LETITIA PARCALABESCU

Letitia Parcalabescu

Contact

+49 6221 54-3483
Email (click to copy)
Institut für Computerlinguistik
Im Neuenheimer Feld 325
D-69120 Heidelberg
Germany

About me:

🎓 I've defended my PhD Thesis on the 19th of July 2024! 🎓

Since January 1st, 2019 I have been working under the supervision of Prof. Dr. Anette Frank at the Department of Computational Linguistics at Heidelberg University. My research focused on measuring how well vision and language models integrate their modalities for achieving multimodal understanding. Specifically, I analysed this through model benchmarking, interpretation, and self-explanations.

Also check out my 📺 YouTube channel and personal web page.

NEWS

🎓 PhD Defense

19th July 2024

I've successfully defended my PhD Thesis in July 2024, and I will stop updating this website from now on. Everything you see below will serve as a snapshot of my achievements during my time as a PhD student. If you are curios about the general picture of this time, I ahave made a video about it!

AI Coffee Break with Letitia YouTube Channel

31st May 2020

Ms. Coffee Bean The birth of Ms. Coffee Bean. 📺

🫂 Multi3Generation COST action

7th March 2021

Anette Frank and Letitia Parcalabescu are participating since its beginning in the COST Action Multi3Generation: Multi-task, Multilingual, Multi-modal Language Generation, initiated by Isabelle Augenstein. We are members of the management committee as representatives for Germany. We are looking forward to collaborating in this great network activity.

🗣️ Invited Talk at the Datafest Conference 2024

6 September 2024

Invited talk at the Datafest Yerevan Conference (2024). I'll be talking about own work measuring the self-consistency of self-explanations generated by LLMs and VLMs.

🎙️ Podcast Interview: Deep Learning with Letitia Parcalabescu

6 June 2024

Appeared at the Deep Learning with Letitia Parcalabescu - Weaviate Podcast #96! where we discussed YouTube in the ML space, current trends and the evolution of deep learning. Tune in!

🔉 Panel and Keynote 🗣️: Which skills do I need to benefit from AI technology?

15 March 2024

Happening in Florence at the VOICES festival! 🗺️ I will speak in a panel about Artificial Intelligence to a broad audience. We'll explore the question "Which skills do I need to benefit from AI technology?"

🗣️ Invited Talk: Aleph Alpha

30 January 2024

Invited talk at Aleph Alpha – AI startup in Heidelberg.

🗣️ Invited Talk: cogsys-group, CLASP, Gothenburg, Sweden

26 January 2024

Invited talk at cogsys-group, CLASP, Gothenburg, Sweden.

🗣️ Invited Talk: heidelberg.ai @ DKFZ

28 November 2023

Invited to talk about on work at heidelberg.ai at the DKFZ.

🔉 Panelist: Tbh-conference 2023

8 December 2023

Invited to talk in a panel about Digital Tools & AI in Research at To be honest Conference.

🎙️ Guest: ARD MixTalk

18 October 2023

Invited to talk (in German) about AI for a general audience at ARD MixTalk.

🎙️Interview: Handelsblatt

13 October 2023

Podcast (in German) about AI for the Handelsblatt.

🗣️ Keynote @LIMO 2023 Workshop at KONVENS2023

28th November 2022

Giving a keynote at the LIMO 2023 Workshop About Vision and Language (VL) models.

🎙️Guest: Transformative Ideas Podcast

7 August 2023

Invited to talk about own work, the quickly shifting landscape around LLMs and multimodal models, AI, science communication and more on the Transformative Ideas Podcast.

🗣️ Invited Talk @ICDM 2022 Workshop

28th November 2022

Invited Talk at the ICDM 2022 Workshop Foundation Models in Vision and Language.

🔉 Panelist: Popularization in ML Research Panel

4th November 2022 (upcoming)

Invited to speak in a panel at the ML in PL conference.

🎙️ Ask-me-anything session

26th February 2022

☕⚔️🍵 We had a pair-AMA session with Sanyam Bhutani Sanyam Bhutani hosted by Weights & Biases's Head of Data Science, Dr. Bryan Bischof. We talked about educational content, videos in ML and about the AI and ML community in general. Check it out by clicking on the arrow to the right.

🎙️ Guest: AI misconceptions

30th Movember 2021

We had a conversation about AI (mis)conceptions at EAVI - Media Literacy for Citizenship. Check out the textual summary or the recording by clicking on the arrow to the right.

📺 Host: Machine Learning Street Talk

25th July 2021

Co-hosting the 44th and 57th episodes of the Machine Learning Street Talk YouTube channel and podcast.

🗣️ Invited Talk @Stuts2020

19th November 2020

Invited to speak at The 68th edition of StuTS.

PUBLICATIONS

2024
+ -

On Measuring Faithfulness or Self-consistency of Natural Language Explanations

Parcalabescu, L., Frank, A. (2024)

Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) .

ViLMA: A Zero-Shot Benchmark for Linguistic and Temporal Grounding in Video-Language Models

Kesen, I., Pedrotti, A., Dogan, M., Cafagna, M., Acikgoz, E., Parcalabescu, L., Calixto, I., Frank, A., Gatt, A., Erdem, A., Erdem, E. (2024)

arXiv preprint arXiv:2311.07022.

Do Vision & Language Decoders use Images and Text equally? How Self-consistent are their Explanations?

