SlideShare uma empresa Scribd logo
1 de 31
Data-driven AI for
Self-adaptive
Information Systems
Andreas Metzger
BIOC & FAiSE
Roma, MMXIX
Agenda
2
Motivation and
Background
Case 1:
Deep Learning for
Proactive Process
Adaptation
Case 2:
Policy-based Reinforcement
Learning for
Self-adaptive
Cloud Services
Discussion
and Outlook
BIOC/FAiSE, Roma, MMXIX
Need for Self-Adaptation
3
Engineering
Requirements
Engineering
Design
Coding
Deployment
Operations
Manual information system engineering too slow…
BIOC/FAiSE, Roma, MMXIX
Need for Self-Adaptation
4
Requirements
Engineering
Design
Coding
Deployment
Operations
Monitoring
Adaptation
Self-Adaption Engineering
Automated perception, reasoning, actuation, …
BIOC/FAiSE, Roma, MMXIX
Data-driven AI as Enabling Technique
BIOC/FAiSE, Roma, MMXIX 5
Image: https://www.datarevenue.com/
a.k.a
Data-driven AI
Data-driven as Enabling Technique
Main Drivers
BIOC/FAiSE, Roma, MMXIX 6
2020
2011
44 Zettabyte
1 Zettabyte
Big Data
Hardware
e.g., GPU / TPU
Algorithms
e.g., Deep Learning
Case 2Case 1
Data-driven as Enabling Technique
BIOC/FAiSE, Roma, MMXIX 7
Unsupervised
Learning
Supervised
Learning
Reinforcement
Learning
No Feedback Labelled Training Data Reward for an Action
Agenda
8
Motivation and
Background
Case 1:
Deep Learning for
Proactive Process
Adaptation
Case 2:
Policy-based Reinforcement
Learning for
Self-adaptive
Cloud Services
Discussion
and Outlook
BIOC/FAiSE, Roma, MMXIX
Process
completiontCheckpoint j
Process
start
Proactive Process Adaptation “in a Nutshell”
BIOC/FAiSE, Roma, MMXIX 9
Monitor
Predict
Proactive
adaptation
planned /
acceptable situations
= Violation
= Non-
Violation

