Derivation of User Browsing Model

最新推荐文章于 2023-06-13 02:11:39 发布

Mikesong2csdn

最新推荐文章于 2023-06-13 02:11:39 发布

阅读量947

点赞数

分类专栏：模型推导文章标签： UBM EM

本文链接：https://blog.csdn.net/song0601013ndsc/article/details/47666515

版权

模型推导专栏收录该内容

1 篇文章 0 订阅

订阅专栏

Mathematicians hate words like “trivial, traditional, … “

- Prelude
- Derivation in detail
  - E-step
  - M-step

Prelude

UBM1is a simple but ecient model in estimating position bias. The model
structures as following:

Figure 1 UBM: Multiple Browsing

The complete-data likelihood:

P (c, a, e, m | q, u, r, d, Θ) = P (c | a, e) P (a | q, u) P (e | r, d, m) P (m | q)

The incomplete-data likelihood:

P (c | q, u, r, d, Θ) = \sum a, e P (c | a, e) P (a | q, u) \sum m P (e | r, d, m) P (m | q)

$P(c | q, u, r, d, \Theta) = \sum_{a,e}P( c | a, e)P(a | q, u)\sum_{m}P(e | r, d, m)P(m | q)$

The conditional distribution of $a, e$ is Bernoulli:

P (a | q, u) = {α u q 1 - α u q i f a = 1 i f a = 0

$P(a | q, u) = \begin{cases} \alpha_{uq} & \quad \mathrm{if} \quad a = 1 \\ 1-\alpha_{uq} & \quad \mathrm{if} \quad a = 0 \end{cases}$

P (e | r, d, m) = {γ r d m 1 - γ r d m i f e = 1 i f e = 0

$P(e | r, d, m) = \begin{cases} \gamma_{rdm} & \quad \mathrm{if} \quad e = 1 \\ 1-\gamma_{rdm} & \quad \mathrm{if} \quad e = 0 \\ \end{cases}$

With deterministic assumption:

c = 1 ⟺ a = e = 1

$c = 1 \iff a=e=1$

The complete-data likelihood becomes:

P (c, a, e, m | q, u, r, d, Θ) = ⎧ ⎩ ⎨ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ α u q γ r d m μ m q α u q (1 - γ r d m) μ m q (1 - α u q) γ r d m μ m q (1 - α u q) (1 - γ r d m) μ m q, i f c = 1, i f a = 1 & e = 0, i f a = 0 & e = 1, i f a = 0 & e = 0

$P(c, a, e, m | q, u, r, d, \Theta) = \begin{cases} \alpha_{uq} \gamma_{rdm} \mu_{mq} & ,\quad \mathrm{if} \quad c = 1 \\ \alpha_{uq} (1- \gamma_{rdm}) \mu_{mq} & ,\quad \mathrm{if} \quad a = 1 \quad \& \quad e = 0 \\ ( 1 - \alpha_{uq} ) \gamma_{rdm}\mu_{mq} & ,\quad \mathrm{if} \quad a = 0 \quad \& \quad e = 1 \\ ( 1 - \alpha_{uq} ) ( 1 - \gamma_{rdm} ) \mu_{mq} & ,\quad \mathrm{if} \quad a = 0 \quad \& \quad e = 0 \\ \end{cases}$

The incomplete-data likelihood becomes:

P (c | q, u, r, d, Θ) = {α u q \sum m γ r d m μ m q 1 - α u q \sum m γ r d m μ m q, i f c = 1, i f c = 0

$P(c | q, u, r, d, \Theta) = \begin{cases} \alpha_{uq}\sum_{m}\gamma_{rdm}\mu_{mq} & ,\quad \mathrm{if} \quad c = 1 \\ 1 - \alpha_{uq}\sum_{m}\gamma_{rdm}\mu_{mq} & ,\quad \mathrm{if} \quad c = 0 \\ \end{cases}$

The log-likelihood function becomes:

log L (Θ) = = = \sum all record log P (c | q, u, r, d, Θ) \sum u, q \sum r, d {S ∙ u q r d log P (c = 1 | q, u, r, d, Θ) + S \circ u q r d log P (c = 0 | q, u, r, d, Θ)} \sum u, q \sum r, d {S ∙ u q r d log [α u q \sum m γ r d m μ m q] + S \circ u q r d log [(1 - α u q \sum m γ r d m μ m q)]}

