hebb rule
DESCRIPTION
Hebb Rule. Linear neuron Hebb rule Similar to LTP (but not quite…). Hebb Rule. Average Hebb rule= correlation rule Q: correlation matrix of u. Hebb Rule. Hebb rule with threshold= covariance rule C: covariance matrix of u - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/1.jpg)
Hebb Rule
• Linear neuron
• Hebb rule
• Similar to LTP (but not quite…)
Tv w u
1, w t tw
d tv v
dt w
u w w u
![Page 2: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/2.jpg)
Hebb Rule
• Average Hebb rule= correlation rule
• Q: correlation matrix of u
w
T Tw
T T
dv
dtd
vdt
wu
wu w u u u w u
u u w uu w Qw
![Page 3: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/3.jpg)
Hebb Rule
• Hebb rule with threshold= covariance rule
• C: covariance matrix of u• Note that <(v-< v >)(u-< u >)> would be unrealistic because it predicts
LTP when both u and v are low
Tw
T T
dv
dtC
C
wu u w u u u
w
u u u u u u u
![Page 4: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/4.jpg)
Hebb Rule
• Main problem with Hebb rule: it’s unstable… Two solutions:
1. Bounded weights
2. Normalization of either the activity of the postsynaptic cells or the weights.
![Page 5: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/5.jpg)
BCM rule
• Hebb rule with sliding threshold
• BCM rule implements competition because when a synaptic weight grows, it raises by v2, making more difficult for other weights to grow.
2
v
v
w v
vv
w
dv v
dtd
vdt
wu
![Page 6: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/6.jpg)
Weight Normalization
• Subtractive Normalization:
1
, 11 11
1
1 0
u
wu
Ni
w i kku
wu
u
vdv
dt N
dwvu vu
dt N
vdv
dt N
vN
n u nwu n
n u nn wn u
n nn u
1
Const.uN
ii
w
n w
![Page 7: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/7.jpg)
Weight Normalization
• Multiplicative Normalization:
• Norm of the weights converge to 1/
2
2
222
2
2 1
w
w w
dv v
dt
d ddt dt
dv
dt
wu w
w ww
ww
2
1
Const.uN
ii
w
![Page 8: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/8.jpg)
• Convergence properties:
• Use an eigenvector decomposition:
where e are the eigenvectors of Q
Hebb Rule
w
ddt
wQw
1
uN
t c t
w e
![Page 9: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/9.jpg)
Hebb Rule
e2e1
1>2
![Page 10: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/10.jpg)
Hebb Rule
1
, uN
w
w
w
w
d tt t c t
dt
dc tc t
dtdc t
c tdt
dc tc t
dt
wQw w e
eQ e
e Qe
e e
Equations decouple because
e are the eigenvectors of Q
![Page 11: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/11.jpg)
Hebb Rule
1
1 1
exp 0 exp 0
exp 0
for large , ,
u
w
w
w w
N
w
dc tc t
dtdc t
c tdt
t tc t c w
tt w
t t v
e e
e
w e e
w e e u
![Page 12: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/12.jpg)
Hebb Rule
• The weights line up with first eigenvector and the postsynaptic activity, v, converges toward the projection of u onto the first eigenvector (unstable PCA)
![Page 13: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/13.jpg)
Hebb Rule
• Non zero mean distribution: correlation vs covariance
![Page 14: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/14.jpg)
Hebb Rule
• Limiting weights growth affects the final state
0 0.2 0.4 0.6 0.8 1
0
0.2
0.4
0.6
1
w1/wmax
w2/w
max
0.8 First eigenvector: [1,-1]
![Page 15: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/15.jpg)
Hebb Rule
• Normalization also affects the final state. • Ex: multiplicative normalization. In this case,
Hebb rule extracts the first eigenvector but keeps the norm constant (stable PCA).
![Page 16: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/16.jpg)
• Normalization also affects the final state. • Ex: subtractive normalization.
Hebb Rule
1
111
if
0
wu
wu u
ddt N
ddt N N
w Qn nwQw
e n
e Qn n n Qn neQe Qn
![Page 17: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/17.jpg)
Hebb Rule
1
1
1 1
if
0, 1
wu
u
u
c tdc tc t
dt N
c tc t
N
c tc t
N
c t
e n
e n
e Qn ne Q e
e Qe nQ e
e e nQ e
Q e
![Page 18: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/18.jpg)
• The constrain does not affect the other eigenvector:
• The weights converge to the second eigenvector (the weights need to be bounded to guarantee stability…)
Hebb Rule
1 12
0 exp 0uN
w
tt w w
w e e e e
![Page 19: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/19.jpg)
Ocular Dominance Column
• One unit with one input from right and left eyes
R R L L
s dR R L R
d sR L L L
v w u w u
q qu u u u
q qu u u u
Q uu
s: same eye
d: different eyes
![Page 20: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/20.jpg)
Ocular Dominance Column
• The eigenvectors are:
s dR R L RT
d sR L L L
q qu u u u
q qu u u u
Q uu
1 1
2 2
1,1 / 2,
1, 1 / 2,
s d
s d
q q
q q
e
e
![Page 21: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/21.jpg)
Ocular Dominance Column
• Since qd is likely to be positive, qs+qd>qs-qd. As a result, the weights will converge toward the first eigenvector which mixes the right and left eye equally. No ocular dominance...
