JAISTサマースクール2016「脳を知るための理論」講義04 Neural Networks and Neuroscience

SS2016 Modern Neural
Computation
Lecture 5: Neural Networks
and Neuroscience
Hirokazu Tanaka
School of Information Science
Japan Institute of Science and Technology
Supervised learning as functional approximation.
In this lecture we will learn:
• Single-layer neural networks
Perceptron and the perceptron theorem.
Cerebellum as a perceptron.
• Multi-layer feedforward neural networks
Universal functional approximations, Back-propagation
algorithms
• Recurrent neural networks
Back-propagation-through-time (BPTT) algorithms
• Tempotron
Spike-based perceptron
Gradient-descent learning for optimization.
• Classification problem: to output discrete labels.
For a binary classification (i.e., 0 or 1), a cross-entropy is
often used.
• Regression problem: to output continuous values.
Sum of squared errors is often used.
Cost function: classification and regression.
• Classification problem: to output discrete labels.
For a binary classification (i.e., 0 or 1), a cross-entropy is
often used.
• Regression problem: to output continuous values.
Sum of squared errors is often used.
ˆ:output of network, :desired outputi iy y
( ) ( ) ( )
ˆ1ˆ
: samples: samples
ˆ ˆlog 1 log 1 log 1ii
i i i i i
yy
i
ii
y y y y y y
−
− − =− + − −  ∑∏
( )
: sa p e
2
m l s
ˆi
i
iy y−∑
Perceptron: single-layer neural network.
• Assume a single-layer neural network with an input layer
composed of N units and an output layer composed of
one unit.
• Input units are specified by
and an output unit are determined by
( )1
T
Nx x=x 
( )T
0
1
0
n
i
i iy f w x fw w
=
 
= + = + 
 
∑ w x
( )
1 if 0
0 if 0
u
f
u
u
≥
= 
<
Perceptron: single-layer neural network.
feature 1
feature 2
Perceptron: single-layer neural network.
• [Remark] Instead of using
often, an augmented input vector
are used. Then,
( )1
T
Nx x=x 
( ) ( )T T
0y f w f= + =w x w x
( )11
T
Nx x=x 
( )10
T
Nw w w=w 
Perceptron Learning Algorithm.
( ) ( ) ( ){ }21 1 2, , ,, , ,P Pd d dx x x
• Given a training set:
• Perceptron learning rule:
( )i i iydη −∆ =w x
while err>1e-4 && count<10
y = sign(w'*X)';
wnew = w + X*(d-y)/P;
wnew = wnew/norm(wnew);
count = count+1;
err = norm(w-wnew)/norm(w)
w = wnew;
end
Perceptron Learning Algorithm.
Case 1: Linearly separable case
Perceptron Learning Algorithm.
Case 2: Linearly non-separable case
Perceptron’s capacity: Cover’s Counting Theorem.
• Question: Suppose that there are P vectors in N-
dimensional Euclidean space.
There are 2P possible patterns of two classes. How many
of them are linearly separable?
[Remark] They are assumed to be in general position.
• Answer: Cover’s Counting Theorem.
{ }1, ,, N
P i ∈x x x 
( )
1
0
1
, 2
N
k
P
C P N
k
−
=
− 
=  
 
∑
Perceptron’s capacity: Cover’s Counting Theorem.
• Cover’s Counting Theorem.
• Case 𝑃𝑃 ≤ 𝑁𝑁:
• Case 𝑃𝑃 = 2𝑁𝑁:
• Case 𝑃𝑃 ≫ 𝑁𝑁:
( )
1
0
1
, 2
N
k
P
C P N
k
−
=
− 
=  
 
∑
( ), 2P
C P N =
( ) 1
, 2P
C P N −
=
( ), N
C P N AP≈
Cover (1965) IEEE Information; Sompolinsky (2013) MIT lecture note
Perceptron’s capacity: Cover’s Counting Theorem.
• Case for large P:
Orhan (2014) “Cover’s Function Counting Theorem”
( ) 1 2
1 e
2
rf
,
2 2P
pC P
N
N
p
  
+ −   
   
≈
Cerebellum as a Perceptron.
Llinas (1974) Scientific American
Cerebellum as a Perceptron.
• Cerebellar cortex has a feedforward structure:
mossy fibers -> granule cells -> parallel fibers -> Purkinje
cells
Ito (1984) “Cerebellum and Neural Control”
Cerebellum as a Perceptron (or its extensions)
• Perceptron model
Marr (1969): Long-term potentiation (LTP) learning.
Albus (1971): Long-term depression (LTD) learning.
• Adaptive filter theory
Fujita (1982): Reverberation among granule and Golgi
cells for generating temporal templates.
• Liquid-state machine model
Yamazaki and Tanaka (2007):
Perceptron: a new perspective.
• Evaluation of memory capacity of a Purkinje cell using
perceptron methods (the Gardner limit).
Brunel, N., Hakim, V., Isope, P., Nadal, J. P., & Barbour, B. (2004). Optimal
information storage and the distribution of synaptic weights: perceptron versus
Purkinje cell. Neuron, 43(5), 745-757.
• Estimation of dimensions of neural representations
during visual memory task in the prefrontal cortex using
perceptron methods (Cover’s counting theorem).
Rigotti, M., Barak, O., Warden, M. R., Wang, X. J., Daw, N. D., Miller, E. K., & Fusi,
S. (2013). The importance of mixed selectivity in complex cognitive tasks.
Nature, 497(7451), 585-590.
Limitation of Perceptron.
• Only linearly separable input-output sets can be learned.
• Non-linear sets, even a simple one like XOR, CANNOT be
learned.
Multilayer neural network: feedforward design
( )n
ix
( )1n
jx −
Layer 1 Layer n-1 Layer n Layer N
( )1n
ijw
−
• Feedforward network: a unit in layer n receives inputs
from layer n-1 and projects to layer n+1.
Multilayer neural network: feedforward design
( )n
ix
( )1n
jx −
Layer 1 Layer n-1 Layer n Layer N
( )1n
ijw
−
• Feedforward network: a unit in layer n receives inputs
from layer n-1 and projects to layer n+1.
Multilayer neural network: forward propagation.
( ) ( )
( ) ( ) ( )1 1
1
n n n n
i i ij j
j
x f u f w x− −
=
 
= =  
 
∑
( )
1
1 u
f u
e−
=
+
( )
( )
( ) ( )( )2
1 1
1
1
1
11
u
u uu
f
e
e e
u
e
f u f u
−
− −−
 
= = − = 
+ + +
′ −
Layer n-1 Layer n
( )n
ix
( )1n
jx
−
( )1n
ijw
−
( ) ( ) ( )1 1
1
n n n
i ij j
j
u w x
− −
=
= ∑
In a feedforward multilayer neural network propagates its activities
from one layer to another in one direction:
Inputs to neurons in layer n are a
summation of activities of neurons in
layer n-1:
The function f is called an activation function, and its derivative is
easy to compute:
Multilayer neural network: error backpropagation
• Define an cost function as a squared sum of errors in
output units:
Gradients of cost function with respect to weights:
( )
( ) ( )
( )
2 21 1
2 2
N N
i i i
i i
x z= − = ∆∑ ∑
Layer n-1 Layer n
( ) ( ) ( ) ( )
( ) ( )1 1
1
n n n n n
i j j j ji
j
x x w
− −
∆ = ∆ −∑
( )1n
j
−
∆
( )n
i∆
The neurons in the output layer has
explicit supervised errors (the difference
between the network outputs and the
desired outputs). How, then, to compute
the supervising signals for neurons in
intermediate layers?
Multilayer neural network: error backpropagation
1. Compute activations of units in all layers.
2. Compute errors in the output units, .
3. “Back-propagate” the errors to lower layers using
4. Update the weights
( )
{ } ( )
{ } ( )
{ }1
,, , ,
n N
i i ix x x 
( )
{ }N
i∆
( ) ( ) ( ) ( )
( ) ( )1 1
1n n n n n
i j j j ji
j
x x w
− −
∆ = ∆ −∑
( ) ( ) ( ) ( )
( ) ( )1 1 1
1
n n n n n
ij i i i jw x x xη + + +
∆ =− ∆ −
Multilayer neural network as universal machine for
functional approximation.
A multilayer neural network is in principle able to approximate any
functional relationship between inputs and outputs at any desired
accuracy (Funahashi, 1988).
Intuition: A sum or a difference of two sigmoid functions is a “bump-
like” function. And, a sufficiently large number of bump functions
can approximate any function.
NETtalk: A parallel network that learns to read aloud.
Sejnowski & Rosenberg (1987) Complex Systems
A feedforward three-layer neural network with delay lines.
NETtalk: A parallel network that learns to read aloud.
Sejnowski & Rosenberg (1987) Complex Systems; https://www.youtube.com/watch?v=gakJlr3GecE
A feedforward three-layer neural network with delay lines.
NETtalk: A parallel network that learns to read aloud.
Sejnowski & Rosenberg (1987) Complex Systems
Activations of hidden units for a same sound but different inputs
Hinton diagrams: characterizing and visualizing
connection to and from hidden units.
Hinton (1992) Sci Am
Activations of hidden units for a same sound but different inputs
Autonomous driving learning by backpropagation.
Pomerleau (1991) Neural Comput
Activations of hidden units for a same sound but different inputs
Autonomous driving learning by backpropagation.
Pomerleau (1991) Neural Comput; https://www.youtube.com/watch?v=ilP4aPDTBPE
Gradient vanishing problem: why is training a multi-layer
neural network so difficult?
Hochreiter et al. (1991)
• The back-propagation algorithm works only for neural networks of
three or four layers.
• Training neural networks with many hidden layers – called “deep
neural networks”- is notoriously difficult.
( ) ( ) ( ) ( )
( ) ( )1 1
1N N N N N
j i i i ij
i
x x w− −
∆ = ∆ −∑
( ) ( ) ( ) ( )
( ) ( )
( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( )
2 1 1 1 2
1 1 1 2
1
1 1
N N N N N
k j j j jk
j
N N N N N N N
i i i ij j j jk
j i
x x w
x x w x x w
− − − − −
− − − −
∆ = ∆ −
 
