Sharing research, catalyzing ideas
CRIM’s R&D seminars provide a forum for exchange and reflection, open to our teams, partners and the general public. They provide an opportunity to discover our current projects, explore current issues in applied science and nurture a culture of collective innovation.
Each meeting highlights concrete research projects, emerging technological challenges and innovative methodological approaches.
Some recent examples
Natural language, requirements and agility: a happy marriage (November 2024)
This presentation explores how natural language processing approaches can help extract conceptual models from user stories in an agile development context.
Three techniques are compared: rule-based approaches, large language models (LLMs) and conditional random fields (CRFs). A perspective is also offered on the integration of software safety practices in AI applications.
Understanding and addressing ethical laundering in machine learning (January 2025)
A critical reflection on AI development practices, highlighting the risks of ethical laundering and ways to prevent them in machine learning projects.
Attention Models for Robust Audio-Visual Learning (February 2025)
This presentation addresses the challenges of automatic speech processing in sparse or unannotated data environments. It provides an overview of the approaches used at CRIM, including model adaptation, self-supervised learning, massively multilingual models and synthetic data generation.
Sharing, stimulating, advancing
These seminars illustrate CRIM’s commitment to sharing its advances, stimulating collective thinking and advancing applied research in artificial intelligence.