Posts by Collection

Identifying Contradicted Findings in ACL Anthology Papers

Annotating and building a dataset to identify contradictions among NLP papers on the same topic, with model training and evaluation.

Evaluating MGT Detector Bias: Native vs Non-Native English Authors

Investigating whether machine-generated text detectors exhibit systematic bias against non-native English speakers in scientific writing.

Endi-QM: Energy Efficiency via Intelligent In-Process Quality Monitoring

Applying deep learning and reinforcement learning to industrial sensor data for product quality prediction and machine parameter optimization.

RAG for Implicit Relation Detection Between Texts (M.Sc. Thesis)

A Retrieval-Augmented Generation framework to automate detection of implicit textual links, developed as M.Sc. thesis at UKP Lab.

DP-NMT: Scalable Differentially-Private Machine Translation

Published in EMNLP, 2024

We present DP-NMT, a scalable framework for differentially-private neural machine translation. We adapt and evaluate differentially-private training approaches across multiple NLP models and datasets including WMT16 and BSD.

Download Paper