E.g., Delayed
freight delivery
E.g., Schedule air
instead of land
transport
E.g., Freight
delivery within 2
days
Process
Performance
Considerations for Proactive Adaptation
Accuracy
• False violations  Unnecessary adaptations
• False non-violations  Missed adaptations
Earliness
• Late prediction  no time for adaptation
But: Trade-off between accuracy and earliness
BIOC/FAiSE, Roma, MMXIX 10
[Metzger & Neubauer, 2018] [Teinemaa et al., 2019]
BPIC 2017BPIC 2012Cargo 2000
Process
completion
Accuracy[MCC]
Accuracy[AUC]
Prediction Reliability for Proactive Adaptation
BIOC/FAiSE, Roma, MMXIX 11
Reliability Estimate
• Probability that individual prediction is correct
 Distinguish between more or less reliable predictions
Dynamic Approach
• Use earliest prediction with sufficiently high reliability
Aggregate Accuracy
75%
75%
75%
75%
Prediction #
1
2
3
…
Reliability Estimate
60%
90%
70%
…
Prediction Reliability for Proactive Adaptation
Real-World Example
12BIOC/FAiSE, Roma, MMXIX
Alarm about Delay
Reliability Estimate
Terminal Productivity Cockpit
Data sources
• Data streams from terminal equipment
(1.3 mio states / month)
• Integrated data of container moves
(10,000 moves / month)
Container terminal
Prediction Reliability for Proactive Adaptation
Realization of Dynamic Approach
13BIOC/FAiSE, Roma, MMXIX
Process
monitoring
data at
Checkpoint j
RNN Model 1
RNN Model m
…
Ensemble
Prediction
[Tj = “non-violation”]
Proactive
Process
Adaptation
No Proactive
Process
Adaptation
Prediction Tj
Reliability
estimate j
[Tj =
“violation”]
[j  threshold]
[j < threshold]
Deep Learning
RNNs as Base Learners
RNN = Recurrent Neural Network
Benefits
• High accuracy [Tax et. al. 2017; Evermann et al. 2017, Metzger & Nebauer, 2018]
• Arbitrary length process instances (without sequence encoding)
• Predictions at any checkpoint
Scalability
• Long training time
 Parallelization (Bagging)
 Hardware speedups
14BIOC/FAiSE, Roma, MMXIX
Hardware type Training
time
CPU 25 min
GPU (Nvidia CuDNN) 8 min
Google TPU (Tensorflow) 2 min
Results
15BIOC/FAiSE, Roma, MMXIX
“Cheap”
adaptation
“Expensive”
adaptation
Static checkpoint
Dynamic approach
No proactive
adaptation
Results
16BIOC/FAiSE, Roma, MMXIX
“Cheap”
adaptation
“Expensive”
adaptation
Static checkpoint
Dynamic approach
No proactive
adaptation
Details in A. Metzger, A. Neubauer, P. Bohn, and K. Pohl, “Proactive process adaptation
using deep learning ensembles,” in 31st Int’l Conference on Advanced Information
Systems Engineering (CAiSE 2019), Rome, Italy, June 3-7, 2019, ser. LNCS, P. Giorgini and B.
Weber, Eds., vol. 11483. Springer, 2019.
Open Access: https://link.springer.com/chapter/10.1007%2F978-3-030-21290-2_34
Agenda
17
Motivation and
Background
Case 1:
Deep Learning for
Proactive Process
Adaptation
Case 2:
Policy-based Reinforcement
Learning for
Self-adaptive
Cloud Services
Discussion
and Outlook
BIOC/FAiSE, Roma, MMXIX
Design Time Uncertainty
• What are potential
environment situations?
• How does adaptation
affect quality requirements?
Learning for Self-Adaptation „in a Nutshell“
18
Self-Adaptation Logic
Knowledge
Analyze Plan
Monitor Execute
System Logic
Environment
Solution
Self-Adaptation Logic
Knowledge
Analyze Plan
Monitor Execute
System Logic
Environment
Online Learning
Observe Induce
Self-adaptive System Reference Model
BIOC/FAiSE, Roma, MMXIX
Learning for Self-Adaptation „in a Nutshell“
Most widely used: Reinforcement Learning
19BIOC/FAiSE, Roma, MMXIX
Agent
(with action-
selection policy )
Environment
Action
atrt+1
st+1
State
st
Reward
rt
https://www.youtube.com/watch?v=gn4nRCC9TwQ
Considerations for Self-adaptive Cloud Services
Large, continuous environment space
• Continuous state variables (e.g., workload)
• Many state variables
State of the Art: Value-based RL
• Value function =
expected cumulative reward per state
• Requires manual quantization of states
 error prone
• Requires balancing exploration vs. exploitation
20BIOC/FAiSE, Roma, MMXIX
[Marquezan et al., 2014]
Coping with Large Environment Space
Policy-based reinforcement learning
• Approximates and
generalizes over states
• Probabilistic action selection
Learning via policy gradient methods
• Policy update according to gradient of objective function
21
Environment
Action
at
rt+1
st+1
State
st
Reward
rt
BIOC/FAiSE, Roma, MMXIX
Weights of neural
network
Coping with Large Environment Space
Realization of policy-based RL for self-adaptive cloud services
22
Self-Adaptation Logic
Knowledge
Analyze Plan
Monitor Execute
BIOC/FAiSE, Roma, MMXIX
Self-Adaptation Logic
Knowledge
Analyze Plan
Monitor Execute
Policy-based Reinforcement Learning
at
rt+1
t+1
Monitor Execute
Action-
Selection
Self-Adaptation Logic
Policy 
(Knowledge)
st
Policy Update
st+1
Coping with Large Environment Space
Benchmark Example
State
Reward
23BIOC/FAiSE, Roma, MMXIX
[Klein et al., 2014]Brownout-RUBiS
Nbr Requests
CPU Load
Actual Recomm. Rate
Response time (ms)
Auction Web Application
• User requests served with
recommendations for
similar products
Action
Set dimmer value
= per-request probability that
recommendation engine is activated
Results
24
Stable Workload
Off/on workload
BIOC/FAiSE, Roma, MMXIX
Growing workload
Cyclic workload
red = avg. reward
blue = workload
Results
25
Real-world workload
BIOC/FAiSE, Roma, MMXIX
Agenda
26
Motivation and
Background
Case 1:
Deep Learning for
Proactive Process
Adaptation
Case 2:
Policy-based Reinforcement
Learning for
Self-adaptive
Cloud Services
Discussion
and Outlook
BIOC/FAiSE, Roma, MMXIX
Lessons Learned
Deep learning requires little hyper-parametrization
• If enough good quality data is available / can be collected
Use deep learning to increase productivity
of information system engineering
Prediction reliability offers additional decision support
• Operators put more trust in predictions
• Cost savings
Augment predictions with reliability estimates, confidence intervals,
error ranges, etc.
Data quality is a key concern
• “Garbage in – garbage out”
• Missing data, data accuracy, timeliness, timestamps (clocks), …
• Very resource and time-intensive
Plan sufficient time and effort for data quality, data integration and
refinement of data collection
27BIOC/FAiSE, Roma, MMXIX
Limitations
Risks
• Adaptations may “harm” environment
(embedded systems / CPS)
• Adversaries may manipulate training data
Skills
28BIOC/FAiSE, Roma, MMXIX
(Source: IDC)
Year Gap (total EU)
2020 (baseline) 530,000
2020 (high-growth) 3,500,000
Outlook
29BIOC/FAiSE, Roma, MMXIX
Decision Making
• Deep learning for high
accuracy descriptive and
predictive analytics
• Reinforcement learning
for solving complex
planning and decision
problems
Enactment
• Actuation driven by AI
decisions
• Safety and
trustworthiness as key
requirements for
adoption
Thanks!
BIOC/FAiSE, Roma, MMXIX 30
Research leading to these results has received
funding from the EU’s Horizon 2020 research and
innovation programme under grant agreements no.
731932 – http://www.transformingtransport.eu
732630 – http://www.big-data-value.eu
780351 – https://enact-project.eu
References
[Evermann et al. 2017] Evermann, J., Rehse, J., Fettke, P.: Predicting process behaviour using
deep learning. Decision Support Systems 100 (2017)
[Klein et al., 2014] Klein, C., Maggio, M., Arz´ en, K.E., Hernández-Rodriguez, F.: Brownout:
Building more robust cloud applications. In: 36th Intl Conf. on Software Engineering
(ICSE 2014). pp. 700–711. ACM (2014)
[Metzger & Neubauer, 2018] A. Metzger and A. Neubauer, “Considering non-sequential
control flows for process prediction with recurrent neural networks,” in 44th Euromicro
Conference on Software Engineering and Advanced Applications (SEAA 2018), Prague,
Czech Republic, August 29-31, 2018, T. Bures and L. Angelis, Eds. IEEE Computer
Society, 2018, pp. 268–272.
[Metzger et al., 2019] A. Metzger, A. Neubauer, P. Bohn, and K. Pohl, “Proactive process
adaptation using deep learning ensembles,” in 31st Int’l Conference on Advanced
Information Systems Engineering (CAiSE 2019), Rome, Italy, June 3-7, 2019, ser. LNCS, P.
Giorgini and B. Weber, Eds., vol. 11483. Springer, 2019. Open Access:
https://link.springer.com/chapter/10.1007%2F978-3-030-21290-2_34
[Tax et. al. 2017] Tax, N., Verenich, I., Rosa, M.L., Dumas, M.: Predictive business process
monitoring with LSTM neural networks. In: Dubois, E., Pohl, K. (eds.) CAiSE 2017, Essen,
Germany, June 12-16, 2017. LNCS, vol. 10253. Springer (2017)
[Teinemaa et al. 2019] Teinemaa, I., Dumas, M., Rosa, M.L., Maggi, F.M.: Outcome-oriented
predictive process monitoring: Review and benchmark. ACM Transactions on
Knowledge Discovery from Data (TKDD) 13 (2019)
BIOC/FAiSE, Roma, MMXIX 31

Mais conteúdo relacionado

Mais procurados

Internet of Things (IoT) and Big Data
Internet of Things (IoT) and Big DataInternet of Things (IoT) and Big Data
Internet of Things (IoT) and Big DataGuido Schmutz
 
AI and ML Series - Introduction to Generative AI and LLMs - Session 1
AI and ML Series - Introduction to Generative AI and LLMs - Session 1AI and ML Series - Introduction to Generative AI and LLMs - Session 1
AI and ML Series - Introduction to Generative AI and LLMs - Session 1DianaGray10
 
