Elena Merdjanovska
I am a final-year PhD student in the Machine Learning Group at Humboldt University of Berlin, supervised by Prof. Alan Akbik. I am also a member of the DFG Cluster of Excellence Science of Intelligence.
My research focuses on learning with noisy labels, annotation error detection and confidence estimation in natural language processing. If you are working on an interesting project in these topics, feel free to reach out!
News
-
Attending EMNLP in Budapest! Self-Aware Knowledge Probing: Evaluating Language Models' Relational Knowledge through Confidence Calibration was accepted at the UncertaiNLP Workshop.
-
Top Reviewer at NeurIPS. Was recognized as top reviewer of the Evaluations and Datasets track and awarded free registration.
-
Invited Talk at NEC Labs Europe! Had the pleasure to give a guest talk on LLM reliability, focusing on both LLMs as annotators and as predictors. Watch the talk here.
-
New Preprint on Label-Noise-Robust Fine-Tuning! Our paper Error-Type-Aware Loss Reweighting for Robust Named Entity Recognition with Noisy LLM Labels is available on arxiv.
-
Attending ACL in San Diego! Our paper Evaluation Pitfalls and Sparsity Limitations in LLM-based Confidence Estimates for Classification, which resulted from my internship at Amazon was accepted to ACL.
-
Guest Lecture at a HU Berlin course. Gave a guest lecture on label noise and confidence calibration at the Deep Learning and Natural Language Processing course at HU Berlin.
-
Thursday Morning Talk at SCIoI and TU Berlin! Gave a talk at the Science of Intelligence cluster.
-
Attending EMNLP in Suzhou! Had the opportunity to present our paper Token-Level Metrics for Detecting Incorrect Gold Annotations in Named Entity Recognition at EMNLP.
-
Workshop Paper at ACL in Vienna. Our paper Measuring Label Ambiguity in Subjective Tasks using Predictive Uncertainty Estimation was accepted at the Linguistic Annotation Workshop and was given an oral presentation slot!
-
Guest Lecture at a HU Berlin course. Gave a guest lecture on label noise and confidence calibration at the Current Research in Natural Language Processing course at HU Berlin.
-
Girls' Day at SCIoI! Participated in Science of Intelligence's program for Girls' Day.
-
Attending EMNLP in Miami! Thrilled to present our paper NoiseBench: Benchmarking the Impact of Real Label Noise on Named Entity Recognition at EMNLP.
-
Paper Published in Scientific Reports! Our paper A framework for comparative study of databases and computational methods for arrhythmia detection from single-lead ECG, based on my Master's Thesis, was published in the Scientific Reports journal.
-
Started my PhD in Berlin! At the Humboldt University, as part of Alan Akbik's group.
-
Best Paper Award at MIPRO! Very excited that our paper Benchmarking Deep Learning Methods for Arrhythmia Detection was accepted at MIPRO and received the DS-BE conference best paper award.
-
Paper Published in ESWA! Our paper Comprehensive survey of computational ECG analysis: Databases, methods and applications was published at the Expert Systems with Applications journal.