1
ARTIFICIAL INTELLIGENCE FOR ACCELERATED GROWTH
TRANSVERSAL TECHNOLOGIES ENABLE AMBITIOUS APPLICATION SCENARIOS
AI
AI DATA BIG
COMPUTE POWER
AMAZON GOOGLE FACEBOOK
APPLE
ALIBABA BAIDU TENCENT
XIAOMI
AI DATA BIG
COMPUTE POWER
SECURE DISTRIBUTED AI
APPLICATION SPECIFIC AI
DIFFERENTIATING
FROM OTHER MAJOR WW INITIATIVES
AMBITIOUS
COMPREHENSIVE INCLUSIVE
COMBINING
ACADEMIC EXCELLENCE
DIGITAL TECHNOLOGY LEADERSHIP INDUSTRY LEADERSHIP
APPLICATION POTENTIAL
GROUNDBREAKING FOUNDATIONAL RESEARCH
APPLIED R&D PLATFORM BUILDING
INDUSTRY SUPPORT ACTIONS OUTREACH, EDUCATION
AND POLICY SUPPORT
COMPREHENSIVE & INDUSTRY DRIVEN
GENERAL PUBLIC POLICY MAKERS
ACADEMIC PARTNERS STRATEGIC RESEARCH INDUSTRY
Strengthen AI impact on industrial processes
Strengthen AI impact on Offering of companies Strengthen AI impact on
Business models New start-ups / scale-ups in
AI and related domains
Ad d re ss in g in d us tr y ne ed s
Bu ildi ng o n s tr en gt h s
GROUNDBREAKIN G FOUNDATIONAL
RESEARCH APPLIED R&D
PLATFORM BUILDING INDUSTRY
SUPPORT ACTIONS OUTREACH, EDUCATION AND POLICY SUPPORT
A CONTINUUM FROM EXPLORATION TO IMPLEMENTATION
WITH ALL KEY FLEMISH STAKEHOLDERS
Strategic foundational research and tech. exploration
Applied R&D and platform building
Implementation and demonstration in applications
Outreach – education – Awareness creation
Guiding measures and policies wrt all aspects of society
Application
Domain
11
A CONTINUUM FROM EXPLORATION TO IMPLEMENTATION
WITH KEY FLEMISH STAKEHOLDERS
Application Domain
Application Domain Application
Domain
GENERAL PUBLIC POLICY MAKERS
ACADEMIC PARTNERS STRATEGIC RESEARCH INDUSTRY
PERSONALISED HEALTH SMART
MOBILITY
SMART INDUSTRY
13
SMART INDUSTRY
SMART INDUSTRY
Smaller production series – faster product updates More variability – different product flavors Reduce shipping cost
…
Digital Factory Twin Smarter quick learning cobots Human-machine interaction Augmented and virtual reality
Sensors monitoring the factory floor, Safety Highly skilled workers
SMART INDUSTRY
SMART INDUSTRY
Innovation, validation and demonstration in
vehicle industry, in the machine building industry
and in production environments
15
SMART MOBILITY
SMART MOBILITY
Reduce congestion Improve safety Reduce social exclusion Improve sustainability Regain public space
Shared Situational Awareness Intelligent Intersections Joint & Synchromodal scheduling Spare capacity tracking Mobility shaping Last mile delivery END GOALS
ENABLING USE CASES
mobi
SMART MOBILITY
SMART MOBILITY
SHARED WIRELESS CONNECTIVITY
v 3 v 2
v 1
DISTRIBUTED INTELLIGENCE &
SENSOR FUSION
OPEN DIGITAL TRANSPORT
TWIN
17
PERSONALISED HEALTH
PERSONALISED HEALTH
Smart meter Domotics
Dx IT
PERSONALISED HEALTH
PERSONALISED HEALTH
Data fusion
@datacenter
@extreme edge EHR
AI
Trusted
Secure
19
PERSONALISED HEALTH
PERSONALISED HEALTH
Leading R&D organizations in all disciplines Reputation for clinical trials and biotech ecosystem
FIT FIT
…
…
SMART MOBILITY
SMART INDUSTRY
PERSONALISED HEALTH
TECHNOLOGY
PLATFORMS CYBER
SECURITY
NETWORK TECHNOL
OGY
ARTIFICIAL INTELLIGENCE
SMART MOBILITY
SMART INDUSTRY
PERSONALISED HEALTH SMART
INFOTAINMENT
21
TECHNOLOGY PLATFORMS CYBER
SECURITY
NETWORK TECHNOLOG
Y
ARTIFICIAL INTELLIGENCE SECURE LUMPED &
DISTRIBUTED
SMART MOBILITY
SMART INDUSTRY
SMART ENERGY
PERSONALISED HEALTH SMART
INFOTAINMENT
SMART CITY
ENABLING PLATFORMS
SYSTEM ACTUATORS
DATA OBSERVATIONS DISTURBANCES
A PRIORI INFORMATION
- Objectives - Population based info - Expert’s expertise - Rules, syntaxis - Hypothesis, query, … - …..
SENSING DEVICES
Audio-visual, Chemical, Mechanical , Electrio-Magnetic, Biological,….
HUMAN IN-THE- LOOP
Decision support systems Data based AI
Knowledge based AI Hybrid AI
Explainable AI
LUMPED AI à REAL-WORLD USER-CENTRIC CYBER-PHYSICAL SYSTEMS
23 23
ENABLING PLATFORMS
§distributed collaborative AI system will execute simultaneously on many devices and servers.
