Contrastive Divergence (CD) is a popular method for estimating the parameters of Markov Random Fields (MRFs) by efficiently approximating an intractable term in the gradient of the MRF’s log probability. Despite its empirical success, basic theoretical questions on its convergence properties are currently open. In this paper, we analyze the CD1 update rule for Restricted Boltzmann Machines (RBM...

In this study we show that constituency is of limited importance for a proper treatment of the interaction between the linear position of a wa-marked nominal in a Japanese sentence and possible domains of contrastive focus, and that constraints concerning contrastive focus should be represented in terms of linear order and not constituency. Linearisation HPSG, where linear order is independent ...

Pseudo-likelihood and contrastive divergence are two well-known examples of contrastive methods. These algorithms trade off the probability of the correct label with the probabilities of other “nearby” instantiations. In this paper we explore more general types of contrastive objectives, which trade off the probability of the correct label against an arbitrary set of other instantiations. We pr...

The strong version of the Contrastive Analysis Hypothesis rested on the ‘difference equals difficulty principle.’ That is to say, learner errors can be predicted on the basis of differences between their mother tongue (L1) and the target language (L2). Although this strong version has been abandoned, research in error analysis and more recently, computer-aided error analysis, has demonstrated t...

A tool developed for computer-aided training of spoken language is presented in the paper. The tool’s environment utilizes both visual and auditory feedback information to help a user in learning pronunciation and intonation in L1 or L2. The learning is supported by displaying the user’s speech and its relevant parameters (volume, F0 and spectrum) in parallel with multiple reference templates. ...

This paper reports on a comparative analysis of accentuation strategies within Italian and Dutch NPs. Accent-patterns were obtained in a (semi-)spontaneous way via a simple dialogue-game with 8 Dutch speakers and 4 Italian ones. In this way, target descriptions of all speakers were obtained in the following four contexts: all new, single contrast in the adjective, single contrast in the noun, a...