Optimisation and Robustness of Cellular
Neural Networks
Palestra para o DCA/UFRN base de sistemas inteligentes
Samuel Xavier de Souza
Katholieke Universiteit Leuven Departement Elektrotechniek (ESAT)
SCD - SISTA
Outline
1 Introduction
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
5 Conclusions and Challenges for Future Work
Outline
1 Introduction
The potential of Cellular Neural Networks
How does a CNN work? Goals of this thesis
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
CNN as an alternative to classical digital computation
Cells replication
→ →
Local connections→Large number of cells→High parallelism
Local connections & analog signals→Low area & Low power High parallelism→ supercomputer power→order of Tera ops.
[Chua and Yang, IEEE TCAS, 1988]
CNN as an alternative to classical digital computation
Cells replication
→ →
Local connections→Large number of cells→High parallelism Local connections & analog signals→Low area & Low power
High parallelism→supercomputer power→order of Tera ops.
CNN as an alternative to classical digital computation
Cells replication
→ →
Local connections→Large number of cells→High parallelism Local connections & analog signals→Low area & Low power High parallelism→ supercomputer power→order of Tera ops.
[Chua and Yang, IEEE TCAS, 1988]
CNN are very suitable for image processing
Evolution of CNN chips New Q-Eye CNN chip
176×144 cells
>10,000 fps <100mW
Early vision processing Redundant image data
→relevant information Wide application range:
Visual applications demanding high throughput speed
A wide range of applications
Ultra-high speed event/fault monitoring in production lines
Smart security cameras
Automotive applications such as
smart airbag deployment blind spot detection navigation
collision detection
3D Echocardiogram
Vision for industrial/domestic robots Object tracking
Eye tracking and gaze estimation Visual game/computer interfaces Handsfree wheelchair driving, etc.
A wide range of applications
Ultra-high speed event/fault monitoring in production lines Smart security cameras
Automotive applications such as
smart airbag deployment blind spot detection navigation
collision detection
3D Echocardiogram
Vision for industrial/domestic robots Object tracking
Eye tracking and gaze estimation Visual game/computer interfaces Handsfree wheelchair driving, etc.
A wide range of applications
Ultra-high speed event/fault monitoring in production lines Smart security cameras
Automotive applications such as
smart airbag deployment blind spot detection navigation
collision detection
3D Echocardiogram
Vision for industrial/domestic robots Object tracking
Eye tracking and gaze estimation Visual game/computer interfaces Handsfree wheelchair driving, etc.
A wide range of applications
Ultra-high speed event/fault monitoring in production lines Smart security cameras
Automotive applications such as
smart airbag deployment blind spot detection navigation
collision detection
3D Echocardiogram
Vision for industrial/domestic robots Object tracking
Eye tracking and gaze estimation Visual game/computer interfaces Handsfree wheelchair driving, etc.
A wide range of applications
Ultra-high speed event/fault monitoring in production lines Smart security cameras
Automotive applications such as
smart airbag deployment blind spot detection navigation
collision detection
3D Echocardiogram
Vision for industrial/domestic robots
Object tracking
Eye tracking and gaze estimation Visual game/computer interfaces Handsfree wheelchair driving, etc.
A wide range of applications
Ultra-high speed event/fault monitoring in production lines Smart security cameras
Automotive applications such as
smart airbag deployment blind spot detection navigation
collision detection
3D Echocardiogram
Vision for industrial/domestic robots Object tracking
Eye tracking and gaze estimation Visual game/computer interfaces Handsfree wheelchair driving, etc.
A wide range of applications
Ultra-high speed event/fault monitoring in production lines Smart security cameras
Automotive applications such as
smart airbag deployment blind spot detection navigation
collision detection
3D Echocardiogram
Vision for industrial/domestic robots Object tracking
Eye tracking and gaze estimation
Visual game/computer interfaces Handsfree wheelchair driving, etc.