$\begin{eqnarray*} \log L(\Theta) & = & \sum_{\text{all record}} \log P(c | q, u, r, d, \Theta) \\ \quad & = & \sum_{u,q} \sum_{r, d} \bigg \{ S_{uqrd}^{\bullet} \log P(c = 1 | q, u, r, d, \Theta) + S_{uqrd}^{\circ} \log P(c = 0 | q, u, r, d, \Theta) \bigg \} \\ \quad & = & \sum_{u,q} \sum_{r, d} \bigg \{ S_{uqrd}^{\bullet} \log \left [ \alpha_{uq}\sum_{m}\gamma_{rdm}\mu_{mq} \right ] + S_{uqrd}^{\circ} \log \left [ (1 - \alpha_{uq}\sum_{m}\gamma_{rdm}\mu_{mq}) \right ] \bigg \} \end{eqnarray*}$

The so-called deterministic relationship $c \leftrightarrow (a ,e)$ and upon formula of log-likelihood are so misleading that at first I omited $a , e$ as latent variables. Despite all kinds of tricks tried in M-step, I failed to deduce the iteration formulas appended in the paper.

Derivation in detail

E-step

Posterior distributions of latent variables after $t$ -th iteration is:

Q t (a, e, m | c, u, q, r, d, Θ t) = P ( c , a , e , m | u , q , r , d , Θ t ) \sum a , e , m P ( c , a , e , m | u , q , r , d , Θ t )

$Q^t (a, e, m | c, u, q, r, d, \Theta^t) = \frac{ P(c,a,e,m | u, q, r, d, \Theta^t ) }{\sum_{a, e,m} P(c,a,e,m | u, q, r, d, \Theta^t ) }$

, thus

Q t (a = 1, e = 1, m | c = 1, u, q, r, d, Θ t) = γ r d m μ m q \sum m γ r d m μ m q

$Q^t (a=1, e=1, m | c=1, u,q,r,d,\Theta^t ) = \frac{ \gamma_{rdm} \mu_{mq} }{ \sum_{m} \gamma_{rdm} \mu_{mq} }$

Q t (a = 1, e = 0, m | c = 0, u, q, r, d, Θ t) = α t u q ( 1 - γ t r d m ) μ t m q 1 - α t u q \sum m γ t r d m μ t m q

$Q^t (a=1, e=0, m | c=0, u,q,r,d,\Theta^t ) = \frac{ \alpha_{uq}^t ( 1 - \gamma_{rdm}^t ) \mu_{mq}^t }{ 1 - \alpha_{uq}^t \sum_{m} \gamma_{rdm}^t \mu_{mq}^t}$

Q t (a = 0, e = 1, m | c = 0, u, q, r, d, Θ t) = ( 1 - α t u q ) γ t r d m μ t m q 1 - α t u q \sum m γ t r d m μ t m q

$Q^t (a=0, e=1, m | c=0, u,q,r,d,\Theta^t ) = \frac{ (1 - \alpha_{uq}^t ) \gamma_{rdm}^t \mu_{mq}^t }{ 1 - \alpha_{uq}^t \sum_{m} \gamma_{rdm}^t \mu_{mq}^t}$

Q t (a = 0, e = 0, m | c = 0, u, q, r, d, Θ t) = ( 1 - α t u q ) ( 1 - γ t r d m ) μ t m q 1 - α t u q \sum m γ t r d m μ t m q

$Q^t (a=0, e=0, m | c=0, u,q,r,d,\Theta^t ) = \frac{ (1 - \alpha_{uq}^t ) ( 1 - \gamma_{rdm}^t ) \mu_{mq}^t }{ 1 - \alpha_{uq}^t \sum_{m} \gamma_{rdm}^t \mu_{mq}^t}$

M-step

The free energy with respect to $Q^t$ is:

