FM公式
y
=
w
0
+
∑
i
=
1
n
w
i
x
i
+
∑
i
=
1
n
∑
i
+
1
n
<
v
i
,
v
j
>
x
i
x
j
y = w_0+\sum_{i=1}^nw_ix_i+\sum_{i=1}^{n}\sum_{i+1}^n\lt v_i,v_j\gt x_ix_j
y=w0+i=1∑nwixi+i=1∑ni+1∑n<vi,vj>xixj
化简推导
∑
i
=
1
n
−
1
∑
j
=
i
+
1
n
<
v
i
,
v
j
>
x
i
x
j
=
1
2
∑
i
=
1
n
∑
j
=
1
n
<
v
i
,
v
j
>
x
i
x
j
−
1
2
∑
i
=
1
n
<
v
i
,
v
i
>
x
i
x
i
=
1
2
(
∑
i
=
1
n
∑
j
=
1
n
∑
f
=
1
k
v
i
,
f
v
j
,
f
x
i
x
j
−
∑
i
=
1
n
∑
f
=
1
k
v
i
,
f
v
i
,
f
x
i
x
i
)
=
1
2
∑
f
=
1
k
[
(
∑
i
=
1
n
v
i
,
f
x
i
)
⋅
(
∑
j
=
1
n
v
j
,
f
x
j
)
−
∑
i
=
1
n
v
i
,
f
2
x
i
2
]
=
1
2
∑
f
=
1
k
[
(
∑
i
=
1
n
v
i
,
f
x
i
)
2
−
∑
i
=
1
n
v
i
,
f
2
x
i
2
]
\begin{aligned} \sum_{i=1}^{n-1}{\sum_{j=i+1}^{n}{<v_i,v_j>x_ix_j}} = &\frac{1}{2}\sum_{i=1}^{n}{\sum_{j=1}^{n}{<v_i,v_j>x_ix_j}} - \frac{1}{2} {\sum_{i=1}^{n}{<v_i,v_i>x_ix_i}} \\ =& \frac{1}{2} \left( \sum_{i=1}^{n}{\sum_{j=1}^{n}{\sum_{f=1}^{k}{v_{i,f}v_{j,f}x_ix_j}}} - \sum_{i=1}^{n}{\sum_{f=1}^{k}{v_{i,f}v_{i,f}x_ix_i}} \right) \\ =& \frac{1}{2}\sum_{f=1}^{k}{\left[ \left( \sum_{i=1}^{n}{v_{i,f}x_i} \right) \cdot \left( \sum_{j=1}^{n}{v_{j,f}x_j} \right) - \sum_{i=1}^{n}{v_{i,f}^2 x_i^2} \right]} \\ =& \frac{1}{2}\sum_{f=1}^{k}{\left[ \left( \sum_{i=1}^{n}{v_{i,f}x_i} \right)^2 - \sum_{i=1}^{n}{v_{i,f}^2 x_i^2} \right]} \end{aligned}
i=1∑n−1j=i+1∑n<vi,vj>xixj====21i=1∑nj=1∑n<vi,vj>xixj−21i=1∑n<vi,vi>xixi21⎝⎛i=1∑nj=1∑nf=1∑kvi,fvj,fxixj−i=1∑nf=1∑kvi,fvi,fxixi⎠⎞21f=1∑k[(i=1∑nvi,fxi)⋅(j=1∑nvj,fxj)−i=1∑nvi,f2xi2]21f=1∑k⎣⎡(i=1∑nvi,fxi)2−i=1∑nvi,f2xi2⎦⎤
PNN
内积公式
l
p
n
=
W
p
n
⨀
p
=
∑
i
=
1
N
∑
j
=
1
N
(
W
p
n
)
i
,
j
p
i
,
j
=
∑
i
=
1
N
∑
j
=
1
N
(
W
p
n
)
i
,
j
⟨
f
i
,
f
j
⟩
\begin{aligned} l_p^n=W_p^n \bigodot p=\sum_{i=1}^N\sum_{j=1}^N(W_p^n)_{i,j}p_{i,j}= \sum_{i=1}^N\sum_{j=1}^N(W_p^n)_{i,j}\langle f_i,f_j \rangle \end{aligned}
lpn=Wpn⨀p=i=1∑Nj=1∑N(Wpn)i,jpi,j=i=1∑Nj=1∑N(Wpn)i,j⟨fi,fj⟩
由于W是对称矩阵，可以简化推导
l
p
n
=
W
p
n
⨀
p
=
∑
i
=
1
N
∑
j
=
1
N
(
W
p
n
)
i
,
j
⟨
f
i
,
f
j
⟩
=
∑
i
=
1
N
∑
j
=
1
N
θ
i
n
θ
j
n
⟨
f
i
,
f
j
⟩
=
∑
i
=
1
N
∑
j
=
1
N
⟨
θ
i
n
f
i
,
θ
j
n
f
j
⟩
=
⟨
∑
i
=
1
N
θ
i
n
f
i
,
∑
j
=
1
N
θ
j
n
f
j
⟩
=
⟨
∑
i
=
1
N
δ
i
n
,
∑
j
=
1
N
δ
j
n
⟩
=
∥
∑
i
=
1
N
δ
i
n
∥
2
\begin{aligned} l_p^n ={} & W_p^n \bigodot p =\sum_{i=1}^N\sum_{j=1}^N(W_p^n)_{i,j}\langle f_i,f_j \rangle \\ ={} & \sum_{i=1}^N\sum_{j=1}^N \theta_i^n \theta_j^n \langle f_i,f_j \rangle \\ ={} &\sum_{i=1}^N\sum_{j=1}^N \langle \theta_i^nf_i, \theta_j^nf_j \rangle \\ ={} & \langle \sum_{i=1}^N\theta_i^nf_i, \sum_{j=1}^N\theta_j^nf_j \rangle \\ ={} & \langle \sum_{i=1}^N\delta_i^n, \sum_{j=1}^N\delta_j^n \rangle \\ ={} & \Vert \sum_{i=1}^N\delta_i^n \Vert^2 \\ \end{aligned}
lpn======Wpn⨀p=i=1∑Nj=1∑N(Wpn)i,j⟨fi,fj⟩i=1∑Nj=1∑Nθinθjn⟨fi,fj⟩i=1∑Nj=1∑N⟨θinfi,θjnfj⟩⟨i=1∑Nθinfi,j=1∑Nθjnfj⟩⟨i=1∑Nδin,j=1∑Nδjn⟩∥i=1∑Nδin∥2
4978
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
