Upload
vincent-chu
View
222
Download
0
Embed Size (px)
Citation preview
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 1/25
Handling Concept Drift:An Application Perspective
M kola Pechenizki TU Eindhoven the Netherlands
Indrė Žliobaitė, TU Eindhoven, the Netherlands
TU/e, Eindhoven
Aug. 20th, 2010
DHDHD 2010
Workshop
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 2/25
Learning from Evolving Data
Joao GamaUniversity of Porto, PORTUGAL
Myra Spiliopoulou
Ernestina MenasalvasTechnical Univ. of Madrid, SPAIN
Athena Vakali
and the Chairs of the HaCDAIS Workshop:
Mykola Pechenizkiy, Eidhoven Univ. of Technology, NETHERLANDS
Indrė Žliobaitė, Eidhoven Univ. of Technology, NETHERLANDS
University of Magdeburg, GERMANY University of Thessaloniki, GREECE
ECML-PKDD Conference
TutorialBarcelona, Sept. 24th, 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 3/25
CD refers to non-stationary supervised learning problems
but there are different types of CD
and different types of applicationsPersonal recommenders, spam filters, fraud detection,navigation are affected by drifts coming from different sources
Concept Drift: Application PerspectiveConcept Drift: Application PerspectiveConcept Drift: Application PerspectiveConcept Drift: Application Perspective
(3)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 4/25
View CD research from an application perspective
What is the match between the mainstream CD researchassumptions and properties of the applications?
Identify promising future research directions from theapplication perspective
MotivationMotivationMotivationMotivation
(4)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
We will talk about
Why changes appear in different applications?
What are the properties of CD application tasks?
How the application tasks can be categorized in terms of these basic properties?
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 5/25
Categorization of applications (Žliobaitė& Pechenizkiy, 2010)
(5)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 6/25
DATA task (detection, classification, prediction, ranking)
type (time series, relational, mix)
organization (stream/batches, data re-access, missing)
DRIFT change type (sudden, gradual, incremental, reoccurring)
source (adversary, interests, population, complexity)
Properties of the tasksProperties of the tasksProperties of the tasksProperties of the tasks
(6)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
expec a on unpre c a e, pre c a e, en a e
DECISIONS and GROUND TRUTH
labels (real time, on demand, fixed lag, delay)
decision speed (real time, analytical)
ground truth labels (soft, hard)
costs of mistakes (balanced, unbalanced)
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 7/25
Categorization of applications (Žliobaitė& Pechenizkiy, 2010)
(7)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 8/25
Landscape of applicationsLandscape of applicationsLandscape of applicationsLandscape of applications
Types of appsIndustries
Monitoring/control
Personal assistance/personalization
Managementand planning
Ubiquitousapplications
Security, Police Fraud detection,
insider trading
detection,
adversary actions
detection
-------- Crime volume
prediction
Authentica-
tion,
Intrusion
detection
Finance, Banking,
Telecom, Credit
Monitoring &
management of
Product or service
recommendation,
Demand
prediction,
Location
based
(9)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
Scoring, Insurance,Direct Marketing,
Retail, Advertising, e-
Commerce
customersegments,
bankruptcy
prediction
includingcomplimentary
response rateprediction,
budget
planning
services,related ads,
mobile apps
Education (higher,
professional, children,e-Learning)
Entertainment, Media
Gaming the
system,Drop out
prediction
Music, VOD, movie,
learning objectrecommendation,
adaptive news
access, personalized
search
Player-
centeredgame design,
learner-
centered
education
Virtual reality,
simulations
… … … … …
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 9/25
Categorization of applications (Žliobaitė& Pechenizkiy, 2010)
(10)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
STREAMS/
SENSORS
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 10/25
CFB Boiler OptimizationCFB Boiler OptimizationCFB Boiler OptimizationCFB Boiler Optimization
(11)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 11/25
data collected from a typical experimentation with CFB boiler
Online mass flow prediction in CFB boilersOnline mass flow prediction in CFB boilersOnline mass flow prediction in CFB boilersOnline mass flow prediction in CFB boilers
(12)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
asymmetric natureof the outliers
short consumptionperiods within
feeding stages
Pechenizkiy et al. 2009
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 12/25
PREDICTIVE
Categorization of applications (Žliobaitė& Pechenizkiy, 2010)
(13)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 13/25
Food sales prediction: utility of Belgium milk in Sep. 2009Food sales prediction: utility of Belgium milk in Sep. 2009Food sales prediction: utility of Belgium milk in Sep. 2009Food sales prediction: utility of Belgium milk in Sep. 2009
(14)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 14/25
Challenges in food sales prediction (Zliobaite et al., 2009)Challenges in food sales prediction (Zliobaite et al., 2009)Challenges in food sales prediction (Zliobaite et al., 2009)Challenges in food sales prediction (Zliobaite et al., 2009)
(15)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 15/25
ReoccuringReoccuringReoccuringReoccuring andandandand suddentsuddentsuddentsuddent dritftdritftdritftdritft in food salesin food salesin food salesin food sales
(16)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 16/25
DIAGNOSTIC S/
Categorization of applications (Žliobaitė& Pechenizkiy, 2010)
(17)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
DECISION MAKING
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 17/25
predict the sensitivity of a pathogen to an antibiotic based on dataabout the antibiotic, the isolated pathogen, and the demographic andclinical features of the patient.
Antibiotic Resistance Prediction (Antibiotic Resistance Prediction (Antibiotic Resistance Prediction (Antibiotic Resistance Prediction (TsymbalTsymbalTsymbalTsymbal et al., 2008)et al., 2008)et al., 2008)et al., 2008)
(18)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 18/25
How Antibiotic Resistance HappensHow Antibiotic Resistance HappensHow Antibiotic Resistance HappensHow Antibiotic Resistance Happens
(19)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 19/25
PERSONALIZATION /
RECOMMENDERS
Categorization of applications (Žliobaitė& Pechenizkiy, 2010)
(20)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 20/25
We Know What You Ought
Recommender Systems
Lessons learnt from Netflix:Temporal dynamics is important
Classical CD approaches may not work
(Koren, SIGKDD 2009)
(21)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 21/25
Something Happened in Early 2004
(22)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 20102004
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 22/25
Are movies getting better with time?
(23)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 23/25
Multiple sources of temporal dynamics
Both items and users are changing over time
Item-side effects:
Product perception and popularity are constantly changing
Seasonal patterns influence items’ popularity
User-side effects:
Customers ever redefine their taste
(24)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
Transient, short-term bias; anchoring Drifting rating scale
Change of rater within household
Common “concept drift” methodologies won’t hold.
E.g., underweighting older instances is unappealing
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 24/25
OutlookOutlookOutlookOutlook
From general methods to more specific application orientedproblems like
delayed labeling,label availability,
cost-benefit trade off of the model update.
(25)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010
Changing the focus to
change description,
prediction reoccurring contexts and
meta learning in addition to change detection.
8/7/2019 Handling Concept Drift: An Application Perspective
http://slidepdf.com/reader/full/handling-concept-drift-an-application-perspective 25/25
Take Bowling MessageTake Bowling MessageTake Bowling MessageTake Bowling Message
There is no uniform concept as ‘concept drift’
(26)(c) Gama, Menasalvas, Spiliopoulou, Vakali – Barcelona, 24th Sept. 2010