Uploaded image for project: 'Machine Learning Library'
  1. Machine Learning Library
  2. ML-289

Add MCMC Topic Estimation Method

    XMLWordPrintable

    Details

    • Type: Improvement
    • Status: Resolved
    • Priority: Not specified
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: 6.4.6
    • Component/s: None

      Description

      Need to speed up Topic Estimation. The current variational inference method is O(K*W) (where K is the number of topics and W is the the number of word occurrences in the corpus per iteration and typically fewer than 200 iterations are required for convergence.

      A Markov Chain Monte Carlo method using a collapsed Gibbs sampler is O(W) per iteration, with several thousand iterations required.

      The MCMC method will be much faster for large numbers of topics.

        Attachments

          Activity

            People

            • Assignee:
              johnholt John Holt
              Reporter:
              johnholt John Holt
            • Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: