We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.AP

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Applications

Title: $\ell_1$ Adaptive Trend Filter via Fast Coordinate Descent

Abstract: Identifying the unknown underlying trend of a given noisy signal is extremely useful for a wide range of applications. The number of potential trends might be exponential, which can be computationally exhaustive even for short signals. Another challenge, is the presence of abrupt changes and outliers at unknown times which impart resourceful information regarding the signal's characteristics. In this paper, we present the $\ell_1$ Adaptive Trend Filter, which can consistently identify the components in the underlying trend and multiple level-shifts, even in the presence of outliers. Additionally, an enhanced coordinate descent algorithm which exploit the filter design is presented. Some implementation details are discussed and a version in the Julia language is presented along with two distinct applications to illustrate the filter's potential.
Comments: 9 pages, 2 figures
Subjects: Applications (stat.AP); Optimization and Control (math.OC); Machine Learning (stat.ML)
Cite as: arXiv:1603.03799 [stat.AP]
  (or arXiv:1603.03799v2 [stat.AP] for this version)

Submission history

From: Gustavo C. Amaral Dr. [view email]
[v1] Fri, 11 Mar 2016 21:58:02 GMT (158kb,D)
[v2] Mon, 16 May 2016 19:03:24 GMT (159kb,D)

Link back to: arXiv, form interface, contact.