The Future of AI is Generative not Discriminative 5/26/2021
The Future of AI is Generative not Discriminative 5/26/2021The Future of AI is Generative not Discriminative 5/26/2021
The Future of AI is Generative not Discriminative 5/26/2021Steve Omohundro
 
Smart Data Slides: Machine Learning - Case Studies
Smart Data Slides: Machine Learning - Case StudiesSmart Data Slides: Machine Learning - Case Studies
Smart Data Slides: Machine Learning - Case StudiesDATAVERSITY
 
Data Analytics and Artificial Intelligence in the era of Digital Transformation
Data Analytics and Artificial Intelligence in the era of Digital TransformationData Analytics and Artificial Intelligence in the era of Digital Transformation
Data Analytics and Artificial Intelligence in the era of Digital TransformationJan Wiegelmann
 
Big data analysis and Internet of Things(IoT)
Big data analysis and Internet of Things(IoT)Big data analysis and Internet of Things(IoT)
Big data analysis and Internet of Things(IoT)Monica Kambala
 
1. Data Analytics-introduction
1. Data Analytics-introduction1. Data Analytics-introduction
1. Data Analytics-introductionkrishna singh
 
Understanding GenAI/LLM and What is Google Offering - Felix Goh
Understanding GenAI/LLM and What is Google Offering - Felix GohUnderstanding GenAI/LLM and What is Google Offering - Felix Goh
Understanding GenAI/LLM and What is Google Offering - Felix GohNUS-ISS
 
Automatic machine learning (AutoML) 101
Automatic machine learning (AutoML) 101Automatic machine learning (AutoML) 101
Automatic machine learning (AutoML) 101QuantUniversity
 
Data Mining: Application and trends in data mining
Data Mining: Application and trends in data miningData Mining: Application and trends in data mining
Data Mining: Application and trends in data miningDataminingTools Inc
 
Artificial Intelligence Machine Learning Deep Learning Ppt Powerpoint Present...
Artificial Intelligence Machine Learning Deep Learning Ppt Powerpoint Present...Artificial Intelligence Machine Learning Deep Learning Ppt Powerpoint Present...
Artificial Intelligence Machine Learning Deep Learning Ppt Powerpoint Present...SlideTeam
 
Deep learning health care
Deep learning health care  Deep learning health care
Deep learning health care Meenakshi Sood
 

Mais procurados (20)

Internet of Things (IoT) and Big Data
Internet of Things (IoT) and Big DataInternet of Things (IoT) and Big Data
Internet of Things (IoT) and Big Data
 
AI and ML Series - Introduction to Generative AI and LLMs - Session 1
AI and ML Series - Introduction to Generative AI and LLMs - Session 1AI and ML Series - Introduction to Generative AI and LLMs - Session 1
AI and ML Series - Introduction to Generative AI and LLMs - Session 1
 
The Future of AI is Generative not Discriminative 5/26/2021
The Future of AI is Generative not Discriminative 5/26/2021The Future of AI is Generative not Discriminative 5/26/2021
The Future of AI is Generative not Discriminative 5/26/2021
 
Data analytics
Data analyticsData analytics
Data analytics
 
Smart Data Slides: Machine Learning - Case Studies
Smart Data Slides: Machine Learning - Case StudiesSmart Data Slides: Machine Learning - Case Studies
Smart Data Slides: Machine Learning - Case Studies
 
Data Analytics and Artificial Intelligence in the era of Digital Transformation
Data Analytics and Artificial Intelligence in the era of Digital TransformationData Analytics and Artificial Intelligence in the era of Digital Transformation
Data Analytics and Artificial Intelligence in the era of Digital Transformation
 
Big data analysis and Internet of Things(IoT)
Big data analysis and Internet of Things(IoT)Big data analysis and Internet of Things(IoT)
Big data analysis and Internet of Things(IoT)
 
Presentation on Big Data
Presentation on Big DataPresentation on Big Data
Presentation on Big Data
 
Deep learning and Healthcare
Deep learning and HealthcareDeep learning and Healthcare
Deep learning and Healthcare
 
1. Data Analytics-introduction
1. Data Analytics-introduction1. Data Analytics-introduction
1. Data Analytics-introduction
 
Generative AI
Generative AIGenerative AI
Generative AI
 
Understanding GenAI/LLM and What is Google Offering - Felix Goh
Understanding GenAI/LLM and What is Google Offering - Felix GohUnderstanding GenAI/LLM and What is Google Offering - Felix Goh
Understanding GenAI/LLM and What is Google Offering - Felix Goh
 
Journey of Generative AI
Journey of Generative AIJourney of Generative AI
Journey of Generative AI
 
Automatic machine learning (AutoML) 101
Automatic machine learning (AutoML) 101Automatic machine learning (AutoML) 101
Automatic machine learning (AutoML) 101
 
Data Mining: Application and trends in data mining
Data Mining: Application and trends in data miningData Mining: Application and trends in data mining
Data Mining: Application and trends in data mining
 
Artificial Intelligence Machine Learning Deep Learning Ppt Powerpoint Present...
Artificial Intelligence Machine Learning Deep Learning Ppt Powerpoint Present...Artificial Intelligence Machine Learning Deep Learning Ppt Powerpoint Present...
Artificial Intelligence Machine Learning Deep Learning Ppt Powerpoint Present...
 
Deep learning health care
Deep learning health care  Deep learning health care
Deep learning health care
 
Implementing Ethics in AI
Implementing Ethics in AIImplementing Ethics in AI
Implementing Ethics in AI
 
Iot and Healthcare ppt
Iot and Healthcare pptIot and Healthcare ppt
Iot and Healthcare ppt
 
Predictive Analytics - An Introduction
Predictive Analytics - An IntroductionPredictive Analytics - An Introduction
Predictive Analytics - An Introduction
 

Semelhante a Data-driven AI for Self-adaptive Information Systems

Data-driven AI for Self-Adaptive Software Systems
Data-driven AI for Self-Adaptive Software SystemsData-driven AI for Self-Adaptive Software Systems
Data-driven AI for Self-Adaptive Software SystemsAndreas Metzger
 
Analysis of IT Monitoring Using Open Source Software Techniques: A Review
Analysis of IT Monitoring Using Open Source Software Techniques: A ReviewAnalysis of IT Monitoring Using Open Source Software Techniques: A Review
Analysis of IT Monitoring Using Open Source Software Techniques: A ReviewIJERD Editor
 
