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

We integrate contrastive learning (CL) with adversarial learning toco-optimize the robustness and accuracy of code models. Different from existingworks, we show that code obfuscation, a standard code transformation operation,provides novel means to generate complementary `views' of a code that enable usto achieve both robust and accurate code models. To the best of our knowledge,this is the first systematic study to explore and exploit the robustness andaccuracy benefits of (multi-view) code obfuscations in code models.Specifically, we first adopt adversarial codes as robustness-promoting views inCL at the self-supervised pre-training phase. This yields improved robustnessand transferability for downstream tasks. Next, at the supervised fine-tuningstage, we show that adversarial training with a proper temporally-staggeredschedule of adversarial code generation can further improve robustness andaccuracy of the pre-trained code model. Built on the above two modules, wedevelop CLAWSAT, a novel self-supervised learning (SSL) framework for code byintegrating $\underline{\textrm{CL}}$ with $\underline{\textrm{a}}$dversarialvie$\underline{\textrm{w}}$s (CLAW) with $\underline{\textrm{s}}$taggered$\underline{\textrm{a}}$dversarial $\underline{\textrm{t}}$raining (SAT). Onevaluating three downstream tasks across Python and Java, we show that CLAWSATconsistently yields the best robustness and accuracy ($\textit{e.g.}$ 11$\%$ inrobustness and 6$\%$ in accuracy on the code summarization task in Python). Weadditionally demonstrate the effectiveness of adversarial learning in CLAW byanalyzing the characteristics of the loss landscape and interpretability of thepre-trained models.

CLAWSAT: Towards Both Robust and Accurate Code Models