About: In order to adapt random forests to the dynamic nature of data streams, the state-of-the-art technique discards trained trees and grows new trees when concept drifts are detected. This is particularly wasteful when recurrent patterns exist. In this work, we introduce a novel framework called PEARL, which uses both an exact technique and a probabilistic graphical model with Lossy Counting, to replace drifted trees with relevant trees built from the past. The exact technique utilizes pattern matching to find the set of drifted trees, that co-occurred in predictions in the past. Meanwhile, a probabilistic graphical model is being built to capture the tree replacements among recurrent concept drifts. Once the graphical model becomes stable, it replaces the exact technique and finds relevant trees in a probabilistic fashion. Further, Lossy Counting is applied to the graphical model which brings an added theoretical guarantee for both error rate and space complexity. We empirically show our technique outperforms baselines in terms of cumulative accuracy on both synthetic and real-world datasets.   Goto Sponge  NotDistinct  Permalink

An Entity of Type : fabio:Abstract, within Data Space : covidontheweb.inria.fr associated with source document(s)

AttributesValues
type
value
  • In order to adapt random forests to the dynamic nature of data streams, the state-of-the-art technique discards trained trees and grows new trees when concept drifts are detected. This is particularly wasteful when recurrent patterns exist. In this work, we introduce a novel framework called PEARL, which uses both an exact technique and a probabilistic graphical model with Lossy Counting, to replace drifted trees with relevant trees built from the past. The exact technique utilizes pattern matching to find the set of drifted trees, that co-occurred in predictions in the past. Meanwhile, a probabilistic graphical model is being built to capture the tree replacements among recurrent concept drifts. Once the graphical model becomes stable, it replaces the exact technique and finds relevant trees in a probabilistic fashion. Further, Lossy Counting is applied to the graphical model which brings an added theoretical guarantee for both error rate and space complexity. We empirically show our technique outperforms baselines in terms of cumulative accuracy on both synthetic and real-world datasets.
Subject
  • Bayesian statistics
  • Patent law
  • Digital television
  • Film and video technology
  • Video compression
  • Utility software types
  • Graphical models
  • Videotelephony
  • Conditional constructs
  • Data compression
  • Digital audio
part of
is abstract of
is hasSource of
Faceted Search & Find service v1.13.91 as of Mar 24 2020


Alternative Linked Data Documents: Sponger | ODE     Content Formats:       RDF       ODATA       Microdata      About   
This material is Open Knowledge   W3C Semantic Web Technology [RDF Data]
OpenLink Virtuoso version 07.20.3229 as of Jul 10 2020, on Linux (x86_64-pc-linux-gnu), Single-Server Edition (94 GB total memory)
Data on this page belongs to its respective rights holders.
Virtuoso Faceted Browser Copyright © 2009-2024 OpenLink Software