A wide range of applications
Ultra-high speed event/fault monitoring in production lines Smart security cameras
Automotive applications such as
smart airbag deployment blind spot detection navigation
collision detection
3D Echocardiogram
Vision for industrial/domestic robots Object tracking
Eye tracking and gaze estimation Visual game/computer interfaces
A wide range of applications
Ultra-high speed event/fault monitoring in production lines Smart security cameras
Automotive applications such as
smart airbag deployment blind spot detection navigation
collision detection
3D Echocardiogram
Vision for industrial/domestic robots Object tracking
Eye tracking and gaze estimation Visual game/computer interfaces Handsfree wheelchair driving, etc.
Outline
1 Introduction
The potential of Cellular Neural Networks
How does a CNN work?
Goals of this thesis
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
A template controls space-invariant CNNs
The equation of a cell
dxi,j dt = −xi,j+ X |k,l|6r ak,lyi−k,j−l+ X |k,l|6r bk,lui−k,j−l+z, with yi,j(t) =f(xi,j)
Atemplatespecifies acoupling law
19 template values(9+9+1)
Controls the entire network
CNN computer Template Digital computer Instruction CNN template values A= 2 4 a−1,−1 a−1,0 a−1,1 a0,−1 a0,0 a0,1 a1,−1 a1,0 a1,1 3 5 ,B= 2 4 b−1,−1 b−1,0 b−1,1 b0,−1 b0,0 b0,1 b1,−1 b1,0 b1,1 3 5 ,z 8 / 54
A template controls space-invariant CNNs
The equation of a cell
dxi,j dt = −xi,j+ X |k,l|6r ak,lyi−k,j−l+ X |k,l|6r bk,lui−k,j−l+z, with yi,j(t) =f(xi,j)
Atemplatespecifies acoupling law
19 template values(9+9+1)
Controls the entire network
CNN computer Template Digital computer Instruction CNN template values A= 2 4 a−1,−1 a−1,0 a−1,1 a0,−1 a0,0 a0,1 a1,−1 a1,0 a1,1 3 5 ,B= 2 4 b−1,−1 b−1,0 b−1,1 b0,−1 b0,0 b0,1 b1,−1 b1,0 b1,1 3 5 ,z
A template controls space-invariant CNNs
The equation of a cell
dxi,j dt = −xi,j+ X |k,l|6r ak,lyi−k,j−l+ X |k,l|6r bk,lui−k,j−l+z, with yi,j(t) =f(xi,j)
Atemplatespecifies acoupling law
19 template values(9+9+1)
Controls the entire network
CNN computer Template Digital computer Instruction CNN template values A= 2 4 a−1,−1 a−1,0 a−1,1 a0,−1 a0,0 a0,1 a1,−1 a1,0 a1,1 3 5 ,B= 2 4 b−1,−1 b−1,0 b−1,1 b0,−1 b0,0 b0,1 b1,−1 b1,0 b1,1 3 5 ,z 8 / 54
A template controls space-invariant CNNs
The equation of a cell
dxi,j dt = −xi,j+ X |k,l|6r ak,lyi−k,j−l+ X |k,l|6r bk,lui−k,j−l+z, with yi,j(t) =f(xi,j)
Atemplatespecifies acoupling law
19 template values(9+9+1)
Controls the entire network
CNN computer Template Digital computer Instruction CNN template values A= 2 4 a−1,−1 a−1,0 a−1,1 a0,−1 a0,0 a0,1 a1,−1 a1,0 a1,1 3 5 ,B= 2 4 b−1,−1 b−1,0 b−1,1 b0,−1 b0,0 b0,1 b1,−1 b1,0 b1,1 3 5 ,z
A template controls space-invariant CNNs
The equation of a cell
dxi,j dt = −xi,j+ X |k,l|6r ak,lyi−k,j−l+ X |k,l|6r bk,lui−k,j−l+z, with yi,j(t) =f(xi,j)
Atemplatespecifies acoupling law
19 template values(9+9+1)
Controls the entire network
CNN computer Template Digital computer Instruction CNN template values A= 2 4 a−1,−1 a−1,0 a−1,1 a0,−1 a0,0 a0,1 a1,−1 a1,0 a1,1 3 5 ,B= 2 4 b−1,−1 b−1,0 b−1,1 b0,−1 b0,0 b0,1 b1,−1 b1,0 b1,1 3 5 ,z 8 / 54
A programmable CNN is ideal for image processing
Templates provide an ideal framework for Single-Instruction Multiple-Data computation Image processing operations are mostly SIMD
Most of these tasks have a corresponding CNN template CNN-UM allows execution of a sequence of templates
[Roska and Chua, IEEE TCAS, 1993]
One-to-one cell-pixel mapping Advantages
high parallelism much faster less silicon area very low power Theonlydrawback isrobustness
CNN chips do not behave as ideal CNNs
Example (d) (b) (a) (c) Causes of errors Parameter spread Electrical noiseImperfect loading from off-chip to on-chip memory Temperature variation
Chip-independent approaches
Design of robust templates Template decomposition Stochastic template optimisation
Outline
1 Introduction
The potential of Cellular Neural Networks How does a CNN work?
