CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
Growth Technologies
1. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGEAARHUS
UNIVERSITY
DEPARTMENT OF ENGINEERING
AU
TECHNOLOGIES
OF THE FUTURE
EXPONENTIAL GROWTH
2. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DEPARTMENT OF ENGINEERING
To Educate more Outstanding Engineers on All Levels
To Perform Engineering Science Research at the Highest International Level
Learn about us – follow the link.
3. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
EXPONENTIAL GROWTH
An analysis of the history of technology shows that technological change is exponential,
contrary to the common-sense “intuitive linear” view.
The exponential function: C is the initial amount. t is the time period.
(1 + r) is the growth factor, r is the growth rate.
y = C (1 + r) t
4.
5.
6. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DEPARTMENT OF ENGINEERING
To Educate more Outstanding Engineers on All Levels
To Perform Engineering Science Research at the Highest International Level
Associate Professor Martijn J. R. Heck
7. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
• MULTI-CORE processors allow for the continuous scaling of Moore’s Law.
• Because they overcome bottlenecks of heat dissipation and data synchronization.
• Currently: 80 cores are at the high end. It is expected that the number of cores will
increase to around one hundred in the next few years and multiple hundreds by 2020.
• The maximum number of cores is expected to be around a thousand cores.
• Heck’s group is Developing Chip Layouts which incorporate
Optical Sources seaking further improvements in Performance
and Energy consumption.
ENG ECE: PHOTONICS GROUP
Heck & Bowers, 2014 Journal of Selected Topics in Quantum Electronics
8. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
New Chip Designs - new paradigms
• Optical approaches as by Heck et al.
• Chips are primarely 2D design and it make sence to use the third dimension
• Mimicking the Brain and its design
The impact of computers will become…
COMPUTATIONAL POWER WILL
CONTINUE TO INCREASE
9. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
Most computers will NOT be computers.
How many electric motors do you have in your home?
How many did you buy as electric motors?
I control several computers, but most computers I control
are embedded in cars, remote controls, refrigerators, etc.
We will see Smart, or apparently Intelligent Products.
Intels computer on stick, 2015
10. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DEPARTMENT OF ENGINEERING
To Educate more Outstanding Engineers on All Levels
To Perform Engineering Science Research at the Highest International Level
11. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
THE CAR AS AN EXAMPLE
TWT in Germany is one of the project’s case companies:
They helps car-manufacturers (BMW / Audi) with simulations and models of prototypes.
Peter Gorm Larsen:
A car built around the turn of the millennium, didn’t contain many computers controlling its functions.
A modern car, involves more than 100 different small computers controlling everything from the keys and brakes to
cruise control and automatic parking functionality.
It is clear that this development provides the car industry with completely new challenges in both the development and
the test stages.
What do you do when the many computers need to talk to each other?
How do you ensure that they adapt as well as possible to the context in which they will be used?
12. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
INTELLIGENCE THROUGH ALGORITHMS
Agricultural machinery of the future needs to communicate
Resources and time to be saved during harvesting
if the different field operations and harvesting
equipment can be coordinated in a more
optimal way
13. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
CONSEQUENCE
With increasing computational power, system complexity, and with improved algorithms
technology will become responsive and demonstrate more complex behaviour:
This AUDI autonomously drove across the US
14. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DEPARTMENT OF ENGINEERING
To Educate more Outstanding Engineers on All Levels
To Perform Engineering Science Research at the Highest International Level
Preben Kidmose
15. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
INTERFACING WITH THE BRAIN
Preben Kidmose builds this in-the-ear EEG device:
It is a Brain-computer interface developed with
micro-computers and advanced algorithms that can
interpret electrical activity in the brain cells and
thereby reveal some mental states
An insulin shock detector is one of the many potential applications of the technology
The device will protects diabetes patients. Small electrodes measure electrical brain
signals and warn the patient if their blood sugar drops to a critically low level
16. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DEPARTMENT OF ENGINEERING
To Educate more Outstanding Engineers on All Levels
To Perform Engineering Science Research at the Highest International Level
Rune Hylsberg Jakobson
17. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
FLEXIBLE USE OF ELECTRICITY NEEDED?