= ∆ − − 
 
∑
∑ ∑
( )
( ) ( ) ( ) ( )( 1) ( 1) ( 1) ( 1) ( ) ( )
~ 1 1 1
n Nn n N N N N
x x x x x x+ + − −
∆ − × × − × − ×∆
Multilayer neural network: recurrent connections
• A feedforward neural network can represent an
instantaneous relationship between inputs and outputs
- memoryless: it depends on current inputs but not on
previous inputs.
• In order to describe a history, a neural network should
have its own dynamics.
• One way to incorporate dynamics into a neural network
is to introduce recurrent connections between units.
Working memory in the parietal cortex.
• A feedforward neural network can represent an
instantaneous relationship between inputs and outputs
- memoryless: it depends on current inputs x(t) but not
on previous inputs x(t-1), x(t-2), ...
• In order to describe a history, a neural network should
have its own dynamics.
• One way to incorporate dynamics into a neural network
is to introduce recurrent connections between units.
Multilayer neural network: recurrent connections
( ) ( )( ) ( ) ( )( )( )1 1ii i
x t f u t f t t+= += +Wx Ua
( ) ( )( )iz t g t= Vx
Recurrent dynamics of neural network:
Output readout:
a x z
U VW
Temporal unfolding: backpropagation through time (BPTT)
1t−a
1t−x tztx
{ }10 2 1,, , ,, ,t T −a a a aa  
{ }1 2 3, , , ,, ,t Tzz z zz  
,U W V
Training set for a recurrent network:
Input series:
Output series:
Optimize the weight matrices so as to approximate the training set:
Temporal unfolding: backpropagation through time (BPTT)
0a 1z1x,U W V
0a
2z1x,U W
V,U W
1a 2x
0a
3z
1x,U W
V
,U W
1a 3x2x
,U W
2a
1t−a
1t−x tztx,U W V
Working-memory related activity in parietal cortex.
Gnadt & Andersen (1988) Exp Brain Res
Temporal unfolding: backpropagation through time (BPTT)
Zipser (1991) Neural Comput
Temporal unfolding: backpropagation through time (BPTT)
Zipser (1991) Neural Comput
Model
Experiment
Model
Experiment
Spike pattern discrimination in humans.
Johansson & Birznieks (2004); Johansson & Flanagan (2009)
Spike pattern discrimination in dendrites.
Branco et al. (2009) Science
Tempotron: Spike-based perceptron.
Consider five neurons and each emitting one spike but at different timings:
Rate coding: Information is coded in numbers of spikes in a given period.
( ) ( )31 2 4 5, , , , 1,1,1,1,1r r r r r =
Temporal coding: Information is coded in temporal patterns of spiking.
Tempotron: Spike-based perceptron.
Consider five neurons and each emitting one spike but at different timings:
Tempotron: Spike-based perceptron.
Basic idea: Expand the spike pattern into time:
N
T
N×T
Now
Tempotron: Spike-based perceptron.
3
1 1
t t
w e w e− ∆ −∆
+
2 2
2 t
w e w− ∆
+
2
1 1
t
w e w− ∆
+
3
2 2
t t
w e w e− ∆ −∆
+
( ) ( )2
1 2
3 2
1t t t
w e e w e θ− ∆ − ∆ − ∆
+ + + > ( ) ( )2
1 2
2 3
1t t t
w e w e e θ− ∆ − ∆ − ∆
+ + + <
( ) ( )
3 2 2
1
2 3 2
1 2
2
1
, ,
1
t t t
t t t
w e e e
w e e e
− ∆ − ∆ − ∆
− ∆ − ∆ − ∆
   + + 
= = =    
+ +     
w x x
( ) ( )T T1 2
,θ θ> <w x w x
Consider a classification problem of two spike patterns:
If a vector notation is introduced:
This classification problem is reduced to a perceptron problem:
Tempotron: Spike-based perceptron.
3
1 1
t t
w e w e− ∆ −∆
+
2 2
2 t
w e w− ∆
+
2
1 1
t
w e w− ∆
+
3
2 2
t t
w e w e− ∆ −∆
+
( ) ( )2
1 2
3 2
1t t t
w e e w e θ− ∆ − ∆ − ∆
+ + + > ( ) ( )2
1 2
2 3
1t t t
w e w e e θ− ∆ − ∆ − ∆
+ + + <
( ) ( )
3 2 2
1
2 3 2
1 2
2
1
, ,
1
t t t
t t t
w e e e
w e e e
− ∆ − ∆ − ∆
− ∆ − ∆ − ∆
   + + 
= = =    
+ +     
w x x
( ) ( )T T1 2
,θ θ> <w x w x
Consider a classification problem of two spike patterns:
If a vector notation is introduced:
This classification problem is reduced to a perceptron problem:
Learning a tempotron: intuition.
3
1 1
t t
w e w e− ∆ −∆
+
2 2
2 t
w e w− ∆
+
2
1 1
t
w e w− ∆
+
3
2 2
t t
w e w e− ∆ −∆
+
( ) ( )2
1 2
3 2
1t t t
w e e w e θ− ∆ − ∆ − ∆
+ + + > ( ) ( )2
1 2
2 3
1t t t
w e w e e θ− ∆ − ∆ − ∆
+ + >+
What was wrong if the second pattern was misclassified?
The last spike of neuron #1 (red one) is most responsible for the error, so
the synaptic strength of this neuron should be reduced.
1w λ∆ = −
Learning a tempotron: intuition.
3
1 1
t t
w e w e− ∆ −∆
+
2 2
2 t
w e w− ∆
+
2
1 1
t
w e w− ∆
+
3
2 2
t t
w e w e− ∆ −∆
+
( ) ( )2
1 2
3 2
1t t t
w e e w e θ− ∆ − ∆ − ∆
+ + <+ ( ) ( )2
1 2
2 3
1t t t
w e w e e θ− ∆ − ∆ − ∆
+ + + <
What was wrong if the second pattern was misclassified?
The last spike of neuron #2 (red one) is most responsible for the error, so
the synaptic strength of this neuron should be potentiated.
2w λ∆ = +
Exercise: Capacity of perceptron.
• Generate a set of random vectors.
• Write a code for the Perceptron learning algorithm.
• By randomly relabeling, count how many of them are
linearly separable.
Rigotti, M., Barak, O., Warden, M. R., Wang, X. J., Daw, N. D., Miller, E. K., & Fusi, S.
(2013). The importance of mixed selectivity in complex cognitive tasks. Nature,
497(7451), 585-590.
Exercise: Training of recurrent neural networks.
0
α
=
I
P
T
1 T
1
n n n n
n n
n n n
+= −
+
P r r P
P P
r P r
Goal: Investigate the effects of chaos and feedback in a recurrent
network.
( )1t n n n t+= −+ + ∆x x x Mr
T
tanhnn nz = w x
tanhn n=r x
1 nn n n ne+= −w w P r
nn ne z f= −
Recurrent dynamics without feedback:
Update of covariance matrix:
Update of weight matrix:
force_internal_all2all.m
Exercise: Training of recurrent neural networks.
0
α
=
I
P
T
1 T
1
n n n n
n n
n n n
+= −
+
P r r P
P P
r P r
Goal: Investigate the effects of chaos and feedback in a recurrent
network.
( )1
f
t n nn n n tz+= − ++ + ∆x x Mr wx
T
tanhnn nz = w x
tanhn n=r x
1 nn n n ne+= −w w P r
nn ne z f= −
Recurrent dynamics with feedback:
Update of covariance matrix:
Update of weight matrix:
force_external_feedback_loop.m
Exercise: Training of recurrent neural networks.
Goal: Investigate the effects of chaos and feedback in a recurrent
network.
• Investigate the effect of output feedback. Are there any difference
in the activities of recurrent units?
• Investigate the effect of gain parameter g. What happens if the gain
parameter is smaller than 1?
• Try to approximate some other time series such as chaotic ones.
Use the Lorentz model, for example.
References
• Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1988). Learning representations by back-
propagating errors. Cognitive modeling, 5(3), 1.
• Sejnowski, T. J., & Rosenberg, C. R. (1987). Parallel networks that learn to pronounce English text.
Complex systems, 1(1), 145-168.
• Funahashi, K. I. (1989). On the approximate realization of continuous mappings by neural networks.
Neural networks, 2(3), 183-192.
• S. Hochreiter, Y. Bengio, P. Frasconi, and J. Schmidhuber. Gradient flow in recurrent nets: the
difficulty of learning long-term dependencies
• Zipser, D. (1991). Recurrent network model of the neural mechanism of short-term active memory.
Neural Computation, 3(2), 179-193.
• Johansson, R. S., & Birznieks, I. (2004). First spikes in ensembles of human tactile afferents code
complex spatial fingertip events. Nature neuroscience, 7(2), 170-177.
• Branco, T., Clark, B. A., & Häusser, M. (2010). Dendritic discrimination of temporal input sequences
in cortical neurons. Science, 329(5999), 1671-1675.
• Gütig, R., & Sompolinsky, H. (2006). The tempotron: a neuron that learns spike timing–based
decisions. Nature neuroscience, 9(3), 420-428.
• Sussillo, D., & Abbott, L. F. (2009). Generating coherent patterns of activity from chaotic neural
networks. Neuron, 63(4), 544-557.
1 de 53

