La protection des données personnelles à l’heure du RGPD

Seminar from Félicien Vallet, from CNIL   Date: 17/09/2021 Time: 11h00 Localization: IC2 Salle des conseils, online Speakert: Félicien Vallet   Personal data protection in the era of the RGPD   The CNIL is the authority in charge of personal data protection in France.This presentation will review the birth of the CNIL in 1978, explain […]

Explicabilité et Interprétabilité des modèles IA : concepts, approches et adaptabilité au domaine du traitement de la parole

Seminar from Jean-François Bonastre, professor at LIA, Avignon Université   Date: 03/09/2021 Time: 14h00 Localization: IC2 Boardroom, online Speaker: Jean-François Bonastre   Explicability and Interpretability of AI models: concepts, approaches and adaptability to the domain of speech processing   This presentation proposes to define the notions of explicability and interpretability in AI and to give […]

Joint effort on data loading process and its application to VAD, speaker turn detection and overlap detection

Seminar from Martin Lebourdais and Théo Mariotte, PhD students at LIUM   Date: 18/06/2021 Time: 11h00 Localization:IC2 Salle des conseils, online Speakers: Martin Lebourdais and Théo Mariotte   Joint effort on data loading process and its application to VAD, speaker turn detection and overlap detection   Diarization is the task of finding “Who spoke when?” […]

Human assisted speaker diarization, approach and demonstrator

Seminar from diarization team   Date: 21/05/2021 Time: 11h00 Localization: online Speakers: Meysam Shamsi, Yevhenii Prokopalo and Anthony Larcher   Human assisted speaker diarization, approach and demonstrator   ALLIES and ESPERANTO projects are targeting human assisted speaker diarization across time. To address this task step by step, our team proposes a within-show human assisted correction […]

End2End Acoustic to Semantic Transduction

Seminar from Valentin Pelloin, PhD student at LIUM   Date: 12/05/2021 Time: 11h00 Localization: online Speaker: Valentin Pelloin   End2End Acoustic to Semantic Transduction   We propose a novel end-to-end sequence-to-sequence spoken language understanding model using an attention mechanism. It reliably selects contextual acoustic features in order to hypothesize semantic contents. An initial architecture capable […]

SINr: fast computing of Sparse Interpretable Node Representations is not a sin!

Seminar from Thibault Prouteau, PhD student at LIUM   Date: 19/04/2021 Time: 11h00 Localization: online Speaker: Thibault Prouteau   SINr: fast computing of Sparse Interpretable Node Representations is not a sin!   While graph embedding aims at learning low-dimensional representations of nodes encompassing the graph topology, word embedding focus on learning word vectors that encode […]

MiniBERT: a simple and explainable BERT model

Seminar from Gaëtan Caillaut, Post-Doctoral researcher at LIUM   Date: 12/03/2021 Time: 11h00 Localization: online Speaker: Gaëtan Caillaut   MiniBERT: a simple and explainable BERT model   As part of the PolysEmY project, we work with the SNCF (French railway company) to produce “polysemic-aware” word embeddings. Documents provided by the SNCF are written in technical […]

Approches de transfert en traduction automatique neuronale

Seminar by Adrien Bardet, PhD Student at LIUM, Team LST   Date: 19/02/2021 Time: 11h00 Localization: online Speaker: Adrien Bardet   Multilingual neural architectures for natural language processing   Machine translation using little data leads to poor performance. The use of multilingual systems is one solution to this problem. Multilingual machine translation systems make it […]

Les learning analytics seraient-elles promises à l’échec ?

Seminar from Rémi Venant, Lecturer at LIUM, team TEL   Date: 22/01/2021 Time: 11h00 Localization: video-conference Speaker: Rémi Venant   Are learning analytics doomed to failure?   learning analytics (LA) are undergoing increasing development, and following the COVID-19 pandemic and the forced introduction of large-scale distance learning, it is reasonable to anticipate an acceleration in […]