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

In this paper, we study the multi-task sentiment classification problem inthe continual learning setting, i.e., a model is sequentially trained toclassifier the sentiment of reviews of products in a particular category. Theuse of common sentiment words in reviews of different product categories leadsto large cross-task similarity, which differentiates it from continual learningin other domains. This knowledge sharing nature renders forgetting reductionfocused approaches less effective for the problem under consideration. Unlikeexisting approaches, where task-specific masks are learned with specificallypresumed training objectives, we propose an approach called Task-aware Dropout(TaskDrop) to generate masks in a random way. While the standard dropoutgenerates and applies random masks for each training instance per epoch foreffective regularization, TaskDrop applies random masking for task-wisecapacity allocation and reuse. We conducted experimental studies on threemulti-task review datasets and made comparison to various baselines andstate-of-the-art approaches. Our empirical results show that regardless ofsimplicity, TaskDrop overall achieved competitive performances for all thethree datasets, especially after relative long term learning. This demonstratesthat the proposed random capacity allocation mechanism works well for continualsentiment classification.

TaskDrop: A Competitive Baseline for Continual Learning of Sentiment Classification