Recomendados

JAISTサマースクール2016「脳を知るための理論」講義03 Network Dynamics por
JAISTサマースクール2016「脳を知るための理論」講義03 Network DynamicsJAISTサマースクール2016「脳を知るための理論」講義03 Network Dynamics
JAISTサマースクール2016「脳を知るための理論」講義03 Network Dynamicshirokazutanaka
665 visualizações53 slides
JAISTサマースクール2016「脳を知るための理論」講義02 Synaptic Learning rules por
JAISTサマースクール2016「脳を知るための理論」講義02 Synaptic Learning rulesJAISTサマースクール2016「脳を知るための理論」講義02 Synaptic Learning rules
JAISTサマースクール2016「脳を知るための理論」講義02 Synaptic Learning ruleshirokazutanaka
911 visualizações49 slides
JAISTサマースクール2016「脳を知るための理論」講義01 Single neuron models por
JAISTサマースクール2016「脳を知るための理論」講義01 Single neuron modelsJAISTサマースクール2016「脳を知るための理論」講義01 Single neuron models
JAISTサマースクール2016「脳を知るための理論」講義01 Single neuron modelshirokazutanaka
639 visualizações48 slides
Computational Motor Control: State Space Models for Motor Adaptation (JAIST s... por
Computational Motor Control: State Space Models for Motor Adaptation (JAIST s...Computational Motor Control: State Space Models for Motor Adaptation (JAIST s...
Computational Motor Control: State Space Models for Motor Adaptation (JAIST s...hirokazutanaka
1.4K visualizações35 slides
Computational Motor Control: Kinematics & Dynamics (JAIST summer course) por
Computational Motor Control: Kinematics & Dynamics (JAIST summer course)Computational Motor Control: Kinematics & Dynamics (JAIST summer course)
Computational Motor Control: Kinematics & Dynamics (JAIST summer course)hirokazutanaka
954 visualizações29 slides
Computational Motor Control: Optimal Control for Deterministic Systems (JAIST... por
Computational Motor Control: Optimal Control for Deterministic Systems (JAIST...Computational Motor Control: Optimal Control for Deterministic Systems (JAIST...
Computational Motor Control: Optimal Control for Deterministic Systems (JAIST...hirokazutanaka
1K visualizações34 slides

Mais conteúdo relacionado

Mais procurados

Neural Processes Family por
Neural Processes FamilyNeural Processes Family
Neural Processes FamilyKota Matsui
2.7K visualizações61 slides
Random Matrix Theory and Machine Learning - Part 4 por
Random Matrix Theory and Machine Learning - Part 4Random Matrix Theory and Machine Learning - Part 4
Random Matrix Theory and Machine Learning - Part 4Fabian Pedregosa
22K visualizações41 slides
Backpropagation in Convolutional Neural Network por
Backpropagation in Convolutional Neural NetworkBackpropagation in Convolutional Neural Network
Backpropagation in Convolutional Neural NetworkHiroshi Kuwajima
81.8K visualizações14 slides
Random Matrix Theory and Machine Learning - Part 3 por
Random Matrix Theory and Machine Learning - Part 3Random Matrix Theory and Machine Learning - Part 3
Random Matrix Theory and Machine Learning - Part 3Fabian Pedregosa
22K visualizações109 slides
Random Matrix Theory and Machine Learning - Part 1 por
Random Matrix Theory and Machine Learning - Part 1Random Matrix Theory and Machine Learning - Part 1
Random Matrix Theory and Machine Learning - Part 1Fabian Pedregosa
22.9K visualizações66 slides
Random Matrix Theory and Machine Learning - Part 2 por
Random Matrix Theory and Machine Learning - Part 2Random Matrix Theory and Machine Learning - Part 2
Random Matrix Theory and Machine Learning - Part 2Fabian Pedregosa
22.2K visualizações39 slides

Mais procurados(20)

Neural Processes Family por Kota Matsui
Neural Processes FamilyNeural Processes Family
Neural Processes Family
Kota Matsui2.7K visualizações
Random Matrix Theory and Machine Learning - Part 4 por Fabian Pedregosa
Random Matrix Theory and Machine Learning - Part 4Random Matrix Theory and Machine Learning - Part 4
Random Matrix Theory and Machine Learning - Part 4
Fabian Pedregosa22K visualizações
Backpropagation in Convolutional Neural Network por Hiroshi Kuwajima
Backpropagation in Convolutional Neural NetworkBackpropagation in Convolutional Neural Network
Backpropagation in Convolutional Neural Network
Hiroshi Kuwajima81.8K visualizações
Random Matrix Theory and Machine Learning - Part 3 por Fabian Pedregosa
Random Matrix Theory and Machine Learning - Part 3Random Matrix Theory and Machine Learning - Part 3
Random Matrix Theory and Machine Learning - Part 3
Fabian Pedregosa22K visualizações
Random Matrix Theory and Machine Learning - Part 1 por Fabian Pedregosa
Random Matrix Theory and Machine Learning - Part 1Random Matrix Theory and Machine Learning - Part 1
Random Matrix Theory and Machine Learning - Part 1
Fabian Pedregosa22.9K visualizações
Random Matrix Theory and Machine Learning - Part 2 por Fabian Pedregosa
Random Matrix Theory and Machine Learning - Part 2Random Matrix Theory and Machine Learning - Part 2
Random Matrix Theory and Machine Learning - Part 2
Fabian Pedregosa22.2K visualizações
Neural Processes por Sangwoo Mo
Neural ProcessesNeural Processes
Neural Processes
Sangwoo Mo682 visualizações
Annintro por kaushaljha009
AnnintroAnnintro
Annintro
kaushaljha009179 visualizações
Artificial Neural Network Lect4 : Single Layer Perceptron Classifiers por Mohammed Bennamoun
Artificial Neural Network Lect4 : Single Layer Perceptron ClassifiersArtificial Neural Network Lect4 : Single Layer Perceptron Classifiers
Artificial Neural Network Lect4 : Single Layer Perceptron Classifiers
Mohammed Bennamoun16.6K visualizações
Nn3 por Ruchi Sharma
Nn3Nn3
Nn3
Ruchi Sharma306 visualizações
Dynamics of structures with uncertainties por University of Glasgow
Dynamics of structures with uncertaintiesDynamics of structures with uncertainties
Dynamics of structures with uncertainties
University of Glasgow1.3K visualizações
IJCER (www.ijceronline.com) International Journal of computational Engineerin... por ijceronline
IJCER (www.ijceronline.com) International Journal of computational Engineerin...IJCER (www.ijceronline.com) International Journal of computational Engineerin...
IJCER (www.ijceronline.com) International Journal of computational Engineerin...
ijceronline146 visualizações
Stochastic Alternating Direction Method of Multipliers por Taiji Suzuki
Stochastic Alternating Direction Method of MultipliersStochastic Alternating Direction Method of Multipliers
Stochastic Alternating Direction Method of Multipliers
Taiji Suzuki2.9K visualizações
05 history of cv a machine learning (theory) perspective on computer vision por zukun
05  history of cv a machine learning (theory) perspective on computer vision05  history of cv a machine learning (theory) perspective on computer vision
05 history of cv a machine learning (theory) perspective on computer vision
zukun727 visualizações
Neuronal self-organized criticality (II) por Osame Kinouchi
Neuronal self-organized criticality (II)Neuronal self-organized criticality (II)
Neuronal self-organized criticality (II)
Osame Kinouchi79 visualizações
INFLUENCE OF OVERLAYERS ON DEPTH OF IMPLANTED-HETEROJUNCTION RECTIFIERS por Zac Darcy
INFLUENCE OF OVERLAYERS ON DEPTH OF IMPLANTED-HETEROJUNCTION RECTIFIERSINFLUENCE OF OVERLAYERS ON DEPTH OF IMPLANTED-HETEROJUNCTION RECTIFIERS
INFLUENCE OF OVERLAYERS ON DEPTH OF IMPLANTED-HETEROJUNCTION RECTIFIERS
Zac Darcy144 visualizações
03 image transform por Rumah Belajar
03 image transform03 image transform
03 image transform
Rumah Belajar20.4K visualizações
Neuronal self-organized criticality por Osame Kinouchi
Neuronal self-organized criticalityNeuronal self-organized criticality
Neuronal self-organized criticality
Osame Kinouchi114 visualizações
Artificial Neural Networks Lect8: Neural networks for constrained optimization por Mohammed Bennamoun
Artificial Neural Networks Lect8: Neural networks for constrained optimizationArtificial Neural Networks Lect8: Neural networks for constrained optimization
Artificial Neural Networks Lect8: Neural networks for constrained optimization
Mohammed Bennamoun2K visualizações
Deep neural networks & computational graphs por Revanth Kumar
Deep neural networks & computational graphsDeep neural networks & computational graphs
Deep neural networks & computational graphs
Revanth Kumar40 visualizações

