PulseAugur
LIVE 10:43:08
research · [3 sources] ·
0
research

New framework improves deep learning testing by selecting high-quality mutants

By PulseAugur Editorial · Summary by gemini-2.5-flash-lite from 3 sources

Researchers have developed a new probabilistic framework to assess the quality of mutants used in deep learning testing. This framework quantifies mutant quality based on resistance and realism, addressing a gap in current DL literature. The approach allows for the ranking and filtering of low-quality mutation-operator configurations, potentially reducing the number of generated mutants by over 55% while maintaining their effectiveness for testing and debugging. AI

Summary written by gemini-2.5-flash-lite from 3 sources. How we write summaries →

IMPACT Introduces a new method to improve the efficiency and effectiveness of deep learning testing and debugging.

RANK_REASON This is a research paper introducing a new framework for deep learning testing.

Read on arXiv cs.LG →

COVERAGE [3]

  1. arXiv cs.LG TIER_1 · Zaheed Ahmed, Emmanuel Charleson Dapaah, Philip Makedonski, Jens Grabowski ·

    Quality-Driven Selective Mutation for Deep Learning

    arXiv:2604.22640v1 Announce Type: cross Abstract: Mutants support testing and debugging in two roles: (i) as test goals and (ii) as substitutes for real faults. Hard-to-kill mutants provide better guidance for test improvement, while realism is essential when mutants are used to …

  2. arXiv cs.LG TIER_1 · Jens Grabowski ·

    Quality-Driven Selective Mutation for Deep Learning

    Mutants support testing and debugging in two roles: (i) as test goals and (ii) as substitutes for real faults. Hard-to-kill mutants provide better guidance for test improvement, while realism is essential when mutants are used to simulate real bugs. Building on these roles, selec…

  3. Hugging Face Daily Papers TIER_1 ·

    Quality-Driven Selective Mutation for Deep Learning

    Mutants support testing and debugging in two roles: (i) as test goals and (ii) as substitutes for real faults. Hard-to-kill mutants provide better guidance for test improvement, while realism is essential when mutants are used to simulate real bugs. Building on these roles, selec…

RELATED ENTITIES

RELATED TOPICS