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

Can deep learning solve multiple tasks simultaneously, even when they areunrelated and very different? We investigate how the representations of theunderlying tasks affect the ability of a single neural network to learn themjointly. We present theoretical and empirical findings that a single neuralnetwork is capable of simultaneously learning multiple tasks from a combineddata set, for a variety of methods for representing tasks -- for example, whenthe distinct tasks are encoded by well-separated clusters or decision treesover certain task-code attributes. More concretely, we present a novel analysisthat shows that families of simple programming-like constructs for the codesencoding the tasks are learnable by two-layer neural networks with standardtraining. We study more generally how the complexity of learning such combinedtasks grows with the complexity of the task codes; we find that combining manytasks may incur a sample complexity penalty, even though the individual tasksare easy to learn. We provide empirical support for the usefulness of thelearning bounds by training networks on clusters, decision trees, and SQL-styleaggregation.

One Network Fits All? Modular versus Monolithic Task Formulations in Neural Networks