Destaque

KDD2016論文読み会資料(DeepIntent) por
KDD2016論文読み会資料(DeepIntent) KDD2016論文読み会資料(DeepIntent)
KDD2016論文読み会資料(DeepIntent) Sotetsu KOYAMADA(小山田創哲)
2.3K visualizações18 slides
【強化学習】Montezuma's Revenge @ NIPS2016 por
【強化学習】Montezuma's Revenge @ NIPS2016【強化学習】Montezuma's Revenge @ NIPS2016
【強化学習】Montezuma's Revenge @ NIPS2016Sotetsu KOYAMADA(小山田創哲)
2.9K visualizações34 slides
最近のRのランダムフォレストパッケージ -ranger/Rborist- por
最近のRのランダムフォレストパッケージ -ranger/Rborist-最近のRのランダムフォレストパッケージ -ranger/Rborist-
最近のRのランダムフォレストパッケージ -ranger/Rborist-Shintaro Fukushima
19.2K visualizações35 slides
機械学習によるデータ分析 実践編 por
機械学習によるデータ分析 実践編機械学習によるデータ分析 実践編
機械学習によるデータ分析 実践編Ryota Kamoshida
17.6K visualizações72 slides
Kerberos por
KerberosKerberos
KerberosGichelle Amon
2K visualizações17 slides
Os module 2 d por
Os module 2 dOs module 2 d
Os module 2 dGichelle Amon
3.4K visualizações72 slides

Destaque(20)

最近のRのランダムフォレストパッケージ -ranger/Rborist- por Shintaro Fukushima
最近のRのランダムフォレストパッケージ -ranger/Rborist-最近のRのランダムフォレストパッケージ -ranger/Rborist-
最近のRのランダムフォレストパッケージ -ranger/Rborist-
Shintaro Fukushima19.2K visualizações
機械学習によるデータ分析 実践編 por Ryota Kamoshida
機械学習によるデータ分析 実践編機械学習によるデータ分析 実践編
機械学習によるデータ分析 実践編
Ryota Kamoshida17.6K visualizações
Kerberos por Gichelle Amon
KerberosKerberos
Kerberos
Gichelle Amon2K visualizações
Os module 2 d por Gichelle Amon
Os module 2 dOs module 2 d
Os module 2 d
Gichelle Amon3.4K visualizações
Why dont you_create_new_spark_jl por Shintaro Fukushima
Why dont you_create_new_spark_jlWhy dont you_create_new_spark_jl
Why dont you_create_new_spark_jl
Shintaro Fukushima3.4K visualizações
Probabilistic Graphical Models 輪読会 #1 por Takuma Yagi
Probabilistic Graphical Models 輪読会 #1Probabilistic Graphical Models 輪読会 #1
Probabilistic Graphical Models 輪読会 #1
Takuma Yagi6K visualizações
RBM、Deep Learningと学習(全脳アーキテクチャ若手の会 第3回DL勉強会発表資料) por Takuma Yagi
RBM、Deep Learningと学習(全脳アーキテクチャ若手の会 第3回DL勉強会発表資料)RBM、Deep Learningと学習(全脳アーキテクチャ若手の会 第3回DL勉強会発表資料)
RBM、Deep Learningと学習(全脳アーキテクチャ若手の会 第3回DL勉強会発表資料)
Takuma Yagi48.5K visualizações
Women in Tech: How to Build A Human Company por Luminary Labs
Women in Tech: How to Build A Human CompanyWomen in Tech: How to Build A Human Company
Women in Tech: How to Build A Human Company
Luminary Labs77.2K visualizações
Rユーザのためのspark入門 por Shintaro Fukushima
Rユーザのためのspark入門Rユーザのためのspark入門
Rユーザのためのspark入門
Shintaro Fukushima10K visualizações
論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,... por Takuma Yagi
論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,...論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,...
論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,...
Takuma Yagi1.9K visualizações
機械学習によるデータ分析まわりのお話 por Ryota Kamoshida
機械学習によるデータ分析まわりのお話機械学習によるデータ分析まわりのお話
機械学習によるデータ分析まわりのお話
Ryota Kamoshida568.1K visualizações
What is the maker movement? por Luminary Labs
What is the maker movement?What is the maker movement?
What is the maker movement?
Luminary Labs30.1K visualizações
Network security por Gichelle Amon
Network securityNetwork security
Network security
Gichelle Amon215.7K visualizações
The Human Company Playbook, Version 1.0 por Luminary Labs
The Human Company Playbook, Version 1.0The Human Company Playbook, Version 1.0
The Human Company Playbook, Version 1.0
Luminary Labs80.8K visualizações
Hype vs. Reality: The AI Explainer por Luminary Labs
Hype vs. Reality: The AI ExplainerHype vs. Reality: The AI Explainer
Hype vs. Reality: The AI Explainer
Luminary Labs497.8K visualizações
A Non Linear Model to explain persons with Stroke por Hariohm Pandian
A Non Linear Model to explain persons with StrokeA Non Linear Model to explain persons with Stroke
A Non Linear Model to explain persons with Stroke
Hariohm Pandian823 visualizações
From epilepsy to migraine to stroke: A unifying framework. por MPI Dresden / HU Berlin
From epilepsy to migraine to stroke: A unifying framework. From epilepsy to migraine to stroke: A unifying framework.
From epilepsy to migraine to stroke: A unifying framework.
MPI Dresden / HU Berlin724 visualizações

Similar a JAISTサマースクール2016「脳を知るための理論」講義04 Neural Networks and Neuroscience

SOFT COMPUTERING TECHNICS -Unit 1 por
SOFT COMPUTERING TECHNICS -Unit 1SOFT COMPUTERING TECHNICS -Unit 1
SOFT COMPUTERING TECHNICS -Unit 1sravanthi computers
1.5K visualizações77 slides
MLIP - Chapter 2 - Preliminaries to deep learning por
MLIP - Chapter 2 - Preliminaries to deep learningMLIP - Chapter 2 - Preliminaries to deep learning
MLIP - Chapter 2 - Preliminaries to deep learningCharles Deledalle
382 visualizações168 slides
10-Perceptron.pdf por
10-Perceptron.pdf10-Perceptron.pdf
10-Perceptron.pdfESTIBALYZJIMENEZCAST
2 visualizações40 slides
ACUMENS ON NEURAL NET AKG 20 7 23.pptx por
ACUMENS ON NEURAL NET AKG 20 7 23.pptxACUMENS ON NEURAL NET AKG 20 7 23.pptx
ACUMENS ON NEURAL NET AKG 20 7 23.pptxgnans Kgnanshek
2 visualizações128 slides
19_Learning.ppt por
19_Learning.ppt19_Learning.ppt
19_Learning.pptgnans Kgnanshek
4 visualizações31 slides
6 por
66
6Vaibhav Shah
589 visualizações52 slides

Similar a JAISTサマースクール2016「脳を知るための理論」講義04 Neural Networks and Neuroscience (20)

