Heterogeneous Feature-based Transfer Learning

Symmetric transformation takes both the source feature space Xs and target feature space Xt and learns feature transformations as to project each onto a common subspace Xc for adaptation purposes. This derived subspace becomes a domain-invariant feature subspace to associate cross-domain data, and in effect, reduces marginal distribution differences.

References

Wang, Q., Mao, K. Z., Wang, B., & Guan, J. (2017). Big data clustering by hybrid optimization algorithm. Journal of Big Data, 4(1), 25. Link.