cv
Basics
| Name | Eleonora Mancini |
| Label | PhD Candidate · Multimodal AI for Human Expression Understanding |
| e.mancini@unibo.it | |
| Url | https://helemanc.github.io/ |
| Summary | PhD candidate in Computer Science and Engineering at the University of Bologna. Research focus: data representation, fusion, and interpretability for multimodal deep learning with emphasis on audio and speech. |
Work
-
2025.01 - 2025.07 Barcelona, Spain
Visiting Researcher
Sony AI
Transferred speech-model capabilities to music using large ASR models and LLMs for enhanced lyrics representations and multimodal musical version identification.
- ASR+LLMs for lyrics representations
- Multimodal version identification
-
2023.09 - 2023.12 Montréal, Canada
Visiting Researcher
Mila – Quebec AI Institute
Explored interpretability in speech analysis; evaluated multiple post-hoc explainers and developed a post-hoc method to produce listenable time-domain explanations for audio classifiers.
- 100+ experiments on AudioMNIST
- Time-domain listenable explanations
-
2022.01 - 2022.10 Bologna, Italy
Research Fellow
Language Technologies Lab, University of Bologna
Vague clauses detection in privacy policies (GDPR). Built a BIO tagging classifier achieving macro F1 = 0.91 on clause categorization.
- GDPR privacy policy analysis
- BIO tagging classifier (F1=0.91)
-
2021.09 - 2025.11 Bologna, Italy
Teaching Assistant
University of Bologna
Assisted courses in NLP (M.Eng AI), Real Time Systems for Automation, Python Programming (BSc Genomics), and Computer Science (BSc Mathematics). Delivered lectures, managed coursework, and mentored students (AI Camp).
- NLP (M.Eng AI)
- Real Time Systems for Automation
- Python Programming
- Computer Science
-
2021.04 - 2021.09 Barcelona, Spain
Research Intern
i2CAT Foundation
Detected disruptive situations in public transport using Speech Emotion Recognition with robustness to noise and overlapping speakers.
- Speaker/gender-independent SER
- F1 > 90% in noisy, multi-speaker settings
Volunteer
-
2024.07 - 2025.08 Vienna, Austria
-
2024.05 - 2024.05 Turin, Italy
-
2022.09 - 2022.09 Bologna, Italy
-
2022.01 - 2025.11 Bologna, Italy
Education
Awards
- 2023.06.01
Best Scientific Report & 2nd Best Poster
International Semantic Web Research Summer School (ISWS)
Bertinoro, Italy
- 2022.11.01
Publications
-
2025 -
2025 Promoting the Responsible Development of Speech Datasets for Mental Health and Neurological Disorders Research
Journal of Artificial Intelligence Research (JAIR), 82: 937–972
-
2025 LMAC-TD: Producing Time Domain Explanations for Audio Classifiers
ICASSP 2025 — IEEE International Conference on Acoustics, Speech and Signal Processing
Hyderabad, India, pp. 1–5.
-
2025 Investigating the Effectiveness of Explainability Methods in Parkinson’s Detection from Speech
IEEE ICASSP Workshops (ICASSPW 2025)
Hyderabad, India, pp. 1–5.
-
2025 Overview of MM-ArgFallacy2025 on Multimodal Argumentative Fallacy Detection and Classification in Political Debates
Proceedings of the 12th Argument Mining Workshop (ACL 2025)
pp. 358–368, Vienna, Austria.
-
2024 Disruptive Situations Detection on Public Transports through Speech Emotion Recognition
Intelligent Systems with Applications (ISWA) 21: 200305
-
2024 Multimodal Fallacy Classification in Political Debates
Proceedings of the 18th Conference of the European Chapter of the ACL (EACL 2024)
St. Julian’s, Malta.
-
2024 MAMKit: A Comprehensive Multimodal Argument Mining Toolkit
Proceedings of the 11th Workshop on Argument Mining (ArgMining 2024), ACL 2024
pp. 69–82, Bangkok, Thailand.
-
2024 Data Representation, Fusion and Interpretability in Multimodal Deep Learning for Natural Language Processing
ECAI Doctoral Consortium (ECAI 2024)
-
2023 Enriching hate-tuned transformer-based embeddings with emotions for the categorization of sexism
Working Notes of CLEF 2023 — EXIST 2023 Workshop
-
2023 Draw Me Like My Triples: Leveraging Generative AI for Wikidata Image Completion
Proceedings of the 4th Wikidata Workshop at ISWC 2023
-
2023 Towards Symbiotic Creativity: A Methodological Approach to Compare Human and AI Robotic Dance Creations
Proceedings of IJCAI 2023
pp. 5806–5814.
-
2022 Multimodal Argument Mining: A Case Study in Political Debates
Proceedings of the 9th Workshop on Argument Mining, COLING 2022
pp. 158–170, Online and Gyeongju, Republic of Korea.
Skills
| Programming | |
| Python | |
| Java | |
| C | |
| C# | |
| R | |
| Scala | |
| Prolog | |
| SQL | |
| MATLAB |
| ML Frameworks | |
| PyTorch | |
| TensorFlow |
| High-Performance Computing | |
| SLURM | |
| Parallel ML workflows | |
| HPC clusters | |
| Multi-GPU computing |
| Large Language Models | |
| Fine-tuning | |
| Model parallelism | |
| Distributed inference |
| Version Control | |
| GitHub | |
| GitLab | |
| Git | |
| Bitbucket |
| Soft skills | |
| Problem finding | |
| Problem solving | |
| Communication | |
| Teamwork | |
| Leadership |
Languages
| Italian | |
| Native |
| English | |
| Fluent |
| Spanish | |
| Fluent |
Interests
| Multimodal NLP & Speech | |
| Interpretability | |
| Audio ML | |
| Argument Mining | |
| Clinical Speech Analysis |
Projects
- 2022.10 - 2024.12
HumanE-AI-Net (H2020)
PI for 'Promoting Fairness and Diversity in Speech Datasets for Affective Computing'; researcher on 'Emotion Recognition for Human-Centered Conversational Agents'.
- Guidelines for diverse/unbiased speech datasets
- Affective computing for conversational agents
- 2021.04 - 2021.09
5GMED (H2020)
Ambient Intelligence – Disruptive Situations Detection on public transport through Speech Emotion Recognition.
- Noise-robust SER
- Public transport safety