F (Q t, Θ) = = + = + + \sum o b s \sum a, e, m Q t (a, e, m | o b s) log P (c, a, e, m | u, q, r, d, Θ) \sum u, q \sum r, d {S ∙ u q r d \sum m Q t (a = 1, e = 1, m | c = 1, r, d, u, q) log (α u q γ r d m μ m q) S \circ u q r d [\sum m Q t (a = 1, e = 0, m | c = 0, r, d, u, q) log (α u q (1 - γ r d m) μ m q) + \sum m Q t (a = 0, e = 1, m | c = 0, r, d, u, q) log ((1 - α u q) γ r d m μ m q) + \sum m Q t (a = 0, e = 0, m | c = 0, r, d, u, q) log ((1 - α u q) (1 - γ r d m) μ m q)]} \sum u, q \sum r, d [(S ∙ u q r d + S \circ u q r d Q t (a = 1, e = 0 | c = 0, r, d, u, q)) log α u q + S \circ u q r d Q t (a = 0 | c = 0, r, d, u, q) log (1 - α u q)] \sum r, d, m {[\sum u, q (S ∙ u q r d Q t (a = 1, e = 1, m | c = 1, r, d, u, q) + S \circ u q r d Q t (a = 0, e = 1, m | c = 0, r, d, u, q))] log γ r d m + [\sum u, q S \circ u q r d Q t (e = 0, m | c = 0, r, d, u, q)] log (1 - γ r d m)} \sum m, q {\sum u, r, d [S ∙ u q r d Q t (m | c = 1, r, d, u, q) + S \circ u q r d Q t (m | c = 0, r, d, u, q)]} log μ m q

$\begin{eqnarray*} F(Q^t , \Theta) & = & \sum_{obs} \sum_{a,e, m} Q^t (a,e,m | obs) \log P(c, a, e, m | u, q, r, d, \Theta ) \\ \quad & = & \sum_{u, q} \sum_{r, d} \bigg \{ S_{uqrd}^{\bullet} \sum_{m} Q^t (a=1,e=1, m | c=1, r,d,u,q) \log (\alpha_{uq} \gamma_{rdm} \mu_{mq}) \\ \quad & + & S_{uqrd}^{\circ} \Big [ \sum_{m} Q^t (a=1,e=0, m | c=0, r,d,u,q) \log \big ( \alpha_{uq} (1 - \gamma_{rdm}) \mu_{mq} \big ) \\ \quad & \quad & + \sum_{m} Q^t (a=0,e=1, m | c=0, r,d,u,q) \log \big( (1- \alpha_{uq} ) \gamma_{rdm} \mu_{mq} \big ) \\ \quad & \quad & + \sum_{m} Q^t (a=0,e=0, m | c=0, r,d,u,q) \log \big( (1- \alpha_{uq} ) (1 - \gamma_{rdm} ) \mu_{mq} \big ) \Big ] \bigg \} \\ & = & \sum_{u, q} \sum_{r, d} \bigg [ \big ( S_{uqrd}^{\bullet} + S_{uqrd}^{\circ} Q^t (a=1, e=0 | c=0, r,d,u,q) \big ) \log \alpha_{uq} \\ \quad & \quad & \qquad \qquad + S_{uqrd}^{\circ} Q^t (a=0 | c=0, r,d,u,q) \log (1- \alpha_{uq}) \bigg ] \\ \quad & + & \sum_{r, d, m} \bigg \{ \Big [ \sum_{u, q} \big ( S_{uqrd}^{\bullet} Q^t (a=1,e=1, m | c=1, r,d,u,q) \\ \quad & \quad & \qquad \qquad + S_{uqrd}^{\circ} Q^t (a=0,e=1, m | c=0, r,d,u,q) \big ) \Big ] \log \gamma_{rdm} \\ \quad & \quad & \qquad + \Big [ \sum_{u, q} S_{uqrd}^{\circ } Q^t (e=0, m | c=0, r,d,u,q) \Big ] \log ( 1 - \gamma_{rdm} ) \bigg \} \\ \quad & + & \sum_{m, q} \bigg \{ \sum_{u, r, d} \Big [ S_{uqrd}^{\bullet} Q^t (m | c=1, r,d, u,q) + S_{uqrd}^{\circ} Q^t (m | c=0, r,d, u,q) \Big ] \bigg \} \log \mu_{mq} \end{eqnarray*}$