Problem: Solar and wind energy is fluctuating. Consumer demand is fluctuating
How to minimize a surplus of electricity production & reduce emission of greenhouse gases?
Develop an information system placing decision at the consumer level and balance
consumers needs and opportunities given to the consumer with actual energy production
capabilities
Use data analysis to improve understanding – learn - when, how much and in which way
energy is used in households
Develop algorithms that will informing consumers of alternative choices
18. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
CONSEQUENCE
With increasing computational power, and with the human brain in the loop
our decision process change
One impressive technology recently
demoed is Skype Translator
It provide live translation from
one language to another:
This is now, what happens with further
computational power available?
19. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
HYBRID THINKING
Ray Kurzweil elaborated in a
recent TED talk on what happens
when you combine:
› Brain machine interfaces
› Communication networks
› Cloud computing
› Increased computational power
20. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
Martijn Heck: “You only succeed if you can do it well. If you cannot do it well, basically you
have a useless technology”.
This attitude, our curriosity drives evolution. Like the chip design evolving through paradigms.
It is the Evolution Process which is Exponential, not only Moore’s law.
Miniaturization of technology is also an evolution process. The fundamental concepts driving
this is our ability to apply Materials and Control their Arrangement.
TECHNOLOGICAL DEVELOPMENT IS AN
EVOLUTION
21.
22. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
US Initiative by Barack Obama, 2011
For many years, the United States has been a dominant player in the discovery of
transformative materials that are the basis of entirely new products and industries, yet the
time lag between discovery of advanced materials and their use in commercial products
can be 20 years or more.
At the heart of MGI is the Materials Innovation Infrastructure
› Framework for seamlessly integrated Experiments, Computation, Theory
› Access to Digital Data
› A Well-Equipped Workforce
MATERIALS GENOME PROJECT (MGI)
23. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DEPARTMENT OF ENGINEERING
To Educate more Outstanding Engineers on All Levels
To Perform Engineering Science Research at the Highest International Level
Assoc. Prof. Søren P. Madsen
24. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
IMPROVING SOLAR ENERGY
PRODUCTION
Søren Madsen is building Mathematical Models of the Nano Material and its Layout,
solving the Electro Magnetic field equations to get an idea of Electron Transport in the
Titanium dioxide layer as Sun Light hits the Surface
The Material Development Cycle is reduced as a
high number of variation of Silver nanoparticle
shapes and layout is virtually tested prior to
production in the iNANO clean room
We need computational power to test more
advanced material systems
Johannsen et al. 2015 in Applied Physics Letters
25. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
… and you get a self powered camera:
The pixel’s photodiode can be used to not only
measure the incident light level, but also
convert the incident light into electrical energy
MERGING SOLAR ENERGY PRODUCTION
WITH INTEGRATED ELECTRONICS
Nayar, S.K., Sims, D.C. & Fridberg, M., 2015 Towards Self-Powered Cameras.
Proceeding of the International Conference on Computational Photography ICCP.
26. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DEPARTMENT OF ENGINEERING
To Educate more Outstanding Engineers on All Levels
To Perform Engineering Science Research at the Highest International Level
27. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
Microstructure from 3D print:
• Shielding of the nanostructure because of the large differences in
stiffness needed for transduction and native tissue
• Anatomical correct shapes to fit the site of implant
Nano: Micro:Same material
Combined into a single component
Nanostructure:
• Bioactivity induced mechanically
and chemically.
BIOMATERIAL GUIDE CELLS TO BUILD TISSUE
30. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
Functionalized Scaffolds for Tissue Engineering
We have established methods to build 3D scaffolds in a manner that enables us to synthesize
designs which include functions within the 3D scaffold at all length scales from macro to nano:
Advanced functionalization is needed if a scaffold is to succeed with correct tissue regeneration. Regenerated tissue must
include inherent functions realized in native tissue. Thus, an integrated design approach is needed.
Anatomical correct
shapes to fit the site of
implant made by the
surgeon.
Molecular surface modifications
to improve cellular signaling
and / or nanostructural interaction.