SOFT COMPUTERING TECHNICS -Unit 1 por sravanthi computers
SOFT COMPUTERING TECHNICS -Unit 1SOFT COMPUTERING TECHNICS -Unit 1
SOFT COMPUTERING TECHNICS -Unit 1
sravanthi computers1.5K visualizações
MLIP - Chapter 2 - Preliminaries to deep learning por Charles Deledalle
MLIP - Chapter 2 - Preliminaries to deep learningMLIP - Chapter 2 - Preliminaries to deep learning
MLIP - Chapter 2 - Preliminaries to deep learning
Charles Deledalle382 visualizações
ACUMENS ON NEURAL NET AKG 20 7 23.pptx por gnans Kgnanshek
ACUMENS ON NEURAL NET AKG 20 7 23.pptxACUMENS ON NEURAL NET AKG 20 7 23.pptx
ACUMENS ON NEURAL NET AKG 20 7 23.pptx
gnans Kgnanshek2 visualizações
19_Learning.ppt por gnans Kgnanshek
19_Learning.ppt19_Learning.ppt
19_Learning.ppt
gnans Kgnanshek4 visualizações
6 por Vaibhav Shah
66
6
Vaibhav Shah589 visualizações
Machine Learning - Neural Networks - Perceptron por Andrew Ferlitsch
Machine Learning - Neural Networks - PerceptronMachine Learning - Neural Networks - Perceptron
Machine Learning - Neural Networks - Perceptron
Andrew Ferlitsch1.3K visualizações
Machine Learning - Introduction to Neural Networks por Andrew Ferlitsch
Machine Learning - Introduction to Neural NetworksMachine Learning - Introduction to Neural Networks
Machine Learning - Introduction to Neural Networks
Andrew Ferlitsch270 visualizações
Deep learning lecture - part 1 (basics, CNN) por SungminYou
Deep learning lecture - part 1 (basics, CNN)Deep learning lecture - part 1 (basics, CNN)
Deep learning lecture - part 1 (basics, CNN)
SungminYou649 visualizações
Soft Computing-173101 por AMIT KUMAR
Soft Computing-173101Soft Computing-173101
Soft Computing-173101
AMIT KUMAR3.1K visualizações
Neural Networks por Ismail El Gayar
Neural NetworksNeural Networks
Neural Networks
Ismail El Gayar4.9K visualizações
Perceptron por Nagarajan
PerceptronPerceptron
Perceptron
Nagarajan29.5K visualizações
Artificial neural networks (2) por sai anjaneya
Artificial neural networks (2)Artificial neural networks (2)
Artificial neural networks (2)
sai anjaneya717 visualizações
lecture07.ppt por butest
lecture07.pptlecture07.ppt
lecture07.ppt
butest12.9K visualizações
tutorial.ppt por Vara Prasad
tutorial.ppttutorial.ppt
tutorial.ppt
Vara Prasad17 visualizações
Artificial Neural Network seminar presentation using ppt. por Mohd Faiz
Artificial Neural Network seminar presentation using ppt.Artificial Neural Network seminar presentation using ppt.
Artificial Neural Network seminar presentation using ppt.
Mohd Faiz5.7K visualizações
Lecture artificial neural networks and pattern recognition por Hưng Đặng
Lecture   artificial neural networks and pattern recognitionLecture   artificial neural networks and pattern recognition
Lecture artificial neural networks and pattern recognition
Hưng Đặng862 visualizações
Lecture artificial neural networks and pattern recognition por Hưng Đặng
Lecture   artificial neural networks and pattern recognitionLecture   artificial neural networks and pattern recognition
Lecture artificial neural networks and pattern recognition
Hưng Đặng4.2K visualizações

Mais de hirokazutanaka

東京都市大学 データ解析入門 10 ニューラルネットワークと深層学習 1 por
東京都市大学 データ解析入門 10 ニューラルネットワークと深層学習 1東京都市大学 データ解析入門 10 ニューラルネットワークと深層学習 1
東京都市大学 データ解析入門 10 ニューラルネットワークと深層学習 1hirokazutanaka
443 visualizações46 slides
東京都市大学 データ解析入門 9 クラスタリングと分類分析 2 por
東京都市大学 データ解析入門 9 クラスタリングと分類分析 2東京都市大学 データ解析入門 9 クラスタリングと分類分析 2
東京都市大学 データ解析入門 9 クラスタリングと分類分析 2hirokazutanaka
245 visualizações45 slides
東京都市大学 データ解析入門 8 クラスタリングと分類分析 1 por
東京都市大学 データ解析入門 8 クラスタリングと分類分析 1東京都市大学 データ解析入門 8 クラスタリングと分類分析 1
東京都市大学 データ解析入門 8 クラスタリングと分類分析 1hirokazutanaka
384 visualizações50 slides
東京都市大学 データ解析入門 7 回帰分析とモデル選択 2 por
東京都市大学 データ解析入門 7 回帰分析とモデル選択 2東京都市大学 データ解析入門 7 回帰分析とモデル選択 2
東京都市大学 データ解析入門 7 回帰分析とモデル選択 2hirokazutanaka
394 visualizações37 slides
東京都市大学 データ解析入門 6 回帰分析とモデル選択 1 por
東京都市大学 データ解析入門 6 回帰分析とモデル選択 1東京都市大学 データ解析入門 6 回帰分析とモデル選択 1
東京都市大学 データ解析入門 6 回帰分析とモデル選択 1hirokazutanaka
280 visualizações41 slides
東京都市大学 データ解析入門 5 スパース性と圧縮センシング 2 por
東京都市大学 データ解析入門 5 スパース性と圧縮センシング 2東京都市大学 データ解析入門 5 スパース性と圧縮センシング 2
東京都市大学 データ解析入門 5 スパース性と圧縮センシング 2hirokazutanaka
571 visualizações44 slides

Mais de hirokazutanaka(11)

東京都市大学 データ解析入門 10 ニューラルネットワークと深層学習 1 por hirokazutanaka
東京都市大学 データ解析入門 10 ニューラルネットワークと深層学習 1東京都市大学 データ解析入門 10 ニューラルネットワークと深層学習 1
東京都市大学 データ解析入門 10 ニューラルネットワークと深層学習 1
hirokazutanaka443 visualizações
東京都市大学 データ解析入門 9 クラスタリングと分類分析 2 por hirokazutanaka
東京都市大学 データ解析入門 9 クラスタリングと分類分析 2東京都市大学 データ解析入門 9 クラスタリングと分類分析 2
東京都市大学 データ解析入門 9 クラスタリングと分類分析 2
hirokazutanaka245 visualizações
東京都市大学 データ解析入門 8 クラスタリングと分類分析 1 por hirokazutanaka
東京都市大学 データ解析入門 8 クラスタリングと分類分析 1東京都市大学 データ解析入門 8 クラスタリングと分類分析 1
東京都市大学 データ解析入門 8 クラスタリングと分類分析 1
hirokazutanaka384 visualizações
東京都市大学 データ解析入門 7 回帰分析とモデル選択 2 por hirokazutanaka
東京都市大学 データ解析入門 7 回帰分析とモデル選択 2東京都市大学 データ解析入門 7 回帰分析とモデル選択 2
東京都市大学 データ解析入門 7 回帰分析とモデル選択 2
hirokazutanaka394 visualizações
東京都市大学 データ解析入門 6 回帰分析とモデル選択 1 por hirokazutanaka
東京都市大学 データ解析入門 6 回帰分析とモデル選択 1東京都市大学 データ解析入門 6 回帰分析とモデル選択 1
東京都市大学 データ解析入門 6 回帰分析とモデル選択 1
hirokazutanaka280 visualizações
東京都市大学 データ解析入門 5 スパース性と圧縮センシング 2 por hirokazutanaka
東京都市大学 データ解析入門 5 スパース性と圧縮センシング 2東京都市大学 データ解析入門 5 スパース性と圧縮センシング 2
東京都市大学 データ解析入門 5 スパース性と圧縮センシング 2
hirokazutanaka571 visualizações
東京都市大学 データ解析入門 4 スパース性と圧縮センシング1 por hirokazutanaka
東京都市大学 データ解析入門 4 スパース性と圧縮センシング1東京都市大学 データ解析入門 4 スパース性と圧縮センシング1
東京都市大学 データ解析入門 4 スパース性と圧縮センシング1
hirokazutanaka665 visualizações
東京都市大学 データ解析入門 3 行列分解 2 por hirokazutanaka
東京都市大学 データ解析入門 3 行列分解 2東京都市大学 データ解析入門 3 行列分解 2
東京都市大学 データ解析入門 3 行列分解 2
hirokazutanaka647 visualizações
東京都市大学 データ解析入門 2 行列分解 1 por hirokazutanaka
東京都市大学 データ解析入門 2 行列分解 1東京都市大学 データ解析入門 2 行列分解 1
東京都市大学 データ解析入門 2 行列分解 1
hirokazutanaka632 visualizações
Computational Motor Control: Reinforcement Learning (JAIST summer course) por hirokazutanaka
Computational Motor Control: Reinforcement Learning (JAIST summer course) Computational Motor Control: Reinforcement Learning (JAIST summer course)
Computational Motor Control: Reinforcement Learning (JAIST summer course)
hirokazutanaka901 visualizações
Computational Motor Control: Introduction (JAIST summer course) por hirokazutanaka
Computational Motor Control: Introduction (JAIST summer course)Computational Motor Control: Introduction (JAIST summer course)
Computational Motor Control: Introduction (JAIST summer course)
hirokazutanaka339 visualizações