Maximization over $\alpha_{uq}, \gamma_{rdm}, \mu_{mq}$ with contraint $\sum_{m} \mu_{mq} = 1$ leads to the updating formulas:

for $\alpha_{uq}$ ,

$α t + 1 u q = = = \sum r , d ( S ∙ u q r d + S \circ u q r d Q t ( a = 1 , e = 0 | c = 0 , r , d , u , q ) ) \sum r , d [ S ∙ u q r d + S \circ u q r d ( Q t ( a = 1 , e = 0 | c = 0 , r , d , u , q ) + Q t ( a = 0 | c = 0 , r , d , u , q ) ) ] \sum r , d ( S ∙ u q r d + S \circ u q r d Q t ( a = 1 , e = 0 | c = 0 , r , d , u , q ) ) \sum r , d ( S ∙ u q r d + S \circ u q r d ) 1 S u q (\sum r d S \circ u q r d α t u q ( 1 - \sum m γ t r d m μ t m q ) 1 - α t u q \sum m γ t r d m μ t m q + S ∙ u q)$ $\begin{eqnarray*} \alpha_{uq}^{t+1} & = & \frac{ \sum_{r, d} \big ( S_{uqrd}^{\bullet} + S_{uqrd}^{\circ} Q^t (a=1, e=0 | c=0, r,d,u,q) \big ) }{ \sum_{r, d} \Big [ S_{uqrd}^{\bullet} + S_{uqrd}^{\circ} \big ( Q^t (a=1, e=0 | c=0, r,d,u,q) + Q^t (a=0 | c=0, r,d,u,q) \big ) \Big ] } \\ \quad & = & \frac{ \sum_{r, d} \big ( S_{uqrd}^{\bullet} + S_{uqrd}^{\circ} Q^t (a=1, e=0 | c=0, r,d,u,q) \big ) }{ \sum_{r, d} \big ( S_{uqrd}^{\bullet} + S_{uqrd}^{\circ} \big ) } \\ \quad & = & \frac{1}{S_{uq}} \Big ( \sum_{rd} S_{uqrd}^{\circ} \frac{ \alpha_{uq}^t ( 1 - \sum_{m} \gamma_{rdm}^t \mu_{mq}^t ) }{ 1 - \alpha_{uq}^t \sum_{m} \gamma_{rdm}^t \mu_{mq}^t} + S_{uq}^{\bullet} \Big ) \end{eqnarray*}$
for $\gamma_{rdm}$ ,