Artificial vascularisation
to improve perfusion
and cultivation of thick
tissues.
Use of composites to
tailor scaffold morphology
and improve host integration.
mm μm nm
Particles
Fibers Platelet
Porous
31. 3D Characterization
Nygaard et al. in
Biotech. & Bioeng.
The history of LevOss – a journey of innovation
AcademicCommercial
2006 20132012 2014 2015 2020
Filing of IP
Three-dimentional
Nanostructured Hybrid
Scaffold and
Manufacture Thereof.
Patent nr.
PCT/DK2010/050167
Spun out from
University
Jan 2012 with Seed
investment from
SEED capital
2010
HTF Grant
2 M Euro
GMP partner and in
vivo models
Closure of next
Seed Round
Invitation for X M Euro
Building of syndicate
Large animal model
Mini pig study
Weak translation of rabbit data
in larger animals
Multiple tissues
Biomarkers by gene
expression
Scaffold
Production
Chen et al. in
Acta Biomaterialia
MQUL Collaboration
Drug delivery system for OA
MIT Collaboration
Microtia tissue engineering
New SAB
Prof Mats Brittberg
Prof Verdonk
Q3 2015
Mechanotransduction
hypothesis – cartilage ?
Biomarkers by gene expression
AU Thesis
Mechanics of Stem Cells
Filing new IP
Injectable
PA/2013 70759
Bone Repair
Jensen et al. in
J. Biomed. Mater. Res.
BREAK THROUGH
PAPER
ENGLER ET AL. 2006 IN CELL
MATRIX ELASTICITY DIRECTS STEM
CELL LINEAGE SPECIFICATION
Nanoporous
Deformations
Nygaard et al. on
Invited Biomechanics Workshop
AU Thesis
Cartilage signaling
siRNA to Stem
Cells
Andersen et al. in
Molecular Therapy
Vascularization
Wittenborn et al. in
J Magn Reson Imaging
Nanostructured Scaffold
Bjerre et al. on
ORS 55th Annual Meeting
Cartilage Repair
Christensen et al. in
Knee Surg. Sports Trauma. Arthroscopy
B.Sc Course Design of Medical Devices
M.Sc Course Tissue Engineering (first in Denmark)
3D Print with RNA
Andersen et al. in
Adv. Func. Materials
32. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DEPARTMENT OF ENGINEERING
To Educate more Outstanding Engineers on All Levels
To Perform Engineering Science Research at the Highest International Level
33. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
FOCUSING ON A SINGLE CELL
Peter Kristensen is able to find the few
cancer cells between billions of healthy
cells in a blood sample
The key is a small chip that is able to bind the
cancer cells on its surface using antibodies
Through selection (image processing) that single cell of interest is
identified and the surroundings are destroyed with UV light
Access to single cells will enhance our understanding of disease
progression and how cell functions and interact
34. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DEPARTMENT OF ENGINEERING
To Educate more Outstanding Engineers on All Levels
To Perform Engineering Science Research at the Highest International Level
Associate Professor Xuping
Zhang
35. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
ROBOTIC SINGLE-CELL HANDLING
Xuping Zhang builds a robot
that works within a microscope
With it, he can handle cells:
pick-and-place, positioning,
rotating, sorting, patterning,
identification, inspection,
isolation, injection, etc.
Specifically he works on an intra cytoplasmic sperm injection (ICSI) system.
36. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DRUG DELIVERY
NANOCOMPOSITE DELIVER SIRNA TO STEM CELL AND CONTROL THEIR
DIFFERENTIATION
Andersen et al. 2010 Molecular Therapy 18 (11) 2018-2027
37. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DRUG DELIVERY
NANOCOMPOSITE DELIVER SIRNA TO STEM CELL AND CONTROL THEIR
DIFFERENTIATION
Making bone and fat within the same material – in each side, solely by
delivering two different siRNA sequences to stem cells (hMSC):
Histological sections from the mouse stained with von Kossa (mineralization
appears black) and by immunohistochemistry for osteocalcin
(osteocalcin appears brown) Andersen et al. 2010 Molecular Therapy 18 (11) 2018-2027
38. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
MATERIALS ENGINEERING
BIOMATERIALS – SELF ASSEMBLY IN NATURE
Fratzl and Weinkamer (2007)
40. 24TH OF APRIL 2015
FD
HEAD OF MECHANICAL ENGINEERING
JENS VINGE
DEPARTMENT OF ENGINEERING
AARHUS
UNIVERSITYAU
DEPARTMENT OF ENGINEERING
To Educate more Outstanding Engineers on All Levels
To Perform Engineering Science Research at the Highest International Level
Learn about us – follow the link.