Goals of this thesis
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
Methodologies for robustness in CNN chips
Goals
Robust tuning of CNN chips Learning of complex behaviour in CNNs
Optimisation techniques for robust CNN tuning and learning Integrate the methodologies Develop proof-of-concept applications
Outline
1 Introduction
2 Robust CNN Tuning and Learning
Chip-specific Methods
Learning complex dynamics Experiments and Results
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
Optimising with chip measurements works better
Chip-specific
[F ¨oldesy et al., IEEE TCAS-I,1999]
Optimise-decompose-optimise approach
Uses chip measurements to calculate error Searches the minimum cost via gradient search
Non-optimal templates are decomposed
Disadvantages
Needs explicit calculation of error gradient Only optimises B and z: uncoupled templates May cause unnecessary template
decomposition
The use of chip-measurements does improve robustness
Training set is generated using an ideal CNN simulator
[Xavier-de-Souza et al., 2003]
Generating a training set triplet Training set triplet: input, initial state, and desired images Original template
pinit
Ideal CNN simulator
A good starting point
Original template pinit is a good initial approximation
Presenting our tuning setup
[Xavier-de-Souza et al., IEEE TCAS-I, 2004]
Chip-specific CNN tuning setup Optimisation core
A global optimisation method Generates and accepts, or not, a new template p Optimisation problem min p g(p), where g(p) = 1 √ k v u u t k X i=1 (yid−yi(∞))2 16 / 54
Major advantages, a minor drawback
Advantages of this tuning setup
No restrictions related to template type Propagating templates can be tuned too
Poor local solutions are avoided due to global optimisation No premature template decomposition
The drawback:global optimisation is slower
What makes it worth?
Number of unknowns619 compensates this drawback Original template is a good starting point
Optimising for robustness
Typical template chip error Chip-specific robustness
Gross tuning optimum
chip-robust optimum
Max. value Min value pmin pinit pmax
pmin pmax
pinit e
upper bound Theoretical
lower bound Theoretical
Optimisation problem min p 1 r r X j=1 g(p+ej) 18 / 54
Outline
1 Introduction
2 Robust CNN Tuning and Learning Chip-specific Methods
Learning complex dynamics
Experiments and Results
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
Learning is more difficult than tuning
[Xavier-de-Souza et al., 2005]
No know good starting point
Desired output is obtained either manually or via other methodologies Potentially, the entire template parameter space needs to be searched
Helpful strategies
The search space can be reduced with simple initial assumptions, e.g. symmetry, zero values
These assumptions can be relaxed in subsequent optimisation runs
Learning is more difficult than tuning
[Xavier-de-Souza et al., 2005]
No know good starting point
Desired output is obtained either manually or via other methodologies Potentially, the entire template parameter space needs to be searched
Helpful strategies
The search space can be reduced with simple initial assumptions, e.g. symmetry, zero values
These assumptions can be relaxed in subsequent optimisation runs
Learning dynamics requires more desired outputs
Desired output is a sequence of images Example
Example
