning?
Why now?
4
Typical Use Cases for Machine Learning
PredictiveMaintenance
PredictiveQuality
ChurnPrediction Demand /
Revenueforecasting
Price / VolumePrediction
PriceOptimization –
Elasticityunderstanding
FraudDetection
CustomerSegmentation
Root-causeanalysis
RiskManagement
5
Data Science is an iterative process
CRISP-DM | Cross Industry Standard Process for Data Mining
ModelingData
BusinessUnderstanding
DataUnderstanding
DataPreparation
Evaluation
Deployment
Monitoring
6
Is it all about Creating Models?
“Only 20% of the time is spent on modeling / algorithms” Rexer Analytics
7
Data PreparationAutomation of data acquisition andtransformation to create input datasets for data science use cases
Advanced AnalyticsUtilize SAP PA automated modelling, HANAPredictive Analytics Library, R, 3rd party and opensource ML, text analytics or geo spatial engines togenerate new business insight
Business Process IntegrationIntegrate results with existingbusiness processes or develop newbusiness applications using SCP, BI,HANA SQL and HTML5 (SAP UI5)
1
2
3
4
Access Storage Preparation VisualizationAnalysis Process Operate
Value
Data AccessIntegrate and transform data from flatfiles, relational data bases or Hadoop forconsumption in statistical models
Real scenarios require a Data Science Platform
8
The Data Scientists Profiles
The DataScientist
The DataArchitect
The DataEngineer/
Administrator
The Data /Analytics Manager
The Data / BusinessAnalyst
9
Business ExpertBusiness Analyst
Data Scientist
IT
Data Driven Application Development – Process Steps and Owners
BusinessUnderstanding
DataUnderstanding
DataPreparation
Modelling
Evaluation
Analysis(Production)Data Access Data
Preparation Visualization ProcessIntegration
Operate
SAP DataNon-SAPData
Deploy
Analysis(Exploration)
Data
10
INTELLIGENT APPS
Machine Learning @SAP. How do we explain to customers:
DEVELOPER,DATA SCIENTIST
BUSINESS ANALYST,CITIZEN DATA SCIENTIST M
LSK
ILLS#
OF
USER
SMO
DE
1M
OD
E2
PAL, APL,Graph, Text,
Streaming, PAI
Open Source & 3rd PartyMachine Learning
(R, Python, TensorFlow,Spark ML, Hadoop)
Machine LearningCloud Services
SAP PredictiveAnalytics
3rd Party AnalyticalTools
EmbeddedIntelligence(S/4, SAC,
Hybris)
New IntelligentApplications
INTERNAL
December, 2017
Automating predictive quality pipelines based on image featuresSAP Data Hub use case
12PUBLIC© 2017 SAP SE or an SAP affiliate company. All rights reserved. ǀPublic
Efficiently Identifying Faults Earlier
Business Outcome› Reduce waste and rework due to higher accuracy of quality checks
› Early issue detection due to solution performance
Pressure
Temperature
IR image features
ManufacturingMolding press
Sensors
IR cameras
Raw material
ERP Data
13PUBLIC© 2017 SAP SE or an SAP affiliate company. All rights reserved. ǀPublic
SAP Data Hub
SAP HANA Platform
Image ProcessingPipeline
Sensor Pipeline
Predictive
HTML5 User Interface
ERP Data
Feature Extraction
Streaming Data
XSAApplication Server
Business & logistics data
Data Modeling
Orchestration & Pipelines - High Level Architecture
14PUBLIC© 2017 SAP SE or an SAP affiliate company. All rights reserved. ǀPublic
Backend : Data Hub task workflow
2. Extract Features
3. Classify
1. Stream Data
• The workflow simulates the arrival of an image and new sensor data stream
• Process continues with two operators streaming the data to HANA and performing the Python image processingto extract the features
• Finally, a stored procedure in HANA using PAL is used to apply a classification model to evaluate the quality
15PUBLIC© 2017 SAP SE or an SAP affiliate company. All rights reserved. ǀPublic
Backend : Data Hub Pipelines
1. Stream Data
16PUBLIC© 2017 SAP SE or an SAP affiliate company. All rights reserved. ǀPublic
Backend : Data Hub Pipelines2. Extract Features
17PUBLIC© 2017 SAP SE or an SAP affiliate company. All rights reserved. ǀPublic
Backend : Data Hub Pipelines
3. Classify
18PUBLIC© 2017 SAP SE or an SAP affiliate company. All rights reserved. ǀPublic
Frontend : Monitoring UI
Track the products onthe production linewith the quality checkresults
IR Image of theproduction line foroptical validation
Main contributingvariables with theirvalues can be seenhere. If they are overthe limit, it is indicatedby red font
Operator can notifyengineers if someirregularities occur
No part of this publication may be reproduced or transmitted in any form or for any purpose without the express permission of SAP SE or an SAP affiliate company.
The information contained herein may be changed without prior notice. Some software products marketed by SAP SE and its distributors contain proprietary software componentsof other software vendors. National product specifications may vary.
These materials are provided by SAP SE or an SAP affiliate company for informational purposes only, without representation or warranty of any kind, and SAP or its affiliatedcompanies shall not be liable for errors or omissions with respect to the materials. The only warranties for SAP or SAP affiliate company products and services are those that areset forth in the express warranty statements accompanying such products and services, if any. Nothing herein should be construed as constituting an additional warranty.
In particular, SAP SE or its affiliated companies have no obligation to pursue any course of business outlined in this document or any related presentation, or to develop or releaseany functionality mentioned therein. This document, or any related presentation, and SAP SE’s or its affiliated companies’ strategy and possible future developments, products,and/or platforms, directions, and functionality are all subject to change and may be changed by SAP SE or its affiliated companies at any time for any reason without notice. Theinformation in this document is not a commitment, promise, or legal obligation to deliver any material, code, or functionality. All forward-looking statements are subject to various risksand uncertainties that could cause actual results to differ materially from expectations. Readers are cautioned not to place undue reliance on these forward-looking statements, andthey should not be relied upon in making purchasing decisions.
SAP and other SAP products and services mentioned herein as well as their respective logos are trademarks or registered trademarks of SAP SE (or an SAP affiliate company)in Germany and other countries. All other product and service names mentioned are the trademarks of their respective companies.See www.sap.com/corporate-en/legal/copyright/index.epx for additional trademark information and notices.
© 2018 SAP SE or an SAP affiliate company. All rights reserved.