Cyber physical systems and robotics
Cyber physical systems and roboticsCyber physical systems and robotics
Cyber physical systems and roboticstrinhanhtuan247
 
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...ijdpsjournal
 
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...ijdpsjournal
 
Ericsson Support Services
Ericsson Support Services Ericsson Support Services
Ericsson Support Services Ericsson
 
Hardware Design Practices For Modern Hardware
Hardware Design Practices For Modern HardwareHardware Design Practices For Modern Hardware
Hardware Design Practices For Modern HardwareWinstina Kennedy
 
Overview of GovCloud Today
Overview of GovCloud TodayOverview of GovCloud Today
Overview of GovCloud TodayGovCloud Network
 
Next Century Project Overview
Next Century Project OverviewNext Century Project Overview
Next Century Project Overviewjennhunter
 
Big Data, Physics, and the Industrial Internet: How Modeling & Analytics are ...
Big Data, Physics, and the Industrial Internet: How Modeling & Analytics are ...Big Data, Physics, and the Industrial Internet: How Modeling & Analytics are ...
Big Data, Physics, and the Industrial Internet: How Modeling & Analytics are ...mattdenesuk
 
Lancy-Curriculum Vitae
Lancy-Curriculum VitaeLancy-Curriculum Vitae
Lancy-Curriculum VitaeLancy Menezes
 
Svm Classifier Algorithm for Data Stream Mining Using Hive and R
Svm Classifier Algorithm for Data Stream Mining Using Hive and RSvm Classifier Algorithm for Data Stream Mining Using Hive and R
Svm Classifier Algorithm for Data Stream Mining Using Hive and RIRJET Journal
 
Data Ingestion At Scale (CNECCS 2017)
Data Ingestion At Scale (CNECCS 2017)Data Ingestion At Scale (CNECCS 2017)
Data Ingestion At Scale (CNECCS 2017)Jeffrey Sica
 
Face detection on_embedded_systems
Face detection on_embedded_systemsFace detection on_embedded_systems
Face detection on_embedded_systemsr_sadoun
 
A Hierarchical Distributed Processing Framework for Huge Image by using Big Data
A Hierarchical Distributed Processing Framework for Huge Image by using Big DataA Hierarchical Distributed Processing Framework for Huge Image by using Big Data
A Hierarchical Distributed Processing Framework for Huge Image by using Big DataIRJET Journal
 
PERICLES workshop (London 15 October 2015) - Digital Ecosystem Model
PERICLES workshop (London 15 October 2015) - Digital Ecosystem ModelPERICLES workshop (London 15 October 2015) - Digital Ecosystem Model
PERICLES workshop (London 15 October 2015) - Digital Ecosystem ModelPERICLES_FP7
 
Predictive Maintenance by analysing acoustic data in an industrial environment
Predictive Maintenance by analysing acoustic data in an industrial environmentPredictive Maintenance by analysing acoustic data in an industrial environment
Predictive Maintenance by analysing acoustic data in an industrial environmentCapgemini
 
MN691 Assignment 3 - Final Report 2
MN691 Assignment 3 - Final Report 2MN691 Assignment 3 - Final Report 2
MN691 Assignment 3 - Final Report 2Abi Reddy
 
IRJET - A Research on Eloquent Salvation and Productive Outsourcing of Massiv...
IRJET - A Research on Eloquent Salvation and Productive Outsourcing of Massiv...IRJET - A Research on Eloquent Salvation and Productive Outsourcing of Massiv...
IRJET - A Research on Eloquent Salvation and Productive Outsourcing of Massiv...IRJET Journal
 

Semelhante a Data-driven AI for Self-adaptive Information Systems (20)

Data-driven AI for Self-Adaptive Software Systems
Data-driven AI for Self-Adaptive Software SystemsData-driven AI for Self-Adaptive Software Systems
Data-driven AI for Self-Adaptive Software Systems
 
Analysis of IT Monitoring Using Open Source Software Techniques: A Review
Analysis of IT Monitoring Using Open Source Software Techniques: A ReviewAnalysis of IT Monitoring Using Open Source Software Techniques: A Review
Analysis of IT Monitoring Using Open Source Software Techniques: A Review
 
Cyber physical systems and robotics
Cyber physical systems and roboticsCyber physical systems and robotics
Cyber physical systems and robotics
 
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...
 
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...
SEAMLESS AUTOMATION AND INTEGRATION OF MACHINE LEARNING CAPABILITIES FOR BIG ...
 
Ericsson Support Services
Ericsson Support Services Ericsson Support Services
Ericsson Support Services
 
Hardware Design Practices For Modern Hardware
Hardware Design Practices For Modern HardwareHardware Design Practices For Modern Hardware
Hardware Design Practices For Modern Hardware
 
Overview of GovCloud Today
Overview of GovCloud TodayOverview of GovCloud Today
Overview of GovCloud Today
 
Next Century Project Overview
Next Century Project OverviewNext Century Project Overview
Next Century Project Overview
 
Big Data, Physics, and the Industrial Internet: How Modeling & Analytics are ...
Big Data, Physics, and the Industrial Internet: How Modeling & Analytics are ...Big Data, Physics, and the Industrial Internet: How Modeling & Analytics are ...
Big Data, Physics, and the Industrial Internet: How Modeling & Analytics are ...
 
rerngvit_phd_seminar
rerngvit_phd_seminarrerngvit_phd_seminar
rerngvit_phd_seminar
 
Lancy-Curriculum Vitae
Lancy-Curriculum VitaeLancy-Curriculum Vitae
Lancy-Curriculum Vitae
 
Svm Classifier Algorithm for Data Stream Mining Using Hive and R
Svm Classifier Algorithm for Data Stream Mining Using Hive and RSvm Classifier Algorithm for Data Stream Mining Using Hive and R
Svm Classifier Algorithm for Data Stream Mining Using Hive and R
 
Data Ingestion At Scale (CNECCS 2017)
Data Ingestion At Scale (CNECCS 2017)Data Ingestion At Scale (CNECCS 2017)
Data Ingestion At Scale (CNECCS 2017)
 
Face detection on_embedded_systems
Face detection on_embedded_systemsFace detection on_embedded_systems
Face detection on_embedded_systems
 