Último

JRN 362 - Lecture Twenty-Two por
JRN 362 - Lecture Twenty-TwoJRN 362 - Lecture Twenty-Two
JRN 362 - Lecture Twenty-TwoRich Hanley
39 visualizações157 slides
BUSINESS ETHICS MODULE 1 UNIT I_A.pdf por
BUSINESS ETHICS MODULE 1 UNIT I_A.pdfBUSINESS ETHICS MODULE 1 UNIT I_A.pdf
BUSINESS ETHICS MODULE 1 UNIT I_A.pdfDr Vijay Vishwakarma
40 visualizações25 slides
ANGULARJS.pdf por
ANGULARJS.pdfANGULARJS.pdf
ANGULARJS.pdfArthyR3
51 visualizações10 slides
Creative Restart 2023: Leonard Savage - The Permanent Brief: Unearthing unobv... por
Creative Restart 2023: Leonard Savage - The Permanent Brief: Unearthing unobv...Creative Restart 2023: Leonard Savage - The Permanent Brief: Unearthing unobv...
Creative Restart 2023: Leonard Savage - The Permanent Brief: Unearthing unobv...Taste
55 visualizações21 slides
12.5.23 Poverty and Precarity.pptx por
12.5.23 Poverty and Precarity.pptx12.5.23 Poverty and Precarity.pptx
12.5.23 Poverty and Precarity.pptxmary850239
381 visualizações30 slides
The Future of Micro-credentials: Is Small Really Beautiful? por
The Future of Micro-credentials:  Is Small Really Beautiful?The Future of Micro-credentials:  Is Small Really Beautiful?
The Future of Micro-credentials: Is Small Really Beautiful?Mark Brown
75 visualizações35 slides

Último(20)

JRN 362 - Lecture Twenty-Two por Rich Hanley
JRN 362 - Lecture Twenty-TwoJRN 362 - Lecture Twenty-Two
JRN 362 - Lecture Twenty-Two
Rich Hanley39 visualizações
BUSINESS ETHICS MODULE 1 UNIT I_A.pdf por Dr Vijay Vishwakarma
BUSINESS ETHICS MODULE 1 UNIT I_A.pdfBUSINESS ETHICS MODULE 1 UNIT I_A.pdf
BUSINESS ETHICS MODULE 1 UNIT I_A.pdf
Dr Vijay Vishwakarma40 visualizações
ANGULARJS.pdf por ArthyR3
ANGULARJS.pdfANGULARJS.pdf
ANGULARJS.pdf
ArthyR351 visualizações
Creative Restart 2023: Leonard Savage - The Permanent Brief: Unearthing unobv... por Taste
Creative Restart 2023: Leonard Savage - The Permanent Brief: Unearthing unobv...Creative Restart 2023: Leonard Savage - The Permanent Brief: Unearthing unobv...
Creative Restart 2023: Leonard Savage - The Permanent Brief: Unearthing unobv...
Taste55 visualizações
12.5.23 Poverty and Precarity.pptx por mary850239
12.5.23 Poverty and Precarity.pptx12.5.23 Poverty and Precarity.pptx
12.5.23 Poverty and Precarity.pptx
mary850239381 visualizações
The Future of Micro-credentials: Is Small Really Beautiful? por Mark Brown
The Future of Micro-credentials:  Is Small Really Beautiful?The Future of Micro-credentials:  Is Small Really Beautiful?
The Future of Micro-credentials: Is Small Really Beautiful?
Mark Brown75 visualizações
What is Digital Transformation? por Mark Brown
What is Digital Transformation?What is Digital Transformation?
What is Digital Transformation?
Mark Brown41 visualizações
NodeJS and ExpressJS.pdf por ArthyR3
NodeJS and ExpressJS.pdfNodeJS and ExpressJS.pdf
NodeJS and ExpressJS.pdf
ArthyR348 visualizações
Nelson_RecordStore.pdf por BrynNelson5
Nelson_RecordStore.pdfNelson_RecordStore.pdf
Nelson_RecordStore.pdf
BrynNelson546 visualizações
Guidelines & Identification of Early Sepsis DR. NN CHAVAN 02122023.pptx por Niranjan Chavan
Guidelines & Identification of Early Sepsis DR. NN CHAVAN 02122023.pptxGuidelines & Identification of Early Sepsis DR. NN CHAVAN 02122023.pptx
Guidelines & Identification of Early Sepsis DR. NN CHAVAN 02122023.pptx
Niranjan Chavan40 visualizações
Java Simplified: Understanding Programming Basics por Akshaj Vadakkath Joshy
Java Simplified: Understanding Programming BasicsJava Simplified: Understanding Programming Basics
Java Simplified: Understanding Programming Basics
Akshaj Vadakkath Joshy653 visualizações
Introduction to AERO Supply Chain - #BEAERO Trainning program por Guennoun Wajih
Introduction to AERO Supply Chain  - #BEAERO Trainning programIntroduction to AERO Supply Chain  - #BEAERO Trainning program
Introduction to AERO Supply Chain - #BEAERO Trainning program
Guennoun Wajih95 visualizações
ICS3211_lecture 09_2023.pdf por Vanessa Camilleri
ICS3211_lecture 09_2023.pdfICS3211_lecture 09_2023.pdf
ICS3211_lecture 09_2023.pdf
Vanessa Camilleri141 visualizações
INT-244 Topic 6b Confucianism por S Meyer
INT-244 Topic 6b ConfucianismINT-244 Topic 6b Confucianism
INT-244 Topic 6b Confucianism
S Meyer45 visualizações
UNIDAD 3 6º C.MEDIO.pptx por MarcosRodriguezUcedo
UNIDAD 3 6º C.MEDIO.pptxUNIDAD 3 6º C.MEDIO.pptx
UNIDAD 3 6º C.MEDIO.pptx
MarcosRodriguezUcedo146 visualizações
Berry country.pdf por MariaKenney3
Berry country.pdfBerry country.pdf
Berry country.pdf
MariaKenney375 visualizações
Guess Papers ADC 1, Karachi University por Khalid Aziz
Guess Papers ADC 1, Karachi UniversityGuess Papers ADC 1, Karachi University
Guess Papers ADC 1, Karachi University
Khalid Aziz99 visualizações
Six Sigma Concept by Sahil Srivastava.pptx por Sahil Srivastava
Six Sigma Concept by Sahil Srivastava.pptxSix Sigma Concept by Sahil Srivastava.pptx
Six Sigma Concept by Sahil Srivastava.pptx
Sahil Srivastava44 visualizações
Education of marginalized and socially disadvantages segments.pptx por GarimaBhati5
Education of marginalized and socially disadvantages segments.pptxEducation of marginalized and socially disadvantages segments.pptx
Education of marginalized and socially disadvantages segments.pptx
GarimaBhati543 visualizações
Monthly Information Session for MV Asterix (November) por Esquimalt MFRC
Monthly Information Session for MV Asterix (November)Monthly Information Session for MV Asterix (November)
Monthly Information Session for MV Asterix (November)
Esquimalt MFRC107 visualizações

