We propose a distribution-based pruning of n-gram backoff language models. Instead of the conventional approach of pruning n-grams that are infrequent in training data, we prune n...
Current models of story comprehension have three major deficiencies: (1) lack of experimental support for the inference processes they involve (e.g. reliance on prediction); (2) i...
This work presents a model of a distributed database system which provides the framework to study the performance of timestamp ordering concurrency control. We exhibit an analytic...
The processing of Japanese text is complicated by the fact that there are no word delimiters. To segment Japanese text, systems typically use knowledge-based methods and large lex...
In this paper, a new tracking algorithm, based on local minimum energy, is proposed for matching between a projected model and corresponding image features in real-time. The algor...