ResearchTrend.AI
  • Communities
  • Connect sessions
  • AI calendar
  • Organizations
  • Contact Sales
Papers
Communities
Social Events
Terms and Conditions
Pricing
Contact Sales
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 1210.0508
205
12
v1v2v3v4v5 (latest)

Inference algorithms for pattern-based CRFs on sequence data

1 October 2012
Rustem Takhanov
V. Kolmogorov
ArXiv (abs)PDFHTML
Abstract

We consider Conditional Random Fields (CRFs) with pattern-based potentials defined on a chain. In this model the energy of a string (labeling) x1...xnx_1...x_nx1​...xn​ is the sum of terms over intervals [i,j][i,j][i,j] where each term is non-zero only if the substring xi...xjx_i...x_jxi​...xj​ equals a prespecified pattern α\alphaα. Such CRFs can be naturally applied to many sequence tagging problems. We present efficient algorithms for the three standard inference tasks in a CRF, namely computing (i) the partition function, (ii) marginals, and (iii) computing the MAP. Their complexities are respectively O(nL)O(n L)O(nL), O(nLℓmax)O(n L \ell_{max})O(nLℓmax​) and O(nLmin⁡{∣D∣,log⁡(ℓmax+1)})O(n L \min\{|D|,\log (\ell_{max}+1)\})O(nLmin{∣D∣,log(ℓmax​+1)}) where LLL is the combined length of input patterns, ℓmax\ell_{max}ℓmax​ is the maximum length of a pattern, and DDD is the input alphabet. This improves on the previous algorithms of (Ye et al., 2009) whose complexities are respectively O(nL∣D∣)O(n L |D|)O(nL∣D∣), O(n∣Γ∣L2ℓmax2)O(n |\Gamma| L^2 \ell_{max}^2)O(n∣Γ∣L2ℓmax2​) and O(nL∣D∣)O(n L |D|)O(nL∣D∣), where ∣Γ∣|\Gamma|∣Γ∣ is the number of input patterns. In addition, we give an efficient algorithm for sampling. Finally, we consider the case of non-positive weights. (Komodakis & Paragios, 2009) gave an O(nL)O(n L)O(nL) algorithm for computing the MAP. We present a modification that has the same worst-case complexity but can beat it in the best case.

View on arXiv
Comments on this paper