Sharing information between multiple tasks enables algorithms to achieve good generalization performance even from small amounts of training data. However, in a realistic scenario of multi-task learning not all tasks are equally related to each other, hence it could be advantageous to transfer information only between the most related tasks. In this work we propose an approach that processes multiple tasks in a sequence with sharing between subsequent tasks instead of solving all tasks jointly. Subsequently, we address the question of curriculum learning of tasks, i.e. finding the best order of tasks to be learned. Our approach is based on a generalization bound criterion for choosing the task order that optimizes the average expected classification performance over all tasks. Our experimental results show that learning multiple related tasks sequentially can be more effective than learning them jointly, the order in which tasks are being solved affects the overall performance, and that our model is able to automatically discover the favourable order of tasks.
5492 - 5500
CVPR: Computer Vision and Pattern Recognition
Boston, MA, United States
2015-06-07 – 2015-06-12
Pentina A, Sharmanska V, Lampert C. Curriculum learning of multiple tasks. In: IEEE; 2015:5492-5500. doi:10.1109/CVPR.2015.7299188
Pentina, A., Sharmanska, V., & Lampert, C. (2015). Curriculum learning of multiple tasks (pp. 5492–5500). Presented at the CVPR: Computer Vision and Pattern Recognition, Boston, MA, United States: IEEE. https://doi.org/10.1109/CVPR.2015.7299188
Pentina, Anastasia, Viktoriia Sharmanska, and Christoph Lampert. “Curriculum Learning of Multiple Tasks,” 5492–5500. IEEE, 2015. https://doi.org/10.1109/CVPR.2015.7299188.
A. Pentina, V. Sharmanska, and C. Lampert, “Curriculum learning of multiple tasks,” presented at the CVPR: Computer Vision and Pattern Recognition, Boston, MA, United States, 2015, pp. 5492–5500.
Pentina A, Sharmanska V, Lampert C. 2015. Curriculum learning of multiple tasks. CVPR: Computer Vision and Pattern Recognition 5492–5500.
Pentina, Anastasia, et al. Curriculum Learning of Multiple Tasks. IEEE, 2015, pp. 5492–500, doi:10.1109/CVPR.2015.7299188.
Link(s) to Main File(s)