Mitigating Concept Drift for Time-varying Domain through Adaptive Learning 6a.022.UL


Project Start Date: Jul 1, 2017
Research Areas: Analytics, Analytics - Machine Learning
Funding: Member Funded
Project Tags: ,

Project Summary

The main goals of this project:

  1. Discovery of emerging and declining trends =  association analysis + neural network predictors
  2. Historical spatio-temporal data model à concept drift instances
  3. Ensure valid concepts are not prematurely declared non-viable
  4. Ensure concepts that are on the verge of depreciating are removed

Evaluate the developed algorithm upon:

  1. Our repository of datasets (real and synthetic)
  2. IAB datasets (in order to meet the IAB needs)