A Hierarchical Distributed Processing Framework for Huge Image by using Big Data
A Hierarchical Distributed Processing Framework for Huge Image by using Big DataA Hierarchical Distributed Processing Framework for Huge Image by using Big Data
A Hierarchical Distributed Processing Framework for Huge Image by using Big Data
 
PERICLES workshop (London 15 October 2015) - Digital Ecosystem Model
PERICLES workshop (London 15 October 2015) - Digital Ecosystem ModelPERICLES workshop (London 15 October 2015) - Digital Ecosystem Model
PERICLES workshop (London 15 October 2015) - Digital Ecosystem Model
 
Predictive Maintenance by analysing acoustic data in an industrial environment
Predictive Maintenance by analysing acoustic data in an industrial environmentPredictive Maintenance by analysing acoustic data in an industrial environment
Predictive Maintenance by analysing acoustic data in an industrial environment
 
MN691 Assignment 3 - Final Report 2
MN691 Assignment 3 - Final Report 2MN691 Assignment 3 - Final Report 2
MN691 Assignment 3 - Final Report 2
 
IRJET - A Research on Eloquent Salvation and Productive Outsourcing of Massiv...
IRJET - A Research on Eloquent Salvation and Productive Outsourcing of Massiv...IRJET - A Research on Eloquent Salvation and Productive Outsourcing of Massiv...
IRJET - A Research on Eloquent Salvation and Productive Outsourcing of Massiv...
 

Mais de Andreas Metzger

Explainable Online Reinforcement Learning for Adaptive Systems
Explainable Online Reinforcement Learning for Adaptive SystemsExplainable Online Reinforcement Learning for Adaptive Systems
Explainable Online Reinforcement Learning for Adaptive SystemsAndreas Metzger
 
Data Quality Issues in Online Reinforcement Learning for Self-Adaptive System...
Data Quality Issues in Online Reinforcement Learning for Self-Adaptive System...Data Quality Issues in Online Reinforcement Learning for Self-Adaptive System...
Data Quality Issues in Online Reinforcement Learning for Self-Adaptive System...Andreas Metzger
 
Explaining Online Reinforcement Learning Decisions of Self-Adaptive Systems
Explaining Online Reinforcement Learning Decisions of Self-Adaptive SystemsExplaining Online Reinforcement Learning Decisions of Self-Adaptive Systems
Explaining Online Reinforcement Learning Decisions of Self-Adaptive SystemsAndreas Metzger
 
Antrittsvorlesung - APL.pptx
Antrittsvorlesung - APL.pptxAntrittsvorlesung - APL.pptx
Antrittsvorlesung - APL.pptxAndreas Metzger
 
Feature Model-Guided Online Reinforcement Learning for Self-Adaptive Services
Feature Model-Guided Online Reinforcement Learning for Self-Adaptive ServicesFeature Model-Guided Online Reinforcement Learning for Self-Adaptive Services
Feature Model-Guided Online Reinforcement Learning for Self-Adaptive ServicesAndreas Metzger
 
Triggering Proactive Business Process Adaptations via Online Reinforcement Le...
Triggering Proactive Business Process Adaptations via Online Reinforcement Le...Triggering Proactive Business Process Adaptations via Online Reinforcement Le...
Triggering Proactive Business Process Adaptations via Online Reinforcement Le...Andreas Metzger
 
Data-driven Deep Learning for Proactive Terminal Process Management
Data-driven Deep Learning for Proactive Terminal Process ManagementData-driven Deep Learning for Proactive Terminal Process Management
Data-driven Deep Learning for Proactive Terminal Process Management Andreas Metzger
 
Big Data Technology Insights
Big Data Technology InsightsBig Data Technology Insights
Big Data Technology InsightsAndreas Metzger
 
Proactive Process Adaptation using Deep Learning Ensembles
Proactive Process Adaptation using Deep Learning Ensembles Proactive Process Adaptation using Deep Learning Ensembles
Proactive Process Adaptation using Deep Learning Ensembles Andreas Metzger
 
Towards an End-to-End Architecture for Run-time Data Protection in the Cloud
Towards an End-to-End Architecture for Run-time Data Protection in the Cloud Towards an End-to-End Architecture for Run-time Data Protection in the Cloud
Towards an End-to-End Architecture for Run-time Data Protection in the Cloud Andreas Metzger
 
Considering Non-sequential Control Flows for Process Prediction with Recurren...
Considering Non-sequential Control Flows for Process Prediction with Recurren...Considering Non-sequential Control Flows for Process Prediction with Recurren...
Considering Non-sequential Control Flows for Process Prediction with Recurren...Andreas Metzger
 
Big Data Value in Mobility and Logistics
Big Data Value in Mobility and Logistics Big Data Value in Mobility and Logistics
Big Data Value in Mobility and Logistics Andreas Metzger
 
Predictive Business Process Monitoring considering Reliability and Risk
Predictive Business Process Monitoring considering Reliability and RiskPredictive Business Process Monitoring considering Reliability and Risk
Predictive Business Process Monitoring considering Reliability and RiskAndreas Metzger
 
Risk-based Proactive Process Adaptation
Risk-based Proactive Process AdaptationRisk-based Proactive Process Adaptation
Risk-based Proactive Process AdaptationAndreas Metzger
 
Predictive Process Monitoring Considering Reliability Estimates
Predictive Process Monitoring Considering Reliability EstimatesPredictive Process Monitoring Considering Reliability Estimates
Predictive Process Monitoring Considering Reliability EstimatesAndreas Metzger
 

Mais de Andreas Metzger (15)

Explainable Online Reinforcement Learning for Adaptive Systems
Explainable Online Reinforcement Learning for Adaptive SystemsExplainable Online Reinforcement Learning for Adaptive Systems
Explainable Online Reinforcement Learning for Adaptive Systems
 
Data Quality Issues in Online Reinforcement Learning for Self-Adaptive System...
Data Quality Issues in Online Reinforcement Learning for Self-Adaptive System...Data Quality Issues in Online Reinforcement Learning for Self-Adaptive System...
Data Quality Issues in Online Reinforcement Learning for Self-Adaptive System...
 
