2
$\begingroup$

I was reading Goodfellow, Bengio, and Courville's Deep Learning and in Chapter 20 there is a derivation of the conditional probability $P(h_j=1|\boldsymbol{v})$, the conditional probability of the $j^{th}$ hidden unit being on given a data vector $\boldsymbol{v}$:

Note that $Z'$ is a normalizing constant, and $\tilde{P}(E)$ is the unnormalized probability of event $E$.

The derivation of 20.7 through 20.11 is clear to me, but for equation 20.13, we assume the inside of our product, $\exp{\{c_j h_j + \boldsymbol{v}^\top \boldsymbol{W}_{:,j} \boldsymbol{h}_j\}}$, is exactly equal to our (unnormalized) unit probability $\tilde{P}(h_j|\boldsymbol{v})$.

(This is described in the paragraph between 20.11 and 20.12, and we evaluate this $\tilde{P}(h_j|\boldsymbol{v})$ for $h_j = 0$ and $h_j = 1$ to get equation 20.13.)


My question is, why can we make this assumption?

The argument the paper seems to make is that because $P(\boldsymbol{h}|\boldsymbol{v})$ takes on the form $\frac{1}{Z'} \prod_{j=1}^{n_h} f(h_j)$, we can safely assume $f(h_j)$ must be our probability for each unit, $\tilde{P}(h_j|\boldsymbol{v})$.

I understand that $P(\boldsymbol{h}|\boldsymbol{v})$ must equal $\frac{1}{Z'} \prod_{j=1}^{n_h} \tilde{P}(h_j|\boldsymbol{v})$, but if we are given the equation $P(\boldsymbol{h}|\boldsymbol{v}) = \frac{1}{Z'} \prod_{j=1}^{n_h} f(h_j)$, couldn't there be several functions $f$ that fit this criterion? $f$ wouldn't necessarily have to be $\tilde{P}(h_j|\boldsymbol{v})$, would it?

$\endgroup$
2
+50
$\begingroup$

20.7-11 give us first, $$ p(\mathbf{h}|\mathbf{v}) \propto \prod_j\exp{\{c_j h_j + \boldsymbol{v}^\top \boldsymbol{W}_{:,j} h_j\}}. $$ After summing out any unwanted variables, we get for any $i$, \begin{align*} p(h_i | \mathbf{v}) &\propto \sum_{\mathbf{h} : \mathbf{h_i} = h_i} \prod_j\exp{\{c_j h_j + \boldsymbol{v}^\top \boldsymbol{W}_{:,j} h_j\}} \\ &= \sum_{\mathbf{h} : \mathbf{h_i} = h_i} \left[ \prod_{j \neq i} \exp{\{c_j h_j + \boldsymbol{v}^\top \boldsymbol{W}_{:,j} h_j\}} \exp{\{c_i h_i + \boldsymbol{v}^\top \boldsymbol{W}_{:,i} h_i\}} \right] \\ &= \exp{\{c_i h_i + \boldsymbol{v}^\top \boldsymbol{W}_{:,i} h_i\}} \sum_{\mathbf{h} : \mathbf{h_i} = h_i} \left[ \prod_{j \neq i} \exp{\{c_j h_j + \boldsymbol{v}^\top \boldsymbol{W}_{:,j} h_j\}} \right] \\ &\propto \exp{\{c_i h_i + \boldsymbol{v}^\top \boldsymbol{W}_{:,i} h_i\}}, \end{align*} assuming all the sums we take are finite.

You ask if $\exp{\{c_i h_i + \boldsymbol{v}^\top \boldsymbol{W}_{:,i} h_i\}}$ constitutes an unnormalized probability distribution. The answer is yes, and these have a special name. Your link calls them "Energy Based Models" (same as exponential family models?). It is easy to see that

$$ p(h_i|\mathbf{v}) = \frac{ \exp{\{c_i h_i + \boldsymbol{v}^\top \boldsymbol{W}_{:,i} h_i\}} }{ \sum_h \exp{\{c_i h + \boldsymbol{v}^\top \boldsymbol{W}_{:,i} h\}} } $$ because $0 < \sum_h \exp{\{c_i h + \boldsymbol{v}^\top \boldsymbol{W}_{:,i} h\}} < \infty$.

$\endgroup$
  • $\begingroup$ What happens if more than one $i$ satisfies $f(h_i) \neq \tilde{P}(h_i|\boldsymbol{v})$? $\endgroup$ – Chandler Watson Feb 6 '17 at 0:12
  • $\begingroup$ @AliasUser I'm going to change this answer a lot. Give me one moment $\endgroup$ – Taylor Feb 6 '17 at 0:45
  • $\begingroup$ @AliasUser ok there. Basically it doesn't matter if you call $\tilde{P}(h_j|\mathbf{v})$ an unnormalized probability or not. The steps to get the marginal are still the same. $\endgroup$ – Taylor Feb 6 '17 at 1:08
  • $\begingroup$ Thank you for updating the answer -- it's incredibly well written. I follow all of it except the "summing out" part. Does the first equation directly imply the second? $\endgroup$ – Chandler Watson Feb 6 '17 at 1:11
  • $\begingroup$ @AliasUser I added some more steps. How's that $\endgroup$ – Taylor Feb 6 '17 at 1:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.