Chia - Bad Request - University of Sunderland

advertisement
Neuromorphic Engineering
University of Oxford
Department of Engineering Science
Natasha Chia
1
Neuromorphic Engineering
Outline
• Definition of neuromorphic systems
• Principles of neuromorphic technology
• Typical applications of neuromorphic
technology
• The basic computational element
• Neural codes and information
representation
• Future of Neuromorphic systems
2
Definitions
Carver Mead introducec the term Neuromorphic Engineering to describe
A new field of engineering whose design principles and architecture are biologically
Inspired.
Neuro “ to do with neurons i.e. neurally inspired”
Morphic “ structure or form”
Emulates the functional structure of neurobiological systems.
3
Principles of neuromorphic
technology
Build machines that have similar perception capabilities as human perception
Adaptable and self organising
Robust to changing environments
Realisation of future “THINKING” machines
(intelligent and interactive systems)
4
What does neuromorphic
Engineering involve?
Neuromorphic
Modeling
Analyze neurophysiological
functions in order to reproduce
neuronal structures and
architectures
Neuromorphic
Computation
Build neural networks
and rely on modellers for
explanation and modelling
of natural phenomena
5
Applications of
Neuromorphic Systems
COMPUTER
SCIENCE
ELECTRICAL
ENGINEERING
NEUROMORPHIC
ENGINEERING
- Sensory systems
- Biorobots
- Neuron modelling
- Unsupervised learning
- Pattern understanding
NEUROSCIENCE
6
Neuromorphic systems
Silicon Retina
Learning and adaptation
silicon systems
Koala-obstacle/tracking robot
Silicon Cochlea
TouchPad
7
Example of Modelling approaches
Silicon Cochlear Modelling
Biological/Physical
models
1D BM model
2D hydrodynamic model
3D immersed boundary
model
Analog VLSI
models
Log domain ccts
Switched capacitor ccts
OTA biquads ccts
WLR ccts
Digital VLSI
models
FPGA digital filters
IIR/FIR
Neuromorphic
models
Parallel processing circuits
Parallel distributed analog correlation based processing
is the basis of VLSI systems that emulate the function
of neural information processing in biological systems
8
The Basic computational element :
The Neuron
9
The Neuron Model
X(nT)
W_linking
T
Y(nT)
X(nT)
W_feeding
T
Dynamic thresholding
X(nT)
W_inhibitary
T
E
dE
  E t   V t   G K t .E K  E t 
dt
10
History of neuron modelling
• 1943
• 1963
• 1970
• 1983
• 1985
• 1992
• 1993
• 1995
McCulloch & Pits
Hodgkin & Huxley modelled axon of
giant squid.
Kiang & Gerstein numerical analysis of
interactions in nerve cells
Cohan & Mpitson Deterministic chaos used
to describe behaviour of single neurons
Meyer et al Discovery of electrical
interaction between neurons
Bower Quantitative evaluation of
functional data
Knopf and Gupta Fundamental
neural processing element
Bressler Parallel processing of
information
11
Neural code and information
representation
How does spikes represent sensory information?
• Models of neuron spiking mechanism
• Stimulus Response features
• Group behaviour
12
Stimulus Response features
• Average firing rate
• Position of each
neuronal discharge
• Instantaneous firing
probability.(A ganglion’s firing
rate depends on stimulation)
13
Spiking Neuron models
• Integrate and fire
Neurons spike regularly in response
to an external current. Rate of
spiking increases with the magnitude
of the stimulus current
• Stimulus Response
Model (SRM)
At    t   he
Bx   k c e

Space
B(x)
t

x2
x2
2 rc 2
2 rs 2
 kse
Firing_R(t)
 
Time
A(t)


Firing _ Ratet   N   I x.t ' .Bx . A t  t ' dxdt'
14
Neural coding
Channel
coding
Temporal pattern
coding
Time of arrival
coding
Joint response
from multiple neurons
coding
Narrow correlations
Medium correlations
Broad correlations
t1
t2
Information is conveyed to the brain in parallel by
spike trains of the nerve fibers.
15
Group behaviour of spikes
One of the challenges has been to understand the relative
contribution of various groups of spikes
Oscillatory coupling effects and amplitude dynamics in two or
more populations of neurons will be important topics for
future research
Some other examples: STDP(Spike Time-dependent plasticity)
Group neurons with correlated inputs
16
Future of neuromorphic systems
Implantable medical electronics
Increased human computer interaction
Intelligent transportantion systems
Learning, pattern recognition
Robot control(self motion estimation)
Learning higher order perceptual computation
17
Conclusion
Electronic
Engineers
Neuroscientists
and Biologists
Reproduce
neurophysiological
phenomena in silicon
Analyze neuronal
information
Computer Scientist
and Mathematicians
Modelling of
neuromorphic
systems
18
Download