Explaining Online Reinforcement Learning Decisions of Self-Adaptive Systems
Explaining Online Reinforcement Learning Decisions of Self-Adaptive SystemsExplaining Online Reinforcement Learning Decisions of Self-Adaptive Systems
Explaining Online Reinforcement Learning Decisions of Self-Adaptive Systems
 
Antrittsvorlesung - APL.pptx
Antrittsvorlesung - APL.pptxAntrittsvorlesung - APL.pptx
Antrittsvorlesung - APL.pptx
 
Feature Model-Guided Online Reinforcement Learning for Self-Adaptive Services
Feature Model-Guided Online Reinforcement Learning for Self-Adaptive ServicesFeature Model-Guided Online Reinforcement Learning for Self-Adaptive Services
Feature Model-Guided Online Reinforcement Learning for Self-Adaptive Services
 
Triggering Proactive Business Process Adaptations via Online Reinforcement Le...
Triggering Proactive Business Process Adaptations via Online Reinforcement Le...Triggering Proactive Business Process Adaptations via Online Reinforcement Le...
Triggering Proactive Business Process Adaptations via Online Reinforcement Le...
 
Data-driven Deep Learning for Proactive Terminal Process Management
Data-driven Deep Learning for Proactive Terminal Process ManagementData-driven Deep Learning for Proactive Terminal Process Management
Data-driven Deep Learning for Proactive Terminal Process Management
 
Big Data Technology Insights
Big Data Technology InsightsBig Data Technology Insights
Big Data Technology Insights
 
Proactive Process Adaptation using Deep Learning Ensembles
Proactive Process Adaptation using Deep Learning Ensembles Proactive Process Adaptation using Deep Learning Ensembles
Proactive Process Adaptation using Deep Learning Ensembles
 
Towards an End-to-End Architecture for Run-time Data Protection in the Cloud
Towards an End-to-End Architecture for Run-time Data Protection in the Cloud Towards an End-to-End Architecture for Run-time Data Protection in the Cloud
Towards an End-to-End Architecture for Run-time Data Protection in the Cloud
 
Considering Non-sequential Control Flows for Process Prediction with Recurren...
Considering Non-sequential Control Flows for Process Prediction with Recurren...Considering Non-sequential Control Flows for Process Prediction with Recurren...
Considering Non-sequential Control Flows for Process Prediction with Recurren...
 
Big Data Value in Mobility and Logistics
Big Data Value in Mobility and Logistics Big Data Value in Mobility and Logistics
Big Data Value in Mobility and Logistics
 
Predictive Business Process Monitoring considering Reliability and Risk
Predictive Business Process Monitoring considering Reliability and RiskPredictive Business Process Monitoring considering Reliability and Risk
Predictive Business Process Monitoring considering Reliability and Risk
 
Risk-based Proactive Process Adaptation
Risk-based Proactive Process AdaptationRisk-based Proactive Process Adaptation
Risk-based Proactive Process Adaptation
 
Predictive Process Monitoring Considering Reliability Estimates
Predictive Process Monitoring Considering Reliability EstimatesPredictive Process Monitoring Considering Reliability Estimates
Predictive Process Monitoring Considering Reliability Estimates
 

Último

Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...amitlee9823
 
Edukaciniai dropshipping via API with DroFx
Edukaciniai dropshipping via API with DroFxEdukaciniai dropshipping via API with DroFx
Edukaciniai dropshipping via API with DroFxolyaivanovalion
 
April 2024 - Crypto Market Report's Analysis
April 2024 - Crypto Market Report's AnalysisApril 2024 - Crypto Market Report's Analysis
April 2024 - Crypto Market Report's Analysismanisha194592
 
Jual Obat Aborsi Surabaya ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...
Jual Obat Aborsi Surabaya ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...Jual Obat Aborsi Surabaya ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...
Jual Obat Aborsi Surabaya ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...ZurliaSoop
 
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...amitlee9823
 
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...amitlee9823
 
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...SUHANI PANDEY
 
Capstone Project on IBM Data Analytics Program
Capstone Project on IBM Data Analytics ProgramCapstone Project on IBM Data Analytics Program
Capstone Project on IBM Data Analytics ProgramMoniSankarHazra
 
VidaXL dropshipping via API with DroFx.pptx
VidaXL dropshipping via API with DroFx.pptxVidaXL dropshipping via API with DroFx.pptx
VidaXL dropshipping via API with DroFx.pptxolyaivanovalion
 
Midocean dropshipping via API with DroFx
Midocean dropshipping via API with DroFxMidocean dropshipping via API with DroFx
Midocean dropshipping via API with DroFxolyaivanovalion
 
BabyOno dropshipping via API with DroFx.pptx
BabyOno dropshipping via API with DroFx.pptxBabyOno dropshipping via API with DroFx.pptx
BabyOno dropshipping via API with DroFx.pptxolyaivanovalion
 
BDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort Service
BDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort ServiceBDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort Service
BDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort ServiceDelhi Call girls
 
Halmar dropshipping via API with DroFx
Halmar  dropshipping  via API with DroFxHalmar  dropshipping  via API with DroFx
Halmar dropshipping via API with DroFxolyaivanovalion
 
Week-01-2.ppt BBB human Computer interaction
Week-01-2.ppt BBB human Computer interactionWeek-01-2.ppt BBB human Computer interaction
Week-01-2.ppt BBB human Computer interactionfulawalesam
 
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...amitlee9823
 
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...Valters Lauzums
 
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...amitlee9823
 

Último (20)

Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
 
Edukaciniai dropshipping via API with DroFx
Edukaciniai dropshipping via API with DroFxEdukaciniai dropshipping via API with DroFx
Edukaciniai dropshipping via API with DroFx
 
April 2024 - Crypto Market Report's Analysis
April 2024 - Crypto Market Report's AnalysisApril 2024 - Crypto Market Report's Analysis
April 2024 - Crypto Market Report's Analysis
 
Predicting Loan Approval: A Data Science Project
Predicting Loan Approval: A Data Science ProjectPredicting Loan Approval: A Data Science Project
Predicting Loan Approval: A Data Science Project
 
Jual Obat Aborsi Surabaya ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...
Jual Obat Aborsi Surabaya ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...Jual Obat Aborsi Surabaya ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...
Jual Obat Aborsi Surabaya ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...
 
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
 
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Thane West Call On 9920725232 With Body to body massage...
 
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...
 