Morning – thank you for taking your time to show up.
My opportunity to let you know about a recent perspective on Technology and let become a bit more familiar with My Department of Engineering
I have taken the liberty to alter the language into English
We came to this organisation from international standards (Body of Knowledge) on how the Engineering Disciplines is established through history
It is combined with the fact coming from that it is people in our organisation who actually do the science.
The current computational power
The computational power to come
Passing the computational power of the brain in a mouse 2020 and humans some 2050 ?
Picture of the Intel Pentium chip
Can we continue to improve on this kind of technology, the densification of transistors?
We agree on that the computational power in computers will continue to grow as predicted.
And it will transcent the capability of the human brain to do calculations
Intel has rather impressively crammed in a quad-core Atom CPU, 32GB of storage and 2GB of RAM, along with a USB port, WiFi and Bluetooth 4.0 support and a mini-USB connector for power. It runs Windows 8.1.
What runs the in the computer is code, software. The smartness comes from the algorithms – the code.
The enabling step in building smart / intelligent products is profound integration of hardware and software.
http://www.telegraph.co.uk/finance/newsbysector/industry/engineering/11495706/Driverless-cars-to-create-320000-UK-jobs-and-save-2500-lives.html
Driverless cars to create 320,000 UK jobs and save 2,500 lives Britain will get huge economic boost from driverless cars which will also cut death toll on roads, new report predict.
Why don’t we see this AUDI on the streets? Because technology is being brought to marked along with comsumer adaptation and willingness to pay for features.
Self driving cars tend to make pasengers sick (12%)
One of the most complex systems that we know of is our brain.
Accumulating information on Energy comsumption and guide consumers in their behaviour.
we create a better balance between our electricity consumption and our electricity production from renewable energy sources.”
Live translation and live information support will change negotiations, lower barriers and improve conflict handling.
The transition towards hybrid thinking starts with the Google Glass and Apple Watch and once these technologies is combined with our knowledge on Medical Devices we will get to the hybrid thinking as presented here.
It will be great as a scientist in 2040 when my brain together with my exponential growing non-biological cloud brain crack problems.
Materials Genome InitiativeNational Science and Technology Council Committee on Technology Subcommittee on the Materials Genome Initiative
DECEMBER 2014
Here in Europe we do the same; we call it something else less coordinated, but the X-Ray facility in Lund is one example.
Ti O2 = Titanium dioxide
This is an example on how Materials Engineering and System Integration lead to innovation.
The continued miniturization provide improved sensors in our microscopes letting us see smaller details leading to improved processing – its an evolution
Materials Engineering harness computation power to predict materials behaviour during processing or in its final state
3D printing of the macro structure rely on Digital Manufacturing converting 3D geometry into physical shapes
Turn through a material investigating a single cells frozen while in the process of developing bone.
XTM Image from ESRF
How long time did it take me to make this image – approximately 2,5 years.
His chip works according to a simple principle. It contains a canal with fluid through which the patient’s blood is injected. There are special antibodies on the inner surface of the canal that are able to capture the cancer cells in the sample.
Cells are like Robots. From the Genome they are program using DNA. During the process of decoding DNA in actions, that the cell take, it splises the DNA into RNA as a natural process.
You can take advantage of this proces and deliver RNA directly into the cell using nanoparticles and program it.
BCL2L2 = Osteogenic (BONE)
TRIB2 = adipose (FAT)
Euplectella :
Not only as miniturization as shown here
But also with increase in system integration (complexity)
The Michigan Micro Mote is thought as sensor for medican devices, collecting information, proces it and transmitting response using its radio layer.