Is it possible to chose the parameters of a RBM to maximize the likelihood of the observed data?

(I follow the notation of the deeplearning tutorial ). Denote the observable data by $x$, hidden data by $h$, the energy function by $E$ and the normalizing constant by $Z$. The probability of $x$ is: \begin{equation} P(x) = \sum_h P(x,h) = \sum_h \frac{e^{-E(x,h)}}{Z}. \end{equation} In order to maximize the probability of $x$ conditional on the parameters of the model $\theta$ one could maximize the sum of the derivatives of the log-likelihood: \begin{equation} \sum_{x_i} \frac{\partial \log p(x_i)}{\partial \theta} \end{equation} There exists many different ways to approximate the derivative of the log-likelihood to facilitate (permit) its computation. For example, Contrastive Divergence and Persistent Contrastive Divergence are used often. I wonder whether it makes sense to estimate the parameters $\theta$ recursively until convergence while continuing to approximate the derivative of the log-likelihood. One could update the parameters after seeing each data point $x_i$ as: \begin{equation} \theta_{i+1} = \theta_{i} - \eta_i \frac{\partial \log p(x_i)}{\partial \theta_i} \end{equation} The practice I learned in Hinton et al. (2006) and in Tieleman (2008) (I'm sorry: I cannot link to that paper due to insufficient reputation) is different though: Both papers define a number of fixed iterations a priori. Could somebody kindly educate me why recursively updating the parameters until convergence is not a good idea? In particular, I'm interested whether there's a theoretical flaw in my reasoning or whether computational capacities dictate sticking to a fixed number of iterations. I am grateful for any help!


Your Answer

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

Browse other questions tagged or ask your own question.