English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

This work proposes Multi-task Meta Learning (MTML), integrating two learningparadigms Multi-Task Learning (MTL) and meta learning, to bring together thebest of both worlds. In particular, it focuses simultaneous learning ofmultiple tasks, an element of MTL and promptly adapting to new tasks, a qualityof meta learning. It is important to highlight that we focus on heterogeneoustasks, which are of distinct kind, in contrast to typically consideredhomogeneous tasks (e.g., if all tasks are classification or if all tasks areregression tasks). The fundamental idea is to train a multi-task model, suchthat when an unseen task is introduced, it can learn in fewer steps whilstoffering a performance at least as good as conventional single task learning onthe new task or inclusion within the MTL. By conducting various experiments, wedemonstrate this paradigm on two datasets and four tasks: NYU-v2 and thetaskonomy dataset for which we perform semantic segmentation, depth estimation,surface normal estimation, and edge detection. MTML achieves state-of-the-artresults for three out of four tasks for the NYU-v2 dataset and two out of fourfor the taskonomy dataset. In the taskonomy dataset, it was discovered thatmany pseudo-labeled segmentation masks lacked classes that were expected to bepresent in the ground truth; however, our MTML approach was found to beeffective in detecting these missing classes, delivering good qualitativeresults. While, quantitatively its performance was affected due to the presenceof incorrect ground truth labels. The the source code for reproducibility canbe found at https://github.com/ricupa/MTML-learn-how-to-adapt-to-unseen-tasks.

Multi-Task Meta Learning: learn how to adapt to unseen tasks