$γ t + 1 r d m = \equiv \sum u , q ( S ∙ u q r d Q t ( m | c = 1 , r , d , u , q ) + S \circ u q r d Q t ( e = 1 , m | c = 0 , r , d , u , q ) ) \sum u , q ( S ∙ u q r d Q t ( m | c = 1 , r , d , u , q ) + S \circ u q r d Q t ( m | c = 0 , r , d , u , q ) ) A / B,$ $\begin{eqnarray*} \gamma_{rdm}^{t+1} & = & \frac{ \sum_{u,q} \left ( S_{uqrd}^{\bullet} Q^t (m | c=1, r,d,u,q) + S_{uqrd}^{\circ} Q^t (e=1, m | c=0, r,d,u,q) \right ) }{ \sum_{u,q} \left ( S_{uqrd}^{\bullet} Q^t (m | c=1, r,d,u,q) + S_{uqrd}^{\circ} Q^t ( m | c=0, r,d,u,q) \right ) } \\ \quad & \quad \equiv \quad & A / B , \end{eqnarray*}$
in which,
$A B = = = = \sum u, q (S ∙ u q r d Q t (m | c = 1, r, d, u, q) + S \circ u q r d Q t (e = 1, m | c = 0, r, d, u, q)) \sum u, q (S \circ u q r d ( 1 - α t u q ) γ t r d m μ t m q 1 - α t u q \sum m γ t r d m μ t m q + S ∙ u q r d γ t r d m μ t m q \sum m γ t r d m μ t m q) \sum u, q (S ∙ u q r d Q t (m | c = 1, r, d, u, q) + S \circ u q r d Q t (m | c = 0, r, d, u, q)) \sum u, q (S \circ u q r d ( 1 - α t u q γ t r d m ) μ t m q 1 - α t u q \sum m γ t r d m μ t m q + S ∙ u q r d γ t r d m μ t m q \sum m γ t r d m μ t m q)$ $\begin{eqnarray*} A & = & \sum_{u,q} \Big ( S_{uqrd}^{\bullet} Q^t (m | c=1, r,d,u,q) + S_{uqrd}^{\circ} Q^t (e=1, m | c=0, r,d,u,q) \Big ) \\ & = & \sum_{u,q} \Big ( S_{uqrd}^{\circ} \frac{ (1 - \alpha_{uq}^t ) \gamma_{rdm}^t \mu_{mq}^t }{1 - \alpha_{uq}^t \sum_{m} \gamma_{rdm}^t \mu_{mq}^t} + S_{uqrd}^{\bullet} \frac{ \gamma_{rdm}^t \mu_{mq}^t }{ \sum_{m} \gamma_{rdm}^t \mu_{mq}^t} \Big ) \\ B & = & \sum_{u,q} \Big ( S_{uqrd}^{\bullet} Q^t (m | c=1, r,d,u,q) + S_{uqrd}^{\circ} Q^t ( m | c=0, r,d,u,q) \Big ) \\ & = & \sum_{u,q} \Big ( S_{uqrd}^{\circ} \frac{ (1 - \alpha_{uq}^t \gamma_{rdm}^t ) \mu_{mq}^t }{1 - \alpha_{uq}^t \sum_{m} \gamma_{rdm}^t \mu_{mq}^t} + S_{uqrd}^{\bullet} \frac{ \gamma_{rdm}^t \mu_{mq}^t }{ \sum_{m} \gamma_{rdm}^t \mu_{mq}^t} \Big ) \\ \end{eqnarray*}$
for $\mu_{mq}$ ,
$μ m q = λ q \sum u, r, d [S ∙ u q r d Q t (m | c = 1, r, d, u, q) + S \circ u q r d Q t (m | c = 0, r, d, u, q)] .$ $\mu_{mq} = \lambda_{q} \sum_{u, r, d} \Big [ S_{uqrd}^{\bullet} Q^t (m | c=1, r,d, u,q) + S_{uqrd}^{\circ} Q^t (m | c=0, r,d, u,q) \Big ] .$
$\sum_{m} \mu_{mq} = 1 \implies \lambda_{q} S_{q} = 1 \implies \lambda_{q} = \frac{1}{S_{q}}$ , thus
$μ m q = = 1 S q \sum u, r, d [S \circ u q r d ( 1 - α t u q γ t r d m ) μ t m q 1 - α t u q \sum m γ t r d m μ t m q + S ∙ u q r d γ t r d m μ t m q \sum m γ t r d m μ t m q] μ t m q S q \sum u, r, d [S \circ u q r d 1 - α t u q γ t r d m 1 - α t u q \sum m γ t r d m μ t m q + S ∙ u q r d γ t r d m \sum m γ t r d m μ t m q]$ $\begin{eqnarray*} \mu_{mq} & = & \frac{1}{S_{q}} \sum_{u,r,d} \Big [ S_{uqrd}^{\circ} \frac{ (1 - \alpha_{uq}^t \gamma_{rdm}^t ) \mu_{mq}^t }{1 - \alpha_{uq}^t \sum_{m} \gamma_{rdm}^t \mu_{mq}^t} + S_{uqrd}^{\bullet} \frac{ \gamma_{rdm}^t \mu_{mq}^t }{ \sum_{m} \gamma_{rdm}^t \mu_{mq}^t} \Big ] \\ & = & \frac{ \mu_{mq}^t }{S_{q}} \sum_{u,r,d} \Big [ S_{uqrd}^{\circ} \frac{ 1 - \alpha_{uq}^t \gamma_{rdm}^t }{1 - \alpha_{uq}^t \sum_{m} \gamma_{rdm}^t \mu_{mq}^t} + S_{uqrd}^{\bullet} \frac{ \gamma_{rdm}^t }{ \sum_{m} \gamma_{rdm}^t \mu_{mq}^t} \Big ] \end{eqnarray*}$

G. Dupret and B. Piwowarski, A user browsing model to predict search engine click data from past observations, In ACM SIGIR Conference, 2008. ↩

Mikesong2csdn

关注

0
点赞
踩
0

收藏

觉得还不错? 一键收藏
0
评论
Derivation of User Browsing Model

Derivation of User Browsing ModelDerivation of User Browsing ModelPreludeDerivation in detailE-stepM-stepRemarkPreludeUBM1is a simple but ecient model in estimating position bias. The model stru
复制链接

扫一扫

专栏目录