Learn More
We study the meter data management systems (MDMS) with a focus on missing data handling, and propose two approaches, called Lookback-N and Sandwich-N, based on the historical data. Using a realistic dataset, we demonstrate that Lookback-N is effective for online processing, and Sandwich-N outperforms the conventional offline approach, in terms of estimation(More)
In this paper, we present an approach to active learning for video annotation. We use active learning to aide in the semantic labeling of video databases. A software library and simulator have been developed as a tool to investigate and measure the results of active learning. A list of confidence values and nearest neighbor values is maintained for each(More)
  • 1