COGNITION COGNITION by by Horst Forster Horst Forster Interfaces, Knowledge and Content Technologies; Interfaces, Knowledge and Content Technologies; Applications, Information Market Applications, Information Market DG Information Society DG Information Society Brussels, 29.01.2003 Brussels, 29.01.2003 Cog0
20
Embed
COGNITION by Horst Forster Interfaces, Knowledge and Content Technologies; Applications, Information Market DG Information Society Brussels, 29.01.2003.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
COGNITIONCOGNITION
byby
Horst ForsterHorst Forster
Interfaces, Knowledge and Content Technologies; Interfaces, Knowledge and Content Technologies; Applications, Information MarketApplications, Information Market
DG Information SocietyDG Information Society
Brussels, 29.01.2003Brussels, 29.01.2003
Cog0
Subject area:
Digital content
Remit:
Make information accessible to all
Overcome multilingual barriers
Three pillars:
Market Applications
Research
Our missionOur mission
Cog1
Scope and instrumentsScope and instruments
Research and Technology DevelopmentResearch and Technology Development
Interfaces and CognitionInterfaces and Cognition
Knowledge management and Content creationKnowledge management and Content creation
Interpretation requires more than ‘data’ computation
knowledge of context / situations
knowledge of tasks / goal
Knowledge-based systems
So where are these systems?
Cog7
the elusive real-world…the elusive real-world…
Process lines in factories: visual recognition capabilities empower robots to do intricate tasks … yet they can’t help out in the home
The Voice Web: telephone speech recognition can allow automation of customer services … but in very specific application domains
The issue at hand:
constrained conditions and artificial environments
versus
the endlessly variable ‘real world’
Cog8
more and more knowledge…more and more knowledge…
‘Real world’ environments and applications:different contexts and situations requires
knowledge to be adapted broader concerns require more and different kinds
of knowledge
systems fail in situations not envisaged in their design (lack of robustness)
systems function within narrow application domain (limited versatility)
Cog9
Systems that ‘know’ what they Systems that ‘know’ what they are doingare doing
Build bigger knowledge-bases?
No! – search becomes computationally intractable!
Cognitive systems?
systems that can extend their knowledge by reasoning and by learning from what they perceive
and use this knowledge for the purpose of achieving their performance goals
Cog10
Cognition - Why now? Cognition - Why now?
progress in artificial intelligence
progress in cognitive neuroscience
cheap and plentiful computing power
commoditisation of miniaturised sensors
Cog11
a window to the world…a window to the world…
Machines that ‘see’ the user: i. interprets gestures / behaviour adaptive,
natural interfaces
Machines that ‘see’ the environment: ii. performs visual tasks that are difficult, tedious or
impossible for humansiii. moves and navigates itself to perform various
tasks for humans (autonomous mobile robots)i and ii ONLY
Cog12
Research agendaResearch agenda
1. multimodal interfacesresearch: recognise speech and gestures; learning and adapting to behaviourtarget: facilitating natural interaction between humans and machines applications:user-friendly PCs(!!),wearable interfaces, smart clothes, intelligent rooms, collaborative work environments
Cog13
Research agendaResearch agenda
2. Semantic–based knowledge systemsresearch: annotation of content with ‘formalised’ knowledge; development of machine-processable vocabularies (ontologies) target: automated systems for knowledge handling applications:
knowledge discovery in databases, dynamic information services, automated diagnosis and decision support systems, Semantic Web (in time!)
Cog14
Research agendaResearch agenda
3. Cognitive systems research: integrate perception, action, representation, reasoning and learning in systems target: systems that do more than ‘canned’ predefined tasks; systems that handle new task specifications on the fly applications:image interpretation (eg: in medical or aerospace domains), behavioural interpretation (eg: in crowd surveillance or traffic monitoring), specialised interfaces (eg: sign language interpretation)
Cog15
progressprogress??
eg: machine vision systems:
10 y ago: maintain a description of the identity of objects or actors in a scene,
5y ago: … identity and the location …,
recently: … identity, location and the role …,
towards systems that are aware
Cog16
… … ultimately ultimately ……
The introduction of cognitive capabilities into systems will provide a quantum leap in functionality and flexibility: