Software engineering, machine learning and security
ISSTA 26 · Oct 2026
AnaDroid trains specialized LLMs to classify failed mobile tests, analyze their root causes and recommend repairs.
ICASSP 26 · May 2026
TRIGGERUE uses class-wise backdoor triggers to construct unlearnable examples and protect data from unauthorized model training.
ICSE 26 · April 2026
Ciallo enables continual learning for industrial fault localization through representative exemplars and adaptive training weights.
TDSC 25 · Dec 2025
TRAP detects poisoned samples early in training and removes backdoors by retraining the classifier on relabeled poisoned samples.
FSE 25 · June 2025
Legato uses tailored semi-supervised graph learning to improve fault localization when only a small portion of failed tests can be labeled.
ASE 24 · Oct 2024
DroidCoder improves Android code completion with context-enriched retrieval-augmented generation and tailored fine-tuning.
ICSE 25 · May 2025
FALCON localizes faults in industrial software from semantic log graphs through contrastive learning and adaptive augmentation.
QRS 23 · Oct 2023
GenDroid translates natural language test intents into mobile test events using language models, random forests and UI transfer graphs.
MobileSoft 2022, Student Research Competition · May 2022
Generating mobile GUI test scripts from natural language descriptions with a pre-trained language model.