JAISTサマースクール2016「脳を知るための理論」講義04 Neural Networks and Neuroscience

  • 1. SS2016 Modern Neural Computation Lecture 5: Neural Networks and Neuroscience Hirokazu Tanaka School of Information Science Japan Institute of Science and Technology
  • 2. Supervised learning as functional approximation. In this lecture we will learn: • Single-layer neural networks Perceptron and the perceptron theorem. Cerebellum as a perceptron. • Multi-layer feedforward neural networks Universal functional approximations, Back-propagation algorithms • Recurrent neural networks Back-propagation-through-time (BPTT) algorithms • Tempotron Spike-based perceptron
  • 3. Gradient-descent learning for optimization. • Classification problem: to output discrete labels. For a binary classification (i.e., 0 or 1), a cross-entropy is often used. • Regression problem: to output continuous values. Sum of squared errors is often used.
  • 4. Cost function: classification and regression. • Classification problem: to output discrete labels. For a binary classification (i.e., 0 or 1), a cross-entropy is often used. • Regression problem: to output continuous values. Sum of squared errors is often used. ˆ:output of network, :desired outputi iy y ( ) ( ) ( ) ˆ1ˆ : samples: samples ˆ ˆlog 1 log 1 log 1ii i i i i i yy i ii y y y y y y − − − =− + − −  ∑∏ ( ) : sa p e 2 m l s ˆi i iy y−∑
  • 5. Perceptron: single-layer neural network. • Assume a single-layer neural network with an input layer composed of N units and an output layer composed of one unit. • Input units are specified by and an output unit are determined by ( )1 T Nx x=x  ( )T 0 1 0 n i i iy f w x fw w =   = + = +    ∑ w x ( ) 1 if 0 0 if 0 u f u u ≥ =  <
  • 6. Perceptron: single-layer neural network. feature 1 feature 2
  • 7. Perceptron: single-layer neural network. • [Remark] Instead of using often, an augmented input vector are used. Then, ( )1 T Nx x=x  ( ) ( )T T 0y f w f= + =w x w x ( )11 T Nx x=x  ( )10 T Nw w w=w 
  • 8. Perceptron Learning Algorithm. ( ) ( ) ( ){ }21 1 2, , ,, , ,P Pd d dx x x • Given a training set: • Perceptron learning rule: ( )i i iydη −∆ =w x while err>1e-4 && count<10 y = sign(w'*X)'; wnew = w + X*(d-y)/P; wnew = wnew/norm(wnew); count = count+1; err = norm(w-wnew)/norm(w) w = wnew; end
  • 9. Perceptron Learning Algorithm. Case 1: Linearly separable case
  • 10. Perceptron Learning Algorithm. Case 2: Linearly non-separable case
  • 11. Perceptron’s capacity: Cover’s Counting Theorem. • Question: Suppose that there are P vectors in N- dimensional Euclidean space. There are 2P possible patterns of two classes. How many of them are linearly separable? [Remark] They are assumed to be in general position. • Answer: Cover’s Counting Theorem. { }1, ,, N P i ∈x x x  ( ) 1 0 1 , 2 N k P C P N k − = −  =     ∑
  • 12. Perceptron’s capacity: Cover’s Counting Theorem. • Cover’s Counting Theorem. • Case 𝑃𝑃 ≤ 𝑁𝑁: • Case 𝑃𝑃 = 2𝑁𝑁: • Case 𝑃𝑃 ≫ 𝑁𝑁: ( ) 1 0 1 , 2 N k P C P N k − = −  =     ∑ ( ), 2P C P N = ( ) 1 , 2P C P N − = ( ), N C P N AP≈ Cover (1965) IEEE Information; Sompolinsky (2013) MIT lecture note
  • 13. Perceptron’s capacity: Cover’s Counting Theorem. • Case for large P: Orhan (2014) “Cover’s Function Counting Theorem” ( ) 1 2 1 e 2 rf , 2 2P pC P N N p    + −        ≈
  • 14. Cerebellum as a Perceptron. Llinas (1974) Scientific American
  • 15. Cerebellum as a Perceptron. • Cerebellar cortex has a feedforward structure: mossy fibers -> granule cells -> parallel fibers -> Purkinje cells Ito (1984) “Cerebellum and Neural Control”
  • 16. Cerebellum as a Perceptron (or its extensions) • Perceptron model Marr (1969): Long-term potentiation (LTP) learning. Albus (1971): Long-term depression (LTD) learning. • Adaptive filter theory Fujita (1982): Reverberation among granule and Golgi cells for generating temporal templates. • Liquid-state machine model Yamazaki and Tanaka (2007):
  • 17. Perceptron: a new perspective. • Evaluation of memory capacity of a Purkinje cell using perceptron methods (the Gardner limit). Brunel, N., Hakim, V., Isope, P., Nadal, J. P., & Barbour, B. (2004). Optimal information storage and the distribution of synaptic weights: perceptron versus Purkinje cell. Neuron, 43(5), 745-757. • Estimation of dimensions of neural representations during visual memory task in the prefrontal cortex using perceptron methods (Cover’s counting theorem). Rigotti, M., Barak, O., Warden, M. R., Wang, X. J., Daw, N. D., Miller, E. K., & Fusi, S. (2013). The importance of mixed selectivity in complex cognitive tasks. Nature, 497(7451), 585-590.
  • 18. Limitation of Perceptron. • Only linearly separable input-output sets can be learned. • Non-linear sets, even a simple one like XOR, CANNOT be learned.
  • 19. Multilayer neural network: feedforward design ( )n ix ( )1n jx − Layer 1 Layer n-1 Layer n Layer N ( )1n ijw − • Feedforward network: a unit in layer n receives inputs from layer n-1 and projects to layer n+1.
  • 20. Multilayer neural network: feedforward design ( )n ix ( )1n jx − Layer 1 Layer n-1 Layer n Layer N ( )1n ijw − • Feedforward network: a unit in layer n receives inputs from layer n-1 and projects to layer n+1.
  • 21. Multilayer neural network: forward propagation. ( ) ( ) ( ) ( ) ( )1 1 1 n n n n i i ij j j x f u f w x− − =   = =     ∑ ( ) 1 1 u f u e− = + ( ) ( ) ( ) ( )( )2 1 1 1 1 1 11 u u uu f e e e u e f u f u − − −−   = = − =  + + + ′ − Layer n-1 Layer n ( )n ix ( )1n jx − ( )1n ijw − ( ) ( ) ( )1 1 1 n n n i ij j j u w x − − = = ∑ In a feedforward multilayer neural network propagates its activities from one layer to another in one direction: Inputs to neurons in layer n are a summation of activities of neurons in layer n-1: The function f is called an activation function, and its derivative is easy to compute:
  • 22. Multilayer neural network: error backpropagation • Define an cost function as a squared sum of errors in output units: Gradients of cost function with respect to weights: ( ) ( ) ( ) ( ) 2 21 1 2 2 N N i i i i i x z= − = ∆∑ ∑ Layer n-1 Layer n ( ) ( ) ( ) ( ) ( ) ( )1 1 1 n n n n n i j j j ji j x x w − − ∆ = ∆ −∑ ( )1n j − ∆ ( )n i∆ The neurons in the output layer has explicit supervised errors (the difference between the network outputs and the desired outputs). How, then, to compute the supervising signals for neurons in intermediate layers?
  • 23. Multilayer neural network: error backpropagation 1. Compute activations of units in all layers. 2. Compute errors in the output units, . 3. “Back-propagate” the errors to lower layers using 4. Update the weights ( ) { } ( ) { } ( ) { }1 ,, , , n N i i ix x x  ( ) { }N i∆ ( ) ( ) ( ) ( ) ( ) ( )1 1 1n n n n n i j j j ji j x x w − − ∆ = ∆ −∑ ( ) ( ) ( ) ( ) ( ) ( )1 1 1 1 n n n n n ij i i i jw x x xη + + + ∆ =− ∆ −
  • 24. Multilayer neural network as universal machine for functional approximation. A multilayer neural network is in principle able to approximate any functional relationship between inputs and outputs at any desired accuracy (Funahashi, 1988). Intuition: A sum or a difference of two sigmoid functions is a “bump- like” function. And, a sufficiently large number of bump functions can approximate any function.
  • 25. NETtalk: A parallel network that learns to read aloud. Sejnowski & Rosenberg (1987) Complex Systems A feedforward three-layer neural network with delay lines.
  • 26. NETtalk: A parallel network that learns to read aloud. Sejnowski & Rosenberg (1987) Complex Systems; https://www.youtube.com/watch?v=gakJlr3GecE A feedforward three-layer neural network with delay lines.
  • 27. NETtalk: A parallel network that learns to read aloud. Sejnowski & Rosenberg (1987) Complex Systems Activations of hidden units for a same sound but different inputs
  • 28. Hinton diagrams: characterizing and visualizing connection to and from hidden units. Hinton (1992) Sci Am Activations of hidden units for a same sound but different inputs
  • 29. Autonomous driving learning by backpropagation. Pomerleau (1991) Neural Comput Activations of hidden units for a same sound but different inputs
  • 30. Autonomous driving learning by backpropagation. Pomerleau (1991) Neural Comput; https://www.youtube.com/watch?v=ilP4aPDTBPE