Example
[Xavier-de-Souza et al., JCTA, 2006]
NewActive Wave Computingoperations Example s p ir a l a u to w a v e 21 / 54
Learning dynamics requires more desired outputs
Desired output is a sequence of images Example
Example
Example
[Xavier-de-Souza et al., JCTA, 2006]
NewActive Wave Computingoperations Example s p ir a l a u to w a v e
Outline
1 Introduction
2 Robust CNN Tuning and Learning Chip-specific Methods
Learning complex dynamics
Experiments and Results
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
5 Conclusions and Challenges for Future Work
Tuning results for ACE4k chip (64
×
64 cells)
Edge detection
Average
Half toning
Tuning results for ACE16k chip (128
×
128 cells)
input image initial state ideal CNN output orig. tem. output final tem. output
Average
Threshold
Sobel
On-chip learning results for ACE4k (64
×
64 cells)
Travelling wave
Combustion wave
Pyramiding up
Simulation learning of a spiral autowave (16
×
16 cells)
D e s ir e d D e s ir e d 1 s t la y e r L e a rn e d G e n e ra lis e d 1 s t la y e r D e s ir e d D e s ir e d 2 n d la y e r L e a rn e d G e n e ra lis e d 2 n d la y e r 26 / 54Outline
1 Introduction
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
Simulated annealing and cooperative behaviour
General principles of Coupled SA Variance control of acceptances Experiments and results
4 Applying CSA to CNN Optimisation
SA is based on two stochastic processes
Temperature Tk Temperature Tac
k Current Cost E(x) Current Solution x Program flow Data flow No Yes No Stopping Criterion Equilibrium Criterion Acceptance x ← y if E(y) ≤ E(x), or Generation yi= f (xi, ε) Temperature Scheduling Tk+1= U(Tk, k) Tac k+1= V (T ac k, k) Yes END Initialization x= x0 Tk= T0 Tac k= T0ac k= 1 assess E(x) ∀i = 1, · · · , D; ε ← gk(Tk)
if r < A(x → y), r ← U[0, 1]
Simulating physical annealing
Physical process:
heat, hold and cool slowly down Mathematical modelling:
Material states → Problem solutions Energy of a state → Cost of a solution
Temperature → Control parameter Convergence proof exists fort→ ∞
Many speed up versions exist: quality of solution↔speed With distributed versions: quality↔speed↔resources
Why a new Simulated Annealing (SA) method?
Example
Adaptive SA: less suitable for CNN tuning
Too many run-modify-rerun loops Problem representing template solutions in Genetic Algorithms
The coupling idea
Coupled Local Minimisers couples local gradient searching processes
Cooperative behaviour in SA
Parallel SA versions
Concurrent processes move independently Best solution is exchanged among process
Ensemble SA
No exchange of best solutions
Entire ensemble moves according to average cost
[Xavier-de-Souza et al., 2006a]
The new Coupled Simulated Annealing class of methods
Coupledacceptance probabilities for SA Efficient parallel/distributed implementation Less sensitive to initialisation parameters
Outline
1 Introduction
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
Simulated annealing and cooperative behaviour
General principles of Coupled SA
Variance control of acceptances Experiments and results
4 Applying CSA to CNN Optimisation
Comparing SA and Coupled SA
Properties of the CSA class of methods
.. . .. . AΘ y1 y2 yi ym x2 Ω SA CSA x y A xi x1 Θ xm E E E E f γ 0 ≤ A(x → y) ≤ 1,
∀ x, y ∈ Ω. Θ ⊂ Ω, and γ = f [E(x1), E(x2), · · · , E(xm)] .
