Gustave Cortal
I'm a third-year PhD student at Laboratoire Méthodes Formelles (École Normale Supérieure Paris-Saclay, Université Paris-Saclay) under the supervision of Alain Finkel, Patrick Paroubek and Lina Ye since 2022.
Contact
For research or teaching purposes, please get in touch with me at: gustave.cortal (at) ens-paris-saclay (dot) fr.
Research Interests
- fundamentals of natural language processing
- natural language processing for subjective analysis
- psychological theories of emotion
- philosophy of mind, language and technology
The Cognitive Analysis of Emotions, developed by Alain Finkel, is a cognitive appraisal theory that explores emotions with their corresponding behavioral, physiological, and cognitive components in guided narratives.
National journals
Gustave Cortal. Mar 2024. Automatisation du codage des personnages et de leurs émotions dans les récits de rêves avec des modèles de langue [Sequence-to-Sequence Language Models for Character and Emotion Detection in Dream Narratives]. In Traitement Automatique des Langues, Volume 65, Numéro 1 : Varia, pages 11-35, France. ATALA (Association pour le Traitement Automatique des Langues).
International conferences
Nicolas Richet, Soufiane Belharbi, Haseeb Aslam, Meike Emilie Schadt, Manuela González-González, Gustave Cortal, Alessandro Lameiras Koerich, Marco Pedersoli, Alain Finkel, Simon Bacon, Eric Granger. Sep 30, 2024. Textualized and Feature-based Models for Compound Multimodal Emotion Recognition in the Wild. In Proceedings of the 7th Workshop and Competition on Affective Behavior Analysis in-the-wild (ABAW), Milan, Italy. European Conference on Computer Vision (ECCV). Our models achieved third place in the competition.
Constant Bonard and Gustave Cortal (equal contribution). Aug 11-16, 2024. Improving Language Models for Emotion Analysis: Insights from Cognitive Science. In Proceedings of the Workshop on Cognitive Modeling and Computational Linguistics, pages 264–277, Bangkok, Thailand. Association for Computational Linguistics. 63.9% acceptance rate.
Gustave Cortal. May 20-25, 2024. Sequence-to-Sequence Language Models for Character and Emotion Detection in Dream Narratives. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 14717–14728, Torino, Italy. ELRA and ICCL. slides. video. 20.6% acceptance rate (oral presentation).
Gustave Cortal, Alain Finkel, Patrick Paroubek, Lina Ye. March 17-22, 2024. Emotion Recognition based on Psychological Components in Guided Narratives for Emotion Regulation. In Proceedings of the 7th Joint SIGHUM Workshop on Computational Linguistics for Cultural Heritage, Social Sciences, Humanities and Literature, pages 72–81, Dubrovnik, Croatia. Association for Computational Linguistics. slides. video. 43.7% acceptance rate (oral presentation).
Gustave Cortal, Alain Finkel, Patrick Paroubek, Lina Ye. Sep 6-7, 2022. Natural Language Processing for Cognitive Analysis of Emotions. Semantics, Memory, and Emotion, Paris, France. slides.
National conferences
Constant Bonard and Gustave Cortal. Jul 8-12, 2024. Améliorer les modèles de langue pour l'analyse des émotions : perspectives venant des sciences cognitives. In Actes de la 31ème Conférence sur le Traitement Automatique des Langues Naturelles, volume 1 : articles longs et prises de position, pages 307–322, Toulouse, France. ATALA and AFPC. 72.7% acceptance rate.
Posters
Constant Bonard and Gustave Cortal. Jul 8-12, 2024. Améliorer les modèles de langue pour l'analyse des émotions : perspectives venant des sciences cognitives. JEP-TALN-RECITAL 2024. 31ème Conférence sur le Traitement Automatique des Langues Naturelles. Toulouse, France.
Gustave Cortal, Alain Finkel, Patrick Paroubek, Lina Ye. Jul 4, 2023. Automatisation du codage des récits de rêve : une approche basée sur la génération de paraphrases. Affects, Compagnons Artificiels et Interactions, Strasbourg, France.