§communication between agents and in the data chain
§reduction of the data to core features
§meet the timing & energy constraints, need for specialized hardware architectures at edge Extreme
Edge Edge
MULTI-AGENT SECURE DISTRIBUTED AI
ENABLING PLATFORMS
Extreme Edge Edge
Storage & Processing €, GB
Latency (us)
Privacy
Adaptive learning Bandwidth €, Mbps
Utility
Security: Confidentiality, Integrity, Availability Reliability
Energy
Sensor Joint
Distributed
Trust
25 25
ENABLING PLATFORMS
SNN CNN LSTM
Deep learning Brain inspired
Cortical
Conventional ML
Support
Vector Markov chains Decision trees
Algorithm selection for customer problem
Network selection and optimization
GPU FPGA
Low precision Selection of
implementation platform
Productization Time line
IC-link 2 years IC-link
18 months Accelerator Software/PC FPGA/board
3 months
IC-link 6-9 months Evaluation
imec 5 years Analog/CIM Lower accuracy acceptable No power budget limitations? Limited power budget
ASIC/board Imec platform Imec platform
… … …
NEW COMPUTE @ EXTREME EDGE
ENABLING PLATFORMS
New ways of acquiring knowledge Efficient collaborative work New medical practices Augmented daily life
Gesture recognition Authentication
Haptic feedback
Medical imaging Directional sound
AR/VR
27 27
INDUSTRY ACCESS AND ADOPTION
Ports of entry Google Cities
MOW CoT
AWV Public Transport
VRT
MAKE MOBI Be-Mobile
Telecom Operators FEBIAC BIPT
Nokia
Agfa KBC EURid
Sony
VRT
BNP Paribas
Proximus ICT control
Microsoft Barco
B-Mobile B-Post PWC
Deloitte Belnet
SilexInside Vasco
Prodata Systems Luciad
Nokia
SDNsquare Televic
UP-nxt
Verizon Febelfin
VOKA Agoria
3if.be
Accenture KPMG
Restore Enervalis
Engie
Vasco NIKO
Energis Eandis
ML6 SMALS
Python Predictions
Sporasteria Nixis
Medialaan Carrefour
Van de Velde Sony Atos
Edebex
KBC Alliance
Argenta Ablynx
Zembro Merck Ventures CMAST
Bayer cropsciences
Vlaams instituut gezond leven
Flanders Bio Inovigate Roche Genzyme
Remynd
NIKO
Melexis
OnSemi Microsoft
Grohe MDxHealth
Peira
Regenesys miDiagnostics
Multiplic Dekimo Anacura
Cochlear Amgen
Pharmafluidics JSR micro
Regional and Academic clinics Fujirebio J&J/Janssens
ATMI/Pall ArgenX
Antleron UCB
Galapagos Biocartis
Biogazelle Apitope
Verzekeraars
Mutualiteiten
RIZIV Happy aging
coma
Barco Materialise
Attentia AMS
Dentsply sirona
Materialise R.A.V.E.N.
Fujifilm JSR
NXP Sentiac
Datwyler ICsense ToWalkAgain
Caeleste
Melexis Ansem
Zenso
OnSemi
Xenics Byteflies QUAD
Barco Nokia
Verhaert
Sentiance
Robovision Microsoft
IBM Amazon Colruyt
DHL Bombardier
CNH
Picanol
Ordina Cegeka
Movilitas
AI innovation services
INDUSTRY ACCESS AND ADOPTION
Artificial Intelligence Innovation Service
AI awareness and outreach
▪ neutral and comprehensive AI ‘starting guide’ (whitepaper); online AI compass chatbot
▪ pro-active outreach (AI 101 trainings, AI Summer School, AI hackathons)
▪ campaigns to raise public awareness, involvement and empowerment
▪ organize an annual competition by defining a challenge that requires a distributed AI solutions
Flemish testbeds and access to AI technology
▪ low-cost AI hard- and software platform, including standard software platform with AI building blocks
▪ BYOData, AI FabLab ; MakeLab (Cobots, AR/VR, AI) – Future assembly and smart productivity labs
▪ access to an extended (mobile) infrastructure for prototyping and testing edge and extreme edge applications
▪ Access to new platforms
Strong AI community
29 29
POLICY THINK TANK
Policy support
Co-design approach
Recommendations avoiding AI bias, Explainable Artificial Intelligence, improving algorithmic/AI transparency and accountability etc.
Link up with EU policy
Involve stakeholders for complex ethical and policy issues Drafting of ethical codes and guidelines
Societal outreach
Make AI developments visible for society, and promote ethical and socially responsible deployment, focusing on both positive and negative points
Public outreach efforts, Public awareness raising and empowerment
Following up on media representation of AI: mapping change in different (mis)representations of AI
Thought leadership
Societal and ethical aspects of AI Attract thought international leaders Programme of work related to societal and ethical aspects
Legal challenges of AI
Baseline principles, constraints and touchstone cases Systematic inventory of the issues to adjust current laws Identify systemic changes requiring legislative shifts New public policy models to regulate such systems Certification of adaptive and non-deterministic systems liability allocation amongst the actors
Openness of AI systems, e.g. source code, data, safety Adherence to the European standards of privacy and data
Policy and society validation/certification
test environment/regulatory sandbox for Human AI data mapping, bias detection, ethical issues, ‘AI exam’
acceptance and behavioural tests, co-working with AI