Parcalabescu, L., Frank, A. (2024)

arXiv preprint arXiv:2404.18624.
2023
+ -

MM-SHAP: A Performance-agnostic Metric for Measuring Multimodal Contributions in Vision and Language Models & Tasks

Parcalabescu, L., Frank, A. (2023)

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (ACL’23), Toronto, Canada.
2022
+ -

MAGMA – Multimodal Augmentation of Generative Models through Adapter-based Finetuning

Eichenberg, C., Black, S., Weinbach, S., Parcalabescu, L., Frank, A. (2022)

Findings of EMNLP.

Neural Natural Language Generation: A Survey on Multilinguality, Multimodality, Controllability and Learning

Erdem, E., Kuyu, M., Yagcioglu, S., Frank, A., Parcalabescu, L., Plank, B., Babii, A., Turuta, O., Erdem, A., Calixto, I., Lloret, E., Apostol, E., Truica, C., Sandrih, B., Gatt, A., Martincic-Ipsic, S., Berend, G., Korvel, G. (2022)

Journal of Artificial Intelligence Research (JAIR).

VALSE💃: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena

Parcalabescu, L., Cafagna, M., Muradjan, L., Frank, A., Calixto, I., Gatt, A. (2022)

Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics.
2021
+ -

What is Multimodality?

Parcalabescu, L., Trost, N., Frank, A. (2021)

Proceedings of the First Workshop on Multimodal Semantic Representations (MMSR), Groningen, The Netherlands (Online).

Seeing Past Words: Testing the Cross-Modal Capabilities of Pretrained V&L Models

Parcalabescu, L., Gatt, A., Frank, A., Calixto, I. (2021)

Proceedings of the First Workshop on Multimodal Semantic Representations (MMSR), Groningen, The Netherlands (Online).

Grounding Plural Phrases: Countering Evaluation Biases by Individuation

Suter, J., Parcalabescu, L., Frank, A. (2021)

Proceedings of the 2nd Workshop on Advances in Language and Vision Research (ALVR 2021) at NAACL 2021, Online.
2020
+ -

AMR Similarity Metrics from Principles

Opitz, J., Parcalabescu, L., Frank, A. (2020)

Transactions of the Association for Computational Linguistics.

Exploring Phrase Grounding without Training: Contextualisation and Extension to Text-Based Image Retrieval

Parcalabescu, L., Frank, A. (2020)

Proceedings of the Workshop on Multimodal Learning (In conjunction with CVPR 2020), Seattle, WA, USA.
2018
+ -

WEADE: a workflow for enrichment analysis and data exploration

Trost, N., Rempel, E., Ermakova, O., Tamirisa, S., Parcalabescu, L., Boutros, M., Lohmann, J., Lohmann, I. (2018)

PloS one.

TEACHING

SS 24
+ -

Methods for Learning without Annotated Data

WS 23/24
+ -

Experimente gestalten fürs Maschinelle Lernen

SS 23
+ -

Methods for Learning without Annotated Data

WS 22/23
+ -

Experimente gestalten fürs Maschinelle Lernen

SS 22
+ -

Methods for Learning without Annotated Data

WS 21/22
+ -

Experimente gestalten fürs Maschinelle Lernen

SS 21
+ -

Methods for Learning without Annotated Data

WS 20/21
+ -

Experimente gestalten fürs Maschinelle Lernen

Einführung in die Nutzung computerlinguistischer Ressourcen

Programmierprüfung

SS 20
+ -

Methods for Learning without Annotated Data

Einführung in die Nutzung computerlinguistischer Ressourcen

Programmierprüfung

WS 19/20
+ -

Integrating Vision and Language: Achievements and Challenges in Multimodal Machine Learning

Mi. 16:15 - 17:45

CURRICULUM VITAE

Find my CV as PDF 📄.

Employment

2021 - today

Co-Founder Explanation Mark

Science Communication, Parcalabescu u. Trost GbR

2019 - 2024

Academic Employee

Department of Computational Linguistics (ICL), Heidelberg University

2017 - 2018

HiWi at COS

Web Programming at the Centrum for Organismal Studies (COS) Heidelberg University; http://beta-weade.cos.uni-heidelberg.de

2017

Tutor AlDa Course

Tutoring the Algorithms and Data Structures course at Heidelberg University

2016

Tutor Python Course

Tutoring the Python course at Heidelberg University

2015 - 2016

HiWi at HCI

Research and software programming for ilastik at the Heidelberg Collaboratory for Image Processing (HCI), Heidelberg University

Education

2019 - 2024

PhD in NLP

Thesis: “Measuring the Contributions of Vision and Text Modalities in Multimodal Transformers” focusing on benchmarking, interpretability and explainability of multimodal models (grade: summa cum laude).

Published 12 papers in leading AI/ML conferences and workshops.

Department of Computational Linguistics (ICL), Heidelberg University

2019

Qualifying Studies for Research in NLP

Department of Computational Linguistics (ICL), Heidelberg University

2017 - 2018

M.Sc. Physics

Faculty of Physics and Astronomy, Heidelberg University, Germany

2015 - 2017

B.Sc. Applied Computer Science

Faculty of Mathematics and Computer Science, Heidelberg University, Germany

2013 - 2016

B.Sc. Physics

Faculty of Physics and Astronomy, Heidelberg University, Germany

DAAD Scholarship Holder

DAAD Vollstipendium für Absolventen Deutscher Auslandschulen

2013

Highschool Graduation

Liceul Teoretic "Johannes Honterus", Brașov, Romania