Capstone Project on IBM Data Analytics Program
Capstone Project on IBM Data Analytics ProgramCapstone Project on IBM Data Analytics Program
Capstone Project on IBM Data Analytics Program
 
VidaXL dropshipping via API with DroFx.pptx
VidaXL dropshipping via API with DroFx.pptxVidaXL dropshipping via API with DroFx.pptx
VidaXL dropshipping via API with DroFx.pptx
 
Midocean dropshipping via API with DroFx
Midocean dropshipping via API with DroFxMidocean dropshipping via API with DroFx
Midocean dropshipping via API with DroFx
 
BabyOno dropshipping via API with DroFx.pptx
BabyOno dropshipping via API with DroFx.pptxBabyOno dropshipping via API with DroFx.pptx
BabyOno dropshipping via API with DroFx.pptx
 
BDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort Service
BDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort ServiceBDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort Service
BDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort Service
 
Halmar dropshipping via API with DroFx
Halmar  dropshipping  via API with DroFxHalmar  dropshipping  via API with DroFx
Halmar dropshipping via API with DroFx
 
Week-01-2.ppt BBB human Computer interaction
Week-01-2.ppt BBB human Computer interactionWeek-01-2.ppt BBB human Computer interaction
Week-01-2.ppt BBB human Computer interaction
 
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
 
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
 
(NEHA) Call Girls Katra Call Now 8617697112 Katra Escorts 24x7
(NEHA) Call Girls Katra Call Now 8617697112 Katra Escorts 24x7(NEHA) Call Girls Katra Call Now 8617697112 Katra Escorts 24x7
(NEHA) Call Girls Katra Call Now 8617697112 Katra Escorts 24x7
 
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
 
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts ServiceCall Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
 