1 1
2 2
1,1 / 2,
1, 1 / 2,
s d
s d
q q
q q
e
e
![Page 22: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/22.jpg)
Ocular Dominance Column
• To get ocular dominance we need subtractive normalization.
1 1
2 2
1,1 / 2,
1, 1 / 2,
s d
s d
q q
q q
e
e
![Page 23: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/23.jpg)
Ocular Dominance Column
• Note that the weights will be proportional to e2 or –e2 (i.e. the right and left eye are equally likely to dominate at the end). Which one wins depends on the initial conditions.
1 1
2 2
1,1 / 2,
1, 1 / 2,
s d
s d
q q
q q
e
e
![Page 24: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/24.jpg)
Ocular Dominance Column
• Ocular dominance column: network with multiple output units and lateral connections.
![Page 25: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/25.jpg)
Ocular Dominance Column
• Simplified model
u L uR
B
![Page 26: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/26.jpg)
Ocular Dominance Column
• If we use subtractive normalization and no lateral connections, we’re back to the one cell case. Ocular dominance is determined by initial weights, i.e., it is purely stochastic. This is not what’s observed in V1.
• Lateral weights could help by making sure that neighboring cells have similar ocular dominance.
![Page 27: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/27.jpg)
Ocular Dominance Column
• Lateral weights are equivalent to feedforward weights
0?
ii iR R iL L
R R L L
dvv w u w u
dtd
u udt
ddt
+ Mv
vv w w + Mv
v
![Page 28: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/28.jpg)
Ocular Dominance Column
• Lateral weights are equivalent to feedforward weights
1
i iR R iL L
R R L L
RR L
L
v w u w u
u u
u
u
= + Mv
v = w w + Mv
v = I M w w
v = KWu
![Page 29: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/29.jpg)
Ocular Dominance Column
w
w
w
ddt
ddt
ddt
Wvu
v = KWu
WKWuu
WKWQ
![Page 30: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/30.jpg)
Ocular Dominance Column
• We first project the weight vectors of each cortical unit (wiR,wiL) onto the eigenvectors of Q.
1
w
w
w
ddt
ddt
ddt
WKWQ
WKWPΛP
WPKWPΛ
![Page 31: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/31.jpg)
Ocular Dominance Column
• There are two eigenvectors, w+ and w-, with eigenvalues qs+qd and qs-qd:
=
=
R L
R L
R L R L
w w w
w w w
WP w w w w
w w
![Page 32: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/32.jpg)
Ocular Dominance Column
0
0
w
s dw
s d
s dw
s d
ddtd
q qdtq qd
dtd
q qdtq qd
dt
WPKWPΛ
w
K w ww
wK w
K ww
![Page 33: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/33.jpg)
Ocular Dominance Column
• Ocular dominance column: network with multiple output units and lateral connections.
w s d
w s d
dq q
dtd
q qdt
wKw
wKw
![Page 34: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/34.jpg)
Ocular Dominance Column
• Once again we use a subtractive normalization, which holds w+ constant. Consequently, the equation for w- is the only one we need to worry about.
w s d
dq q
dt
wKw
![Page 35: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/35.jpg)
Ocular Dominance Column
• If the lateral weights are translation invariant, Kw- is a convolution. This is easier to solve in the Fourier domain.
( )*
w s d
s d
kw s d k k
dq q
dtq q x
d wq q K w
dt
wKw
K w
![Page 36: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/36.jpg)
Ocular Dominance Column
• The sine function with the highest Fourier coefficient (i.e. the fundamental) growth the fastest.
exp 0
kw s d k k
s d kk k
w
d wq q K w
dt
q q K tw t w
![Page 37: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/37.jpg)
Ocular Dominance Column
• In other words, the eigenvectors of K are sine functions and the eigenvalues are the Fourier coefficients for K.
2cosa
v
ae
N
![Page 38: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/38.jpg)
Ocular Dominance Column
• The dynamics is dominated by the sine function with the highest Fourier coefficients, i.e., the fundamental of K(x) (note that w- is not normalized along the x dimension).
• This results is an alternation of right and left columns with a periodicity corresponding to the frequency of the fundamental of K(x).
![Page 39: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/39.jpg)
Ocular Dominance Column
• If K is a Gaussian kernel, the fundamental is the DC term and w ends up being constant, i.e., no ocular dominance columns (one of the eyes dominate all the cells).
• If K is a mexican hat kernel, w will show ocular dominance column with the same frequency as the fundamental of K.
• Not that intuitive anymore…
![Page 40: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/40.jpg)
Ocular Dominance Column
• Simplified model
-0 .6 -0 .4 -0 .2 0 0.2 0.4 0.6-1
-0 .5
0
0.5
1
0 20 40 600
0.2
0.4
0.6
c o rtic a l d ista nc e (m m ) k (1/m m )
K, e K~
A B
![Page 41: Hebb Rule](https://reader036.vdocuments.us/reader036/viewer/2022062301/56815560550346895dc32adb/html5/thumbnails/41.jpg)
Ocular Dominance Column
• Simplified model: weights matrices for right and left eyes
W L W R W R W LW W W - W