研究如下的Hopfield神经网络
C
i
d
x
i
(
t
)
d
t
=
−
1
R
x
i
(
t
)
+
∑
j
=
1
n
T
i
j
g
j
(
x
j
(
t
)
)
+
I
i
,
(
i
=
1
,
2
,
⋯
,
n
)
(1)
C_{i} \frac{\mathrm{d} x_{i}(t)}{\mathrm{d} t}=-\frac{1}{R} x_{i}(t)+\sum_{j=1}^{n} T_{i j} g_{j}\left(x_{j}(t)\right)+I_{i},\;(i=1,2, \cdots, n) \tag{1}
Cidtdxi(t)=−R1xi(t)+j=1∑nTijgj(xj(t))+Ii,(i=1,2,⋯,n)(1)
其中 T = [ T i j ] n × n T=[T_{ij}]_{n×n} T=[Tij]n×n 是网络连接权系数矩阵， C i > 0 , R i > 0 C_i>0,R_i>0 Ci>0,Ri>0 及 I i I_i Ii 为常数， g i ( ⋅ ) g_i(\cdot) gi(⋅) 为可微的严格单调上升的有界函数，且 lim l → + ∞ g i ( s ) = ± 1 \lim\limits_{l\rightarrow +\infty}g_i(s)=\pm1 l→+∞limgi(s)=±1
我们称
(
x
1
∗
,
x
2
∗
,
.
.
.
,
x
n
∗
)
(x_1^*,x_2^*,...,x_n^*)
(x1∗,x2∗,...,xn∗) 为网络(1)的平衡态，若
(
x
1
∗
,
x
2
∗
,
.
.
.
,
x
n
∗
)
(x_1^*,x_2^*,...,x_n^*)
(x1∗,x2∗,...,xn∗) 满足：
−
1
R
x
i
∗
(
t
)
+
∑
j
=
1
n
T
i
j
g
j
(
x
j
∗
(
t
)
)
+
I
i
=
0
,
(
i
=
1
,
2
,
⋯
,
n
)
-\frac{1}{R} x_{i}^*(t)+\sum_{j=1}^{n} T_{i j} g_{j}\left(x_{j}^*(t)\right)+I_{i}=0,\;(i=1,2, \cdots, n)
−R1xi∗(t)+j=1∑nTijgj(xj∗(t))+Ii=0,(i=1,2,⋯,n)
证明
定理2. 若连接权系数矩阵T是对称的，则网络(1)是稳定的.
证明. 我们先证明网络的一切解有界.由式(1)有
x
i
(
t
)
=
x
i
(
0
)
e
−
1
C
i
R
i
+
1
C
i
∫
0
t
e
−
1
C
i
R
i
(
t
−
s
)
(
∑
j
=
1
n
T
i
j
g
j
(
x
j
(
s
)
)
+
I
i
)
d
s
x_{i}(t)=x_{i}(0) \mathrm{e}^{-\frac{1}{C_{i} R_{i}}}+\frac{1}{C_{i}} \int_{0}^{t} \mathrm{e}^{-\frac{1}{C_{i} R_{i}}(t-s)}\left(\sum_{j=1}^{n} T_{i j} g_{j}\left(x_{j}(s)\right)+I_{i}\right) \mathrm{d} s
xi(t)=xi(0)e−CiRi1+Ci1∫0te−CiRi1(t−s)(j=1∑nTijgj(xj(s))+Ii)ds
于是，对于一切
t
≥
0
t\ge0
t≥0 有
∣
x
i
(
t
)
∣
⩽
∣
x
i
(
0
)
∣
e
−
1
C
i
R
i
+
1
C
i
∫
0
t
e
−
1
C
i
R
i
(
t
−
s
)
∣
∑
j
=
1
n
T
i
j
g
j
(
x
j
(
s
)
)
+
I
i
∣
d
s
⩽
∣
x
i
(
0
)
∣
e
−
1
C
i
R
i
+
R
i
(
∑
j
=
1
n
∣
T
i
j
∣
+
I
i
)
⩽
∣
x
i
(
0
)
∣
+
R
i
(
∑
j
=
1
n
∣
T
i
j
∣
+
I
i
)
,
(
i
=
1
,
2
,
⋯
,
n
)
\begin{aligned} \left|x_{i}(t)\right| & \leqslant\left|x_{i}(0)\right| \mathrm{e}^{-\frac{1}{C_{i} R_{i}}}+\frac{1}{C_{i}} \int_{0}^{t} \mathrm{e}^{-\frac{1}{C_{i} R_{i}}(t-s)}\left|\sum_{j=1}^{n} T_{i j} g_{j}\left(x_{j}(s)\right)+I_{i}\right| \mathrm{d} s \\ & \leqslant\left|x_{i}(0)\right| \mathrm{e}^{-\frac{1}{C_{i} R_{i}} }+R_{i}\left(\sum_{j=1}^{n}\left|T_{i j}\right|+I_{i}\right) \\ & \leqslant\left|x_{i}(0)\right|+R_{i}\left(\sum_{j=1}^{n}\left|T_{i j}\right|+I_{i}\right),(i=1,2, \cdots, n) \end{aligned}
