The Sequence Memoizer

Journal article

Probabilistic models of sequences play a central role in most machine translation, automated speech recognition, lossless compression, spell-checking, and gene identification applications to name but a few. Unfortunately, realworld sequence data often exhibit long range dependencies which can only be captured by computationally challenging, complex models. Sequence data arising from natural processes also often exhibits power-law properties, yet common sequence models do not capture such properties. The sequence memoizer is a new hierarchical Bayesian model for discrete sequence data that captures long range dependencies and power-law characteristics, while remaining computationally attractive. Its utility as a language model and general purpose lossless compressor is demonstrated. © 2011 ACM.

Authors: Wood, F; Gasthaus, J; Archambeau, C; James, L; Teh, Y

• Publication status: Published
• Journal: COMMUNICATIONS OF THE ACM
• Volume: 54
• Issue: 2
• Pages: 91-98
• Publication date: 2011-02-01
• DOI: 10.1145/1897816.1897842
• EISSN: 1557-7317
• ISSN: 0001-0782
• Language: English