GRSI focuses on the research and development of intelligent systems based on (1) extracting interesting patterns from moderate and large complex data (Data Mining) and (2) learning from them (Machine Learning) for helping experts by means of the building of decision support systems. In this framework, GRSI works on different stages of the process of data mining: pre-processing, characterization of data sets, analysis for a better understanding and improvement of machine learning techniques, methodologies to evaluate learners, and post-processing. During the last few years, the research has mainly focused on learning methods inspired by natural principles and analogy. The group is known for its expertise on Evolutionary Computation, Soft Case-Based Reasoning, and Neural Networks.
Research on Data mining at Research Group in Intelligent Systems
1. Data Mining
g
Research, Innovation and Development for
supporting experts in their decisions
pp g p
Supporting experts is helping them to take more
effective,
effective efficient and reliable decisions
Research, Innovation and Development for
Page 1
supporting experts in their decisions
2. Campus laSalle Barcelona
p
• Campus with 5 buildings
p g
• 4000 students.
• More than 100 years training
highly qualified professionals.
Entrepreneurship
Research Differential
ff
Groups methodology
Prestige and International
innovation character
La Salle
Laboratories and
Technova
infrastructures
Barcelona
Research, Innovation and Development for
Page 2
supporting experts in their decisions
3. Outline
• Research Group in Intelligent Systems
– Descriptors
– Research on Data Mining
• Projects related to Health Sciences
– Decision support system for Breast Cancer
– Data Mining as support for melanoma experts
Research, Innovation and Development for
Page 3
supporting experts in their decisions
4. What is Artificial Intelligence ( )
g (AI)?
• John McCarthy coined AI term in 1956 as ‘the science and engineering of
making intelligent machines’ at a conference at Dartmouth College. Intelligent
machine terms refer to the capability of performing intelligent human processes
as:
– Learning
– Reasoning
– Problem solving
– Perception
– Language understanding
• AI has become an essential part of the technology industry, providing the
heavy lifting for many of the most difficult problems in computer science.
– Prediction
– Classification
– Regression
– Clustering
– F ti ti i ti
Function optimization
Research, Innovation and Development for
Page 4
supporting experts in their decisions
5. Why is AI p
y powerful?
• The power resides in the combination of disciplines that tackle the same
problems as AI: learn and understand, to solve problems and to make decisions.
• AI is fed from many disciplines
– Phil
Philosophy: L i methods of reasoning, mind as physical system,
h Logic, th d f i i d h i l t
foundations of learning, language, rationality.
– Mathematics: Formal representation and proof, algorithms, computation,
(un)decidability, (in)tractability.
( ) ( )
– Statistics : Modeling uncertainty, learning from data.
– Economics: Utility, decision theory, rational economic agents.
y y g
– Neuroscience: Neurons as information processing units.
– Psychology / NeuroScience: How do people behave, perceive, process
cognitive information, represent knowledge
information knowledge.
– Computer Engineering: Building fast computers.
– Control Theory: Design systems that maximize an objective function over time.
– Linguistics: Knowledge representation, grammars.
Research, Innovation and Development for
Page 5
supporting experts in their decisions
6. Basis of Artificial Intelligence
g
Philosophy Mathematical. Computational Cognitive Computational
• Discussion about • Philosophic bases linguistic psychology engineering
the
th possibility of a
ibilit f requires f
i formall • Understanding • Behavior theories, • Some mechanism,
mechanical rules. language requires rational behave hardware and tools
intelligence. understanding of basis. are required for AI.
the subject matter
and th context.
d the t t
Research, Innovation and Development for
Page 6
supporting experts in their decisions
7. A possible map of the current AI
p p
• Non monotonic reasoning • Evolutionary Computation
• Model based reasoning • Case-Based Reasoning
• Constraint ti f ti
C t i t satisfaction •R i f
Reinforcement Learning
tL i
• Qualitative reasoning • Neural Network
• Uncertain reasoning • Data Analysis
• Temporal reasoning
• Heuristic search Machine
Reasoning
Learning
Robotics,
perception
Knowledge
and natural
• Logic Management
language
g g
• Multiagents systems processing
• Decision Support System
• Knowledge management • Robotics and control
• Knowledge representation • Natural Language Processing
• Ontology and semantic web • Artificial vision
• Computer-Human interaction • Speech recognition
Research, Innovation and Development for
Page 7
supporting experts in their decisions
8. Research Group in Intelligent Systems
p g y
• GRSI is a research group focused on Machine Learning, especially in
the field of Knowledge Discovery from Databases (KDD) (also
known as Data Mining) for extracting interesting patterns from
moderate and large complex data.
– Created in 1994
– Recognized as consolidated by Generalitat de Catalunya since 2002.
– Group is composed of 18 members.
– F ll professor J
Full f Josep M í G
María Garrell i th head of th group.
ll is the h d f the
– We tackle classification, prediction, regression, optimization,
recommendation and diagnosis problems which occur in complex and
huge volume of data in domains such as….
Health Energy Telematic Learning
Research, Innovation and Development for
Page 8
supporting experts in their decisions
9. 9
Data Mining sets the difference
g
Value
Wisdom How can we help them?
(Knowledge+ experience)
( g )
Knowledge Why they are getting worse?
How far do you (Information
(Information+ rules)
want to go?