∣xi(t)∣⩽∣xi(0)∣e−CiRi1+Ci1∫0te−CiRi1(t−s)∣∣∣∣∣j=1∑nTijgj(xj(s))+Ii∣∣∣∣∣ds⩽∣xi(0)∣e−CiRi1+Ri(j=1∑n∣Tij∣+Ii)⩽∣xi(0)∣+Ri(j=1∑n∣Tij∣+Ii),(i=1,2,⋯,n)
构造Hopfield能量函数
E
(
x
)
=
−
1
2
∑
i
=
1
n
∑
j
=
1
n
T
i
j
v
i
v
j
−
∑
i
=
1
n
I
i
v
i
+
∑
i
=
1
n
1
R
i
∫
0
v
i
g
i
−
1
(
θ
)
d
θ
E(x)=-\frac{1}{2} \sum_{i=1}^{n} \sum_{j=1}^{n} T_{i j} v_{i} v_{j}-\sum_{i=1}^{n} I_{i} v_{i}+\sum_{i=1}^{n} \frac{1}{R_{i}} \int_{0}^{v_{i}} g_{i}^{-1}(\theta) \mathrm{d} \theta
E(x)=−21i=1∑nj=1∑nTijvivj−i=1∑nIivi+i=1∑nRi1∫0vigi−1(θ)dθ
其中
v
i
=
g
i
(
x
i
)
v_i=g_i(x_i)
vi=gi(xi). 计算函数E沿着网络(1)的轨线的导数
d
E
(
x
(
t
)
)
d
t
∣
(
1
)
=
−
1
2
∑
i
=
1
n
∑
j
=
1
n
T
i
j
(
d
v
i
(
t
)
d
t
v
j
(
t
)
+
v
i
(
t
)
d
v
j
(
t
)
d
t
)
−
∑
i
=
1
n
(
I
i
+
1
R
i
x
i
(
t
)
)
d
v
i
(
t
)
d
t
=
−
∑
i
=
1
n
d
v
i
(
t
)
d
t
(
−
1
R
i
x
i
(
t
)
+
∑
j
=
1
n
T
i
j
g
j
(
x
j
(
t
)
)
+
I
i
)
=
−
∑
i
=
1
n
C
i
d
g
i
(
x
i
(
t
)
)
d
x
i
(
t
)
(
d
x
i
(
t
)
d
t
)
2
⩽
0
\begin{aligned} \frac{\mathrm{d} E\left(x(t)\right)}{\mathrm{d} t} &\left.\right|_{(1)}=-\frac{1}{2} \sum_{i=1}^{n} \sum_{j=1}^{n} T_{i j}\left(\frac{\mathrm{d} v_{i}(t)}{\mathrm{d} t} v_{j}(t)+v_{i}(t) \frac{\mathrm{d} v_{j}(t)}{\mathrm{d} t}\right) \\ &-\sum_{i=1}^{n}\left(I_{i}+\frac{1}{R_{i}} x_{i}(t)\right) \frac{\mathrm{d} v_{i}(t)}{\mathrm{d} t} \\ =&-\sum_{i=1}^{n} \frac{\mathrm{d} v_{i}(t)}{\mathrm{d} t}\left(-\frac{1}{R_{i}} x_{i}(t)+\sum_{j=1}^{n} T_{ij} g_{j}\left(x_{j}(t)\right)+I_{i}\right) \\ =&-\sum_{i=1}^{n} C_{i} \frac{\mathrm{d} g_{i}\left(x_{i}(t)\right)}{\mathrm{d} x_{i}(t)}\left(\frac{\mathrm{d} x_{i}(t)}{\mathrm{d} t}\right)^{2} \\ \leqslant & 0 \end{aligned}
dtdE(x(t))==⩽∣(1)=−21i=1∑nj=1∑nTij(dtdvi(t)vj(t)+vi(t)dtdvj(t))−i=1∑n(Ii+Ri1xi(t))dtdvi(t)−i=1∑ndtdvi(t)(−Ri1xi(t)+j=1∑nTijgj(xj(t))+Ii)−i=1∑nCidxi(t)dgi(xi(t))(dtdxi(t))20
显然，
d
E
(
x
(
t
)
)
d
t
∣
(
1
)
=
0
,
当
且
仅
当
d
x
(
t
)
d
t
=
0
\left.\frac{dE\left(x(t)\right)}{dt}\right|_{(1)}=0,\quad 当且仅当\frac{dx(t)}{dt}=0
dtdE(x(t))∣∣∣∣(1)=0,当且仅当dtdx(t)=0
5万+
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
