Boosting for Regression Transfer

Boosting for Regression Transfer,David Pardoe,Peter Stone

Boosting for Regression Transfer   (Citations: 1)
BibTex | RIS | RefWorks Download
The goal of transfer learning is to improve the learning of a new target concept given knowledge of related source concept(s). We introduce the first boosting-based algorithms for transfer learning that apply to regression tasks. First, we describe two existing clas- sification transfer algorithms, ExpBoost and TrAdaBoost, and show how they can be mod- ified for regression. We then introduce exten- sions of these algorithms that improve per- formance significantly on controlled experi- ments in a wide range of test domains.
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
    • ...The main weaknesses of TrAdaBoost are highlighted in the list below: 1. Weight Mismatch: As outlined in [14], when the size of source instances is much larger than that of target instances, many iterations might be required for the total weight of the target instances to approach that of the source instances...
    • ...Their experimental analyses along with the analyses reported by Pardoe and Stone [14] and our own investigation show mixed results...
    • ...This rapid convergence also led Pardoe and Stone [14] to the use of an adjusted error scheme based on experimental approximation...
    • ...TrAdaBoost has been extended to many transfer learning problems including regression transfer [14] and multi-source learning [19]...

    Samir Al-Stouhiet al. Adaptive Boosting for Transfer Learning Using Dynamic Updates

Sort by: