For the first time in 2007, TRECVID considered structured evaluation of automated video summarization, utilizing BBC rushes video. In 2007, we conducted user evaluations with the ...
This paper addresses the challenging problem of learning from multiple annotators whose labeling accuracy (reliability) differs and varies over time. We propose a framework based ...
We approached the problem of classifying papers for the TREC 2004 Genomics Track triage task as a four step process: feature generation, feature selection, classifier training, an...
Aaron M. Cohen, Ravi Teja Bhupatiraju, William R. ...
This paper investigates how the Univariate Marginal Distribution Algorithm (UMDA) behaves in non-stationary environments when engaging in sampling and selection strategies designe...
Despite the recent advances in planning with MDPs, the problem of generating good policies is still hard. This paper describes a way to generate policies in MDPs by (1) determiniz...