Gustave Cortal, Alain Finkel, Patrick Paroubek, Lina Ye. Jan 16-20, 2023. Détection de l'émotion à partir de ses composantes dans des récits émotionnels. Extraction et Gestion des Connaissances, Lyon, France.
Gustave Cortal, Alain Finkel, Patrick Paroubek, Lina Ye. Sep 5-7, 2022. Natural Language Processing for Cognitive Analysis of Emotions. Meeting of the European Mathematical Psychology Group, Rovereto, Italy.
Talks
Gustave Cortal. 31 May, 2024. Statistical Language Modeling: from N-grams to Transformers. Tutorial for Formal Methods and Artificial Intelligence working group of Laboratoire Méthodes Formelles (ENS Paris-Saclay). slides. video.
Gustave Cortal. Feb 6, 2024. Natural Language Processing: Perspectives on Emotion Analysis, Seminar of Laboratoire Méthodes Formelles (ENS Paris-Saclay). slides. video.
Gustave Cortal and Constant Bonard. Jan 18, 2024. Large Language Models and Emotional Understanding, Artificial Intelligence and Philosophy group at Institut Jean-Nicod (EHESS, ENS Ulm). slides.
Gustave Cortal. Dec 5, 2022. Automatic identification of emotions and their causes in texts, Doctoral Students Day 2nd edition (DATAIA). slides
Teaching experience
Primary Instructor, Natural Language Processing, École Pour l'Informatique et les Techniques Avancées, 2024. Semester-long course to 70 final-year engineering students (30h). The course, cited in the French journal Le Monde, is entirely made by myself, with slides and exercises available on my github. n-gram, logistic regression, neural networks, transformer, large language models.
Primary Instructor, Statistical Learning with R, École Normale Supérieure Paris-Saclay, 2023. Semester-long computer science course to 25 Normalien students in economics (21h). The course is entirely made by myself with slides, exercises, and exams available on my github. data visualization, logistic regression, linear regression, tree and random forest, principal component analysis.
Teaching Assistant, Algorithms and Complexity, CentraleSupélec, 2022. Semester-long course to 30 engineering students (21h). graph theory, computational complexity theory, data structure, dynamic programming.
Student supervision
Nicolas Richet (M2, National School of Computer Science for Industry and Business). Multimodal Emotion Analysis (co-supervised with Eric Granger and Muhammad Haseeb Aslam). March - Jul 2024.
Utkarsh Pant (2nd year Bachelor, Bombay Indian Institute of Technology). Natural Language Processing and Formal Methods (co-supervised with Alain Finkel). May - Jul 2024.
Racha Hachem (M1, Paris-Saclay University). Imageability Prediction in Emotional Narratives (co-supervised with Alain Finkel). May - Sep 2023.
Tools
All models trained for my research are made publicly available on Hugging Face. My work was performed using HPC resources (Jean Zay supercomputer) from GENCI-IDRIS (Grant 20XX-AD011014205).
Gustave Cortal. Jun 2024. The Android and The Machine, a dataset composed of 10,000 real and generated dreams. Real dreams are from DreamBank. Generated dreams were produced with Oneirogen.
Gustave Cortal. Jun 2024. Oneirogen, a language model for dream generation. Several sizes are available: 0.5B, 1.5B and 7B. Generation examples are available on my website.
Gustave Cortal. Aug 2023. Language model for character and emotion prediction in dream narratives and +27,000 annotated dream narratives (+600 downloads).
Gustave Cortal, Alain Finkel, Patrick Paroubek, Lina Ye. Nov 2022. Language model for emotion component prediction and language model for discrete emotion prediction in French emotional narratives (+200 downloads).
Gustave Cortal. Mar 2022. Language model for generating French headlines related to Covid-19 (+1100 downloads). This model has been trained on an internal dataset containing over 40,000 unique news articles related to Covid-19 from more than 50 French-speaking online newspapers. Generation examples are available on my website.
Gustave Cortal. Mar 2022. Notebook for finetuning large language models with Low-Rank Adaptation and 8-bit quantization (+60 stars). One of the first implementations of LoRA, before it was implemented on well-known libraries (eg., PEFT).