  • 31. Gradient vanishing problem: why is training a multi-layer neural network so difficult? Hochreiter et al. (1991) • The back-propagation algorithm works only for neural networks of three or four layers. • Training neural networks with many hidden layers – called “deep neural networks”- is notoriously difficult. ( ) ( ) ( ) ( ) ( ) ( )1 1 1N N N N N j i i i ij i x x w− − ∆ = ∆ −∑ ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) 2 1 1 1 2 1 1 1 2 1 1 1 N N N N N k j j j jk j N N N N N N N i i i ij j j jk j i x x w x x w x x w − − − − − − − − − ∆ = ∆ −   = ∆ − −    ∑ ∑ ∑ ( ) ( ) ( ) ( ) ( )( 1) ( 1) ( 1) ( 1) ( ) ( ) ~ 1 1 1 n Nn n N N N N x x x x x x+ + − − ∆ − × × − × − ×∆
  • 32. Multilayer neural network: recurrent connections • A feedforward neural network can represent an instantaneous relationship between inputs and outputs - memoryless: it depends on current inputs but not on previous inputs. • In order to describe a history, a neural network should have its own dynamics. • One way to incorporate dynamics into a neural network is to introduce recurrent connections between units.
  • 33. Working memory in the parietal cortex. • A feedforward neural network can represent an instantaneous relationship between inputs and outputs - memoryless: it depends on current inputs x(t) but not on previous inputs x(t-1), x(t-2), ... • In order to describe a history, a neural network should have its own dynamics. • One way to incorporate dynamics into a neural network is to introduce recurrent connections between units.
  • 34. Multilayer neural network: recurrent connections ( ) ( )( ) ( ) ( )( )( )1 1ii i x t f u t f t t+= += +Wx Ua ( ) ( )( )iz t g t= Vx Recurrent dynamics of neural network: Output readout: a x z U VW
  • 35. Temporal unfolding: backpropagation through time (BPTT) 1t−a 1t−x tztx { }10 2 1,, , ,, ,t T −a a a aa   { }1 2 3, , , ,, ,t Tzz z zz   ,U W V Training set for a recurrent network: Input series: Output series: Optimize the weight matrices so as to approximate the training set:
  • 36. Temporal unfolding: backpropagation through time (BPTT) 0a 1z1x,U W V 0a 2z1x,U W V,U W 1a 2x 0a 3z 1x,U W V ,U W 1a 3x2x ,U W 2a 1t−a 1t−x tztx,U W V
  • 37. Working-memory related activity in parietal cortex. Gnadt & Andersen (1988) Exp Brain Res
  • 38. Temporal unfolding: backpropagation through time (BPTT) Zipser (1991) Neural Comput
  • 39. Temporal unfolding: backpropagation through time (BPTT) Zipser (1991) Neural Comput Model Experiment Model Experiment
  • 40. Spike pattern discrimination in humans. Johansson & Birznieks (2004); Johansson & Flanagan (2009)
  • 41. Spike pattern discrimination in dendrites. Branco et al. (2009) Science
  • 42. Tempotron: Spike-based perceptron. Consider five neurons and each emitting one spike but at different timings: Rate coding: Information is coded in numbers of spikes in a given period. ( ) ( )31 2 4 5, , , , 1,1,1,1,1r r r r r = Temporal coding: Information is coded in temporal patterns of spiking.
  • 43. Tempotron: Spike-based perceptron. Consider five neurons and each emitting one spike but at different timings:
  • 44. Tempotron: Spike-based perceptron. Basic idea: Expand the spike pattern into time: N T N×T Now
  • 45. Tempotron: Spike-based perceptron. 3 1 1 t t w e w e− ∆ −∆ + 2 2 2 t w e w− ∆ + 2 1 1 t w e w− ∆ + 3 2 2 t t w e w e− ∆ −∆ + ( ) ( )2 1 2 3 2 1t t t w e e w e θ− ∆ − ∆ − ∆ + + + > ( ) ( )2 1 2 2 3 1t t t w e w e e θ− ∆ − ∆ − ∆ + + + < ( ) ( ) 3 2 2 1 2 3 2 1 2 2 1 , , 1 t t t t t t w e e e w e e e − ∆ − ∆ − ∆ − ∆ − ∆ − ∆    + +  = = =     + +      w x x ( ) ( )T T1 2 ,θ θ> <w x w x Consider a classification problem of two spike patterns: If a vector notation is introduced: This classification problem is reduced to a perceptron problem:
  • 46. Tempotron: Spike-based perceptron. 3 1 1 t t w e w e− ∆ −∆ + 2 2 2 t w e w− ∆ + 2 1 1 t w e w− ∆ + 3 2 2 t t w e w e− ∆ −∆ + ( ) ( )2 1 2 3 2 1t t t w e e w e θ− ∆ − ∆ − ∆ + + + > ( ) ( )2 1 2 2 3 1t t t w e w e e θ− ∆ − ∆ − ∆ + + + < ( ) ( ) 3 2 2 1 2 3 2 1 2 2 1 , , 1 t t t t t t w e e e w e e e − ∆ − ∆ − ∆ − ∆ − ∆ − ∆    + +  = = =     + +      w x x ( ) ( )T T1 2 ,θ θ> <w x w x Consider a classification problem of two spike patterns: If a vector notation is introduced: This classification problem is reduced to a perceptron problem:
  • 47. Learning a tempotron: intuition. 3 1 1 t t w e w e− ∆ −∆ + 2 2 2 t w e w− ∆ + 2 1 1 t w e w− ∆ + 3 2 2 t t w e w e− ∆ −∆ + ( ) ( )2 1 2 3 2 1t t t w e e w e θ− ∆ − ∆ − ∆ + + + > ( ) ( )2 1 2 2 3 1t t t w e w e e θ− ∆ − ∆ − ∆ + + >+ What was wrong if the second pattern was misclassified? The last spike of neuron #1 (red one) is most responsible for the error, so the synaptic strength of this neuron should be reduced. 1w λ∆ = −
  • 48. Learning a tempotron: intuition. 3 1 1 t t w e w e− ∆ −∆ + 2 2 2 t w e w− ∆ + 2 1 1 t w e w− ∆ + 3 2 2 t t w e w e− ∆ −∆ + ( ) ( )2 1 2 3 2 1t t t w e e w e θ− ∆ − ∆ − ∆ + + <+ ( ) ( )2 1 2 2 3 1t t t w e w e e θ− ∆ − ∆ − ∆ + + + < What was wrong if the second pattern was misclassified? The last spike of neuron #2 (red one) is most responsible for the error, so the synaptic strength of this neuron should be potentiated. 2w λ∆ = +
  • 49. Exercise: Capacity of perceptron. • Generate a set of random vectors. • Write a code for the Perceptron learning algorithm. • By randomly relabeling, count how many of them are linearly separable. Rigotti, M., Barak, O., Warden, M. R., Wang, X. J., Daw, N. D., Miller, E. K., & Fusi, S. (2013). The importance of mixed selectivity in complex cognitive tasks. Nature, 497(7451), 585-590.
  • 50. Exercise: Training of recurrent neural networks. 0 α = I P T 1 T 1 n n n n n n n n n += − + P r r P P P r P r Goal: Investigate the effects of chaos and feedback in a recurrent network. ( )1t n n n t+= −+ + ∆x x x Mr T tanhnn nz = w x tanhn n=r x 1 nn n n ne+= −w w P r nn ne z f= − Recurrent dynamics without feedback: Update of covariance matrix: Update of weight matrix: force_internal_all2all.m
  • 51. Exercise: Training of recurrent neural networks. 0 α = I P T 1 T 1 n n n n n n n n n += − + P r r P P P r P r Goal: Investigate the effects of chaos and feedback in a recurrent network. ( )1 f t n nn n n tz+= − ++ + ∆x x Mr wx T tanhnn nz = w x tanhn n=r x 1 nn n n ne+= −w w P r nn ne z f= − Recurrent dynamics with feedback: Update of covariance matrix: Update of weight matrix: force_external_feedback_loop.m
  • 52. Exercise: Training of recurrent neural networks. Goal: Investigate the effects of chaos and feedback in a recurrent network. • Investigate the effect of output feedback. Are there any difference in the activities of recurrent units? • Investigate the effect of gain parameter g. What happens if the gain parameter is smaller than 1? • Try to approximate some other time series such as chaotic ones. Use the Lorentz model, for example.
  • 53. References • Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1988). Learning representations by back- propagating errors. Cognitive modeling, 5(3), 1. • Sejnowski, T. J., & Rosenberg, C. R. (1987). Parallel networks that learn to pronounce English text. Complex systems, 1(1), 145-168. • Funahashi, K. I. (1989). On the approximate realization of continuous mappings by neural networks. Neural networks, 2(3), 183-192. • S. Hochreiter, Y. Bengio, P. Frasconi, and J. Schmidhuber. Gradient flow in recurrent nets: the difficulty of learning long-term dependencies • Zipser, D. (1991). Recurrent network model of the neural mechanism of short-term active memory. Neural Computation, 3(2), 179-193. • Johansson, R. S., & Birznieks, I. (2004). First spikes in ensembles of human tactile afferents code complex spatial fingertip events. Nature neuroscience, 7(2), 170-177. • Branco, T., Clark, B. A., & Häusser, M. (2010). Dendritic discrimination of temporal input sequences in cortical neurons. Science, 329(5999), 1671-1675. • Gütig, R., & Sompolinsky, H. (2006). The tempotron: a neuron that learns spike timing–based decisions. Nature neuroscience, 9(3), 420-428. • Sussillo, D., & Abbott, L. F. (2009). Generating coherent patterns of activity from chaotic neural networks. Neuron, 63(4), 544-557.