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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

A variety of machine learning applications expect to achieve rapid learningfrom a limited number of labeled data. However, the success of most currentmodels is the result of heavy training on big data. Meta-learning addressesthis problem by extracting common knowledge across different tasks that can bequickly adapted to new tasks. However, they do not fully exploreweakly-supervised information, which is usually free or cheap to collect. Inthis paper, we show that weakly-labeled data can significantly improve theperformance of meta-learning on few-shot classification. We propose prototypepropagation network (PPN) trained on few-shot tasks together with dataannotated by coarse-label. Given a category graph of the targeted fine-classesand some weakly-labeled coarse-classes, PPN learns an attention mechanism whichpropagates the prototype of one class to another on the graph, so that theK-nearest neighbor (KNN) classifier defined on the propagated prototypesresults in high accuracy across different few-shot tasks. The training tasksare generated by subgraph sampling, and the training objective is obtained byaccumulating the level-wise classification loss on the subgraph. The resultinggraph of prototypes can be continually re-used and updated for new tasks andclasses. We also introduce two practical test/inference settings which differaccording to whether the test task can leverage any weakly-supervisedinformation as in training. On two benchmarks, PPN significantly outperformsmost recent few-shot learning methods in different settings, even when they arealso allowed to train on weakly-labeled data.

Prototype Propagation Networks (PPN) for Weakly-supervised Few-shot Learning on Category Graph