0 ≤ AΘ(γ, xi→ yi) ≤ 1, ∀ x ∈ Θ, y ∈ Ω, with Ω AΘ AΘ AΘ 32 / 54
Three CSA instance methods
Method 1 Multi-state SA AΘ(γ,xi →yi) = exp −E(yi) Tac k exp −E(yi) Tkac + exp −E(xi) TkacThree CSA instance methods
Method 1 Multi-state SA AΘ(γ,xi →yi) = exp −E(yi) Tac k exp −E(yi) Tac k + X xj∈Θ exp −E(xj) Tac kGeneralises classical SA acceptance forΘ ≡x1
Three CSA instance methods
Method 1 Multi-state SA AΘ(γ,xi →yi) = exp −E(yi) Tkac exp −E(yi) Tkac + X xj∈Θ exp −E(xj) Tkac | {z } γThree CSA instance methods
Method 2 Blind acceptance AΘ(γ,xi →yi) = 1 − exp −E(xi) Tkac X xj∈Θ exp −E(xj) TkacAcceptance of yi is independent of its energy when
E(xi) <E(yi)
Three CSA instance methods
Method 2 Blind acceptance AΘ(γ,xi →yi) = 1 − exp −E(xi) Tac k X xj∈Θ exp −E(xj) Tac k | {z } γAcceptance of yi is independent of its energy when
Three CSA instance methods
Method 3 Coupled SA-Modified AΘ(γ,xi →yi) = exp E(xi) Tac k X xj∈Θ exp E (xj) Tac kAlso perform blind acceptance
Strategy: probability ofanycurrent stateleave
to a new state sums to 1
Three CSA instance methods
Method 3 Coupled SA-Modified AΘ(γ,xi →yi) = exp E(xi) Tkac X xj∈Θ exp E(x j) Tkac | {z } γAlso perform blind acceptance
Strategy: probability ofanycurrent stateleave
Three CSA instance methods
Method 3 Coupled SA-Modified AΘ(γ,xi →yi) = exp E(xi)−max z∈Θ(E(z)) Tkac X xj∈Θ exp E(xj)−max z∈Θ(E(z)) Tkac | {z } γAlso perform blind acceptance
Strategy: probability ofanycurrent stateleave
to a new state sums to 1
Outline
1 Introduction
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
Simulated annealing and cooperative behaviour General principles of Coupled SA
Variance control of acceptances
Experiments and results
4 Applying CSA to CNN Optimisation
Adapting the schedule of the acceptance temperature
Acceptance variance in CSA-M
06 σ2(AΘ) 6
m−1
m2
Simple way to control the variance ifσ2(AΘ) < σ2D, Tkac=Tkac−1(1− α) ifσ2(AΘ) > σ2D, Tkac=Tkac−1(1+ α) Reduces sensitivity to initial parameters Adaptive schedule for Tkac Experimentally: Best values σD2 ≈ 99%of mm−21 α ≈ 0.05 37 / 54
Outline
1 Introduction
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
Simulated annealing and cooperative behaviour General principles of Coupled SA
Variance control of acceptances
Experiments and results
4 Applying CSA to CNN Optimisation
Presenting the test problems
Test problems group 1
Unimodal and simple multi-modal problems
Test problems group 2
6 multidimensional and multi-modal problems Many local minima
Test problems group 3
6 rotated problems from group 2 Difficult for multiple 1-D searches
Example −3 −2 −1 0 1 2 3 −3 −2 −1 0 1 2 3 0 1000 2000 3000 4000 Example −500 0 500 −500 0 500 −1000 −500 0 500 1000 39 / 54
Presenting the test problems
Test problems group 1
Unimodal and simple multi-modal problems
Test problems group 2
6 multidimensional and multi-modal problems Many local minima
Test problems group 3
6 rotated problems from group 2 Difficult for multiple 1-D searches
Example −3 −2 −1 0 1 2 3 −3 −2 −1 0 1 2 3 0 1000 2000 3000 4000 Example −500 0 500 −500 0 500 −1000 −500 0 500 1000
Presenting the test problems
Test problems group 1
Unimodal and simple multi-modal problems
Test problems group 2
6 multidimensional and multi-modal problems Many local minima
Test problems group 3
6 rotated problems from group 2 Difficult for multiple 1-D searches
Example −3 −2 −1 0 1 2 3 −3 −2 −1 0 1 2 3 0 1000 2000 3000 4000 Example −500 0 500 −500 0 500 −1000 −500 0 500 1000 39 / 54
Comparison experiments
Multi-start SA versus CSA-M with variance control, D=30
sa csa 0 2 4 6 8 10 12 14 16 18 20 f3, D=30 sa csa 0 10 20 30 40 50 60 70 80 f4, D=30 sa csa 0 10 20 30 40 50 f5, D=30 sa csa 0 20 40 60 80 100 120 140 160 180 f6, D=30 sa csa 0 50 100 150 f7, D=30 sa csa 0 10 20 30 40 50 60 70 80 90 f10, D=30 sa csa 50 60 70 80 90 100 110 120 130 f12, D=30 sa csa 30 40 50 60 70 80 f13, D=30
