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

As software projects rapidly evolve, software artifacts become more complexand defects behind get harder to identify. The emerging Transformer-basedapproaches, though achieving remarkable performance, struggle with long codesequences due to their self-attention mechanism, which scales quadraticallywith the sequence length. This paper introduces SparseCoder, an innovativeapproach incorporating sparse attention and learned token pruning (LTP) method(adapted from natural language processing) to address this limitation.Extensive experiments carried out on a large-scale dataset for vulnerabilitydetection demonstrate the effectiveness and efficiency of SparseCoder, scalingfrom quadratically to linearly on long code sequence analysis in comparison toCodeBERT and RoBERTa. We further achieve 50% FLOPs reduction with a negligibleperformance drop of less than 1% comparing to Transformer leveraging sparseattention. Moverover, SparseCoder goes beyond making "black-box" decisions byelucidating the rationale behind those decisions. Code segments that contributeto the final decision can be highlighted with importance scores, offering aninterpretable, transparent analysis tool for the software engineeringlandscape.

SparseCoder: Advancing Source Code Analysis with Sparse Attention and Learned Token Pruning