Data-driven AI for Self-adaptive Information Systems

  • 1. Data-driven AI for Self-adaptive Information Systems Andreas Metzger BIOC & FAiSE Roma, MMXIX
  • 2. Agenda 2 Motivation and Background Case 1: Deep Learning for Proactive Process Adaptation Case 2: Policy-based Reinforcement Learning for Self-adaptive Cloud Services Discussion and Outlook BIOC/FAiSE, Roma, MMXIX
  • 4. Need for Self-Adaptation 4 Requirements Engineering Design Coding Deployment Operations Monitoring Adaptation Self-Adaption Engineering Automated perception, reasoning, actuation, … BIOC/FAiSE, Roma, MMXIX
  • 5. Data-driven AI as Enabling Technique BIOC/FAiSE, Roma, MMXIX 5 Image: https://www.datarevenue.com/ a.k.a Data-driven AI
  • 6. Data-driven as Enabling Technique Main Drivers BIOC/FAiSE, Roma, MMXIX 6 2020 2011 44 Zettabyte 1 Zettabyte Big Data Hardware e.g., GPU / TPU Algorithms e.g., Deep Learning
  • 7. Case 2Case 1 Data-driven as Enabling Technique BIOC/FAiSE, Roma, MMXIX 7 Unsupervised Learning Supervised Learning Reinforcement Learning No Feedback Labelled Training Data Reward for an Action
  • 8. Agenda 8 Motivation and Background Case 1: Deep Learning for Proactive Process Adaptation Case 2: Policy-based Reinforcement Learning for Self-adaptive Cloud Services Discussion and Outlook BIOC/FAiSE, Roma, MMXIX
  • 9. Process completiontCheckpoint j Process start Proactive Process Adaptation “in a Nutshell” BIOC/FAiSE, Roma, MMXIX 9 Monitor Predict Proactive adaptation planned / acceptable situations = Violation = Non- Violation  E.g., Delayed freight delivery E.g., Schedule air instead of land transport E.g., Freight delivery within 2 days Process Performance
  • 10. Considerations for Proactive Adaptation Accuracy • False violations  Unnecessary adaptations • False non-violations  Missed adaptations Earliness • Late prediction  no time for adaptation But: Trade-off between accuracy and earliness BIOC/FAiSE, Roma, MMXIX 10 [Metzger & Neubauer, 2018] [Teinemaa et al., 2019] BPIC 2017BPIC 2012Cargo 2000 Process completion Accuracy[MCC] Accuracy[AUC]
  • 11. Prediction Reliability for Proactive Adaptation BIOC/FAiSE, Roma, MMXIX 11 Reliability Estimate • Probability that individual prediction is correct  Distinguish between more or less reliable predictions Dynamic Approach • Use earliest prediction with sufficiently high reliability Aggregate Accuracy 75% 75% 75% 75% Prediction # 1 2 3 … Reliability Estimate 60% 90% 70% …
  • 12. Prediction Reliability for Proactive Adaptation Real-World Example 12BIOC/FAiSE, Roma, MMXIX Alarm about Delay Reliability Estimate Terminal Productivity Cockpit Data sources • Data streams from terminal equipment (1.3 mio states / month) • Integrated data of container moves (10,000 moves / month) Container terminal
  • 13. Prediction Reliability for Proactive Adaptation Realization of Dynamic Approach 13BIOC/FAiSE, Roma, MMXIX Process monitoring data at Checkpoint j RNN Model 1 RNN Model m … Ensemble Prediction [Tj = “non-violation”] Proactive Process Adaptation No Proactive Process Adaptation Prediction Tj Reliability estimate j [Tj = “violation”] [j  threshold] [j < threshold] Deep Learning
  • 14. RNNs as Base Learners RNN = Recurrent Neural Network Benefits • High accuracy [Tax et. al. 2017; Evermann et al. 2017, Metzger & Nebauer, 2018] • Arbitrary length process instances (without sequence encoding) • Predictions at any checkpoint Scalability • Long training time  Parallelization (Bagging)  Hardware speedups 14BIOC/FAiSE, Roma, MMXIX Hardware type Training time CPU 25 min GPU (Nvidia CuDNN) 8 min Google TPU (Tensorflow) 2 min
  • 16. Results 16BIOC/FAiSE, Roma, MMXIX “Cheap” adaptation “Expensive” adaptation Static checkpoint Dynamic approach No proactive adaptation Details in A. Metzger, A. Neubauer, P. Bohn, and K. Pohl, “Proactive process adaptation using deep learning ensembles,” in 31st Int’l Conference on Advanced Information Systems Engineering (CAiSE 2019), Rome, Italy, June 3-7, 2019, ser. LNCS, P. Giorgini and B. Weber, Eds., vol. 11483. Springer, 2019. Open Access: https://link.springer.com/chapter/10.1007%2F978-3-030-21290-2_34
  • 17. Agenda 17 Motivation and Background Case 1: Deep Learning for Proactive Process Adaptation Case 2: Policy-based Reinforcement Learning for Self-adaptive Cloud Services Discussion and Outlook BIOC/FAiSE, Roma, MMXIX
  • 18. Design Time Uncertainty • What are potential environment situations? • How does adaptation affect quality requirements? Learning for Self-Adaptation „in a Nutshell“ 18 Self-Adaptation Logic Knowledge Analyze Plan Monitor Execute System Logic Environment Solution Self-Adaptation Logic Knowledge Analyze Plan Monitor Execute System Logic Environment Online Learning Observe Induce Self-adaptive System Reference Model BIOC/FAiSE, Roma, MMXIX
  • 19. Learning for Self-Adaptation „in a Nutshell“ Most widely used: Reinforcement Learning 19BIOC/FAiSE, Roma, MMXIX Agent (with action- selection policy ) Environment Action atrt+1 st+1 State st Reward rt https://www.youtube.com/watch?v=gn4nRCC9TwQ
  • 20. Considerations for Self-adaptive Cloud Services Large, continuous environment space • Continuous state variables (e.g., workload) • Many state variables State of the Art: Value-based RL • Value function = expected cumulative reward per state • Requires manual quantization of states  error prone • Requires balancing exploration vs. exploitation 20BIOC/FAiSE, Roma, MMXIX [Marquezan et al., 2014]
  • 21. Coping with Large Environment Space Policy-based reinforcement learning • Approximates and generalizes over states • Probabilistic action selection Learning via policy gradient methods • Policy update according to gradient of objective function 21 Environment Action at rt+1 st+1 State st Reward rt BIOC/FAiSE, Roma, MMXIX Weights of neural network
  • 22. Coping with Large Environment Space Realization of policy-based RL for self-adaptive cloud services 22 Self-Adaptation Logic Knowledge Analyze Plan Monitor Execute BIOC/FAiSE, Roma, MMXIX Self-Adaptation Logic Knowledge Analyze Plan Monitor Execute Policy-based Reinforcement Learning at rt+1 t+1 Monitor Execute Action- Selection Self-Adaptation Logic Policy  (Knowledge) st Policy Update st+1
  • 23. Coping with Large Environment Space Benchmark Example State Reward 23BIOC/FAiSE, Roma, MMXIX [Klein et al., 2014]Brownout-RUBiS Nbr Requests CPU Load Actual Recomm. Rate Response time (ms) Auction Web Application • User requests served with recommendations for similar products Action Set dimmer value = per-request probability that recommendation engine is activated
  • 24. Results 24 Stable Workload Off/on workload BIOC/FAiSE, Roma, MMXIX Growing workload Cyclic workload red = avg. reward blue = workload
  • 26. Agenda 26 Motivation and Background Case 1: Deep Learning for Proactive Process Adaptation Case 2: Policy-based Reinforcement Learning for Self-adaptive Cloud Services Discussion and Outlook BIOC/FAiSE, Roma, MMXIX
  • 27. Lessons Learned Deep learning requires little hyper-parametrization • If enough good quality data is available / can be collected Use deep learning to increase productivity of information system engineering Prediction reliability offers additional decision support • Operators put more trust in predictions • Cost savings Augment predictions with reliability estimates, confidence intervals, error ranges, etc. Data quality is a key concern • “Garbage in – garbage out” • Missing data, data accuracy, timeliness, timestamps (clocks), … • Very resource and time-intensive Plan sufficient time and effort for data quality, data integration and refinement of data collection 27BIOC/FAiSE, Roma, MMXIX
  • 28. Limitations Risks • Adaptations may “harm” environment (embedded systems / CPS) • Adversaries may manipulate training data Skills 28BIOC/FAiSE, Roma, MMXIX (Source: IDC) Year Gap (total EU) 2020 (baseline) 530,000 2020 (high-growth) 3,500,000
  • 29. Outlook 29BIOC/FAiSE, Roma, MMXIX Decision Making • Deep learning for high accuracy descriptive and predictive analytics • Reinforcement learning for solving complex planning and decision problems Enactment • Actuation driven by AI decisions • Safety and trustworthiness as key requirements for adoption
  • 30. Thanks! BIOC/FAiSE, Roma, MMXIX 30 Research leading to these results has received funding from the EU’s Horizon 2020 research and innovation programme under grant agreements no. 731932 – http://www.transformingtransport.eu 732630 – http://www.big-data-value.eu 780351 – https://enact-project.eu
  • 31. References [Evermann et al. 2017] Evermann, J., Rehse, J., Fettke, P.: Predicting process behaviour using deep learning. Decision Support Systems 100 (2017) [Klein et al., 2014] Klein, C., Maggio, M., Arz´ en, K.E., Hernández-Rodriguez, F.: Brownout: Building more robust cloud applications. In: 36th Intl Conf. on Software Engineering (ICSE 2014). pp. 700–711. ACM (2014) [Metzger & Neubauer, 2018] A. Metzger and A. Neubauer, “Considering non-sequential control flows for process prediction with recurrent neural networks,” in 44th Euromicro Conference on Software Engineering and Advanced Applications (SEAA 2018), Prague, Czech Republic, August 29-31, 2018, T. Bures and L. Angelis, Eds. IEEE Computer Society, 2018, pp. 268–272. [Metzger et al., 2019] A. Metzger, A. Neubauer, P. Bohn, and K. Pohl, “Proactive process adaptation using deep learning ensembles,” in 31st Int’l Conference on Advanced Information Systems Engineering (CAiSE 2019), Rome, Italy, June 3-7, 2019, ser. LNCS, P. Giorgini and B. Weber, Eds., vol. 11483. Springer, 2019. Open Access: https://link.springer.com/chapter/10.1007%2F978-3-030-21290-2_34 [Tax et. al. 2017] Tax, N., Verenich, I., Rosa, M.L., Dumas, M.: Predictive business process monitoring with LSTM neural networks. In: Dubois, E., Pohl, K. (eds.) CAiSE 2017, Essen, Germany, June 12-16, 2017. LNCS, vol. 10253. Springer (2017) [Teinemaa et al. 2019] Teinemaa, I., Dumas, M., Rosa, M.L., Maggi, F.M.: Outcome-oriented predictive process monitoring: Review and benchmark. ACM Transactions on Knowledge Discovery from Data (TKDD) 13 (2019) BIOC/FAiSE, Roma, MMXIX 31

Notas do Editor

  1. (1 ZB = 1 Trillion GB)
  2. 1,16 MEUR für paluno
  3. e.g., average reward per learning iteration