Information
How many patients got worse?
(Data + Context)
How many patients are in the
Data Intensive Care Unit?
KDD allow experts to extract useful and hidden knowledge from data.
The approach is valid for any domain
Volume Business, space, communication media, insurance companies,
financial services, health sciences, games, etc.
Research, Innovation and Development for
Page 9
supporting experts in their decisions
10. Research lines
• GRSI works on the different stages of Knowledge Data
Discovery: characterization, pre-processing, analysis for a better
understanding and improvement of machine learning techniques,
methodologies to evaluate learners and post-processing.
Problem Data
Analysis Analysis
Data
D t
Processing
Production Knowledge
Modeling
Evaluation
Research, Innovation and Development for
Page 10
supporting experts in their decisions
11. Data Mining applications
g pp
Clustering
Classification
Knowledge Association
rules
discovery
Regression
Research, Innovation and Development for
Page 11
supporting experts in their decisions
12. Techniques
q
Create
C t computer t Solves
S l new problemsbl Simulate
Si l t some Measure th
M the
Neura Networks
s
Complex metrics
s
omputation
n
Soft Case-Based Reasoning
g
programs inspired by using other previously properties of biological ‘complexity’ of a
the process of natural solved. neural networks to problem in terms of
selection and genetic replicate how ‘our class separability and
xity
E.g. Retrieve a set of
g
al
laws for search
search, neurons works.
neurons’ works the discriminant power
Evolutionary Co
similar mammographic
optimization and images to a expert E.g. Build system that of features.
machine learning. according to a set of is able to replicate a E.g. Relate how the
E.g. Look for the best criteria. behavior based on a data complexity affects
equation that set of inputs and
f the performance of
f f
C
represents a set of outputs previously algorithms in order to
points. known. adjust them properly.
Research, Innovation and Development for
Page 12
supporting experts in their decisions
13. GRSI members
Director Member emeriti
Garrell Guiu Josep M PhD
Guiu, M., Bacardit, Jaume, PhD,
Bacardit Jaume PhD UK
Assistant executive director Castanys Tutzó, Mireia, PhD
Fornells Herrera, Albert, PhD Farguell Matesanz, Enric, PhD
Members
M b Llorà, Xavier, PhD,
Ll à X i PhD USA
Bernadó Mansilla, Ester, PhD Martorell Rodon, Josep Maria, PhD
Camps Dausà, Joan
p , Macià Antolínez, Núria, PhD
Corral Torruella, Guiomar, PhD Nettleton, David, PhD
García Piquer, Álvaro Orriols-Puig, Albert, PhD, USA
Garriga Berga Carles PhD
Berga, Carles, Salamó Llorente Maria PhD
Llorente, Maria,
Golobardes Ribé, Elisabet, PhD Pazienza de Filippis, Giovanni Egidio,
Nicolàs Sans, Rubén PhD, Hungary
Rios Boutín, Joaquim
Sancho Asensio, Andreu
Teixidó Navarro, Francesc
Navarro
Vernet Bellet, David
Research, Innovation and Development for
Page 13
supporting experts in their decisions
14. Outline
• Research Group in Intelligent Systems
– Descriptors
– Research on Data Mining
• Projects related to health Sciences
– Decision support system for Breast Cancer
– Data Mining as support for melanoma experts
Research, Innovation and Development for
Page 14
supporting experts in their decisions
15. Breast cancer diagnosis
g
TIC2002-04160-C02-02
• Goal: Development of a tool for intelligent retrieval of mammographic images
by content analysis in order to help experts in the diagnosis process.
Digitalization,
Di it li ti
Retrieval of
Mammographic capture segmentation and feature Diagnosis support
mammographic records
extraction
Research, Innovation and Development for
Page 15
supporting experts in their decisions
16. Melanoma diagnosis
g TIN2006-15140-C03-03
• Goal: Help experts in the melanoma characterization for improving melanoma diagnosis
Characterization Patterns Medical protocols
Decision support systems
pp y
Research, Innovation and Development for
Page 16
supporting experts in their decisions
17. Telematic vulnerabilities TIN2006-15140-C03-03
• Goal: Provide tools to the security analyst for helping them in the security analysis
tasks by means of the identification of problematic situations which are not
obviously.
CONSENSUS ANALIA
Research, Innovation and Development for
Page 17
supporting experts in their decisions
18. Active Demand Management
g
CEN200710126
• Goal: Integration of system (OS), distribution (OD) and sellers (CM) agents for an
efficient management of demand
demand.
Design and
OS implementation of agent Design and develop
communication intelligent devices for the
g
management of energy
demand at home
OD GAD CM
Develop of specific rates
CL Pattern identification Analysis of client demand
for clients
Research, Innovation and Development for
Page 18
supporting experts in their decisions
19. 19
Integris: INTElligent GRId Sensor
communications FP7 ICT-Energy-2009-1, Objective 6.5, #247938
• Goal: Integration and management of communication technologies in smart-
grids for assuring QoS, security and reliability.
Research, Innovation and Development for
Page 19
supporting experts in their decisions
20. Thanks for your attention
y
For further information visit http://www.salle.url.edu/GRSI
or send an email to afornells@salle url edu
afornells@salle.url.edu
Research, Innovation and Development for
Page 20
supporting experts in their decisions