Comparable performancewithoutvariance control When testing sensitivity against initialisation parameters Clear CSA-Msuperiority with variance control
Other experiments
Effect of variance control
5 10 15 20 25 30 10−5 10−4 10−3 10−2 10−1 100 101 102 Cost Number of optimizers CSAM (a) CSAM (b) CSAM w/ Var. Control (c)
Variance control approximates best average run
Scaling with dimension
4 6 8 10 12 14 16 18 20 101 102 103 104 105 Number of Optimizers
Number of Cost Function Evaluations per Optimizer
D=20 D=16 D=12
D=8
D=4
Final cost improves consistently with m Scales with higher dimensions
Outline
1 Introduction
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
General approach to CNN optimisation
A systematic approach Applications
CSA for tuning and learning
Cell model dxi,j dt = −ˆg(xi,j,ΦSig) +z(ΦTem) + X |k,l|6r ak,l(ΦTem)xi−k,j−l + X |k,l|6rbk,j(ΦTem)ui−k,j−l(ΦOpt,ΦSig)
Hardware parameters
Φ = [ΦSig,ΦTem,ΦOpt]
xi,j
ˆ
g(xi,j)
gmax(ΦSig)
gmin(ΦSig)
The optimisation problem within the unified framework min A,B,z,Φ,∆t1,···,∆tT 1 NSNRNT NS X s=1 NR X n=1 NT X k=1 X i,j (xid ,j;s,k,n−xi,j(A,B,z,Φ,tk)) 2 43 / 54
Outline
1 Introduction
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
General approach to CNN optimisation
A systematic approach
Applications
Unified CNN optimisation framework
Block diagram [Hillier et al., 2006]
CNN optimisation cases Chip-specific template tuning Learning of fixed-point templates Learning of spatio-temporal dynamics Modular, replicable to many CSA processes CSA reduces
run-modify-rerun loops
Guidelines for defining a training set
Example
Include all essential input-output mapping functionalities
Consider also bias and fixed state images
Balance dark and light pixels Embed many functionalities in a single TS instance
Consider exchanging input and initial state
Ensure continuity of dynamics
CNNOPT: Matlab toolbox
Outline
1 Introduction
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
General approach to CNN optimisation A systematic approach
Applications
5 Conclusions and Challenges for Future Work
Real-time object tracking
Tracking the object
Mi Ai Oi Fi None None None i = 0 i = 1 i = 2 i = 3 [Xavier-de-Souza et al., 2004]
Building blockfor other applications e.g. computer/game interfaces, eye tracking, face tracking, gaze estimation, target tracking, multiple object tracking.
More than300 frames/sec. in 2nd generation CNN chips
Handsfree wheelchair driving
Shifting chip window
Face tracking no yes no no yes yes current frame all objects object positions Infer unknown Initialization lost?
Isolate objects in Acquire new frame if necessary Resize objects object lost? Shift tracking window resize object? Adaptive resizing 3-D Testing environment driving action driving action Bi-I Matlab Computer Video 1 Video 2 [Xavier-de-Souza et al., 2006b] 49 / 54
Outline
1 Introduction
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
5 Conclusions and Challenges for Future Work
Conclusions
Conclusions
Conclusions
Conclusions
Conclusions
Conclusions
Conclusions
Conclusions
Outline
1 Introduction
2 Robust CNN Tuning and Learning
3 Coupled Simulated Annealing
4 Applying CSA to CNN Optimisation
5 Conclusions and Challenges for Future Work Conclusions
”To be continued...”
New template decomposition techniques Multi-layer emulation in single-layer CNN chips General tuning of hardware parameters
On-chip tuning of CNN algorithms Self-tuning of CNN chips
Promoters
Johan Suykens, Joos Vandewalle
Advisory committee
Andr ´e Barb ´e, Marc Van Hulle, Jan Van Impe
Co-authors
Desir ´e Boll ´e, D ´aniel Hillier,
Michiel Van Dyck, M ¨us¸tak E. Yalc¸ın
Prof. T ´amas Roska
for the opportunity to visit his lab in Hungary and for the honor to have him in my Ph.D. jury
SISTA colleagues