Tommasi, Tatiana; Quadrianto, Novi; Caputo, Barbara; Chatterjee, KrishnenduIST Austria
Many visual datasets are traditionally used to analyze the performance of different learning techniques. The evaluation is usually done within each dataset, therefore it is questionable if such results are a reliable indicator of true generalization ability. We propose here an algorithm to exploit the existing data resources when learning on a new multiclass problem. Our main idea is to identify an image representation that decomposes orthogonally into two subspaces: a part specific to each dataset, and a part generic to, and therefore shared between, all the considered source sets. This allows us to use the generic representation as un-biased reference knowledge for a novel classification task. By casting the method in the multi-view setting, we also make it possible to use different features for different databases. We call the algorithm MUST, Multitask Unaligned Shared knowledge Transfer. Through extensive experiments on five public datasets, we show that MUST consistently improves the cross-datasets generalization performance.
This work was supported by the PASCAL 2 Network of Excellence (TT) and by the Newton International Fellowship (NQ)
1 - 15
ACCV: Asian Conference on Computer Vision
2012-11-05 – 2012-11-09
Tommasi T, Quadrianto N, Caputo B, Lampert C. Beyond dataset bias: Multi-task unaligned shared knowledge transfer. 2013;7724:1-15. doi:10.1007/978-3-642-37331-2_1
Tommasi, T., Quadrianto, N., Caputo, B., & Lampert, C. (2013). Beyond dataset bias: Multi-task unaligned shared knowledge transfer. Presented at the ACCV: Asian Conference on Computer Vision, Daejeon, Korea: Springer. https://doi.org/10.1007/978-3-642-37331-2_1
Tommasi, Tatiana, Novi Quadrianto, Barbara Caputo, and Christoph Lampert. “Beyond Dataset Bias: Multi-Task Unaligned Shared Knowledge Transfer.” Lecture Notes in Computer Science. Springer, 2013. https://doi.org/10.1007/978-3-642-37331-2_1.
T. Tommasi, N. Quadrianto, B. Caputo, and C. Lampert, “Beyond dataset bias: Multi-task unaligned shared knowledge transfer,” vol. 7724. Springer, pp. 1–15, 2013.
Tommasi T, Quadrianto N, Caputo B, Lampert C. 2013. Beyond dataset bias: Multi-task unaligned shared knowledge transfer. 7724, 1–15.
Tommasi, Tatiana, et al. Beyond Dataset Bias: Multi-Task Unaligned Shared Knowledge Transfer. Vol. 7724, Springer, 2013, pp. 1–15, doi:10.1007/978-3-642-37331-2_1.