Speaker: Tulika Bose Date and place: October 1, 2020 at 10:30 -C005 + VISIO-CONFERENCE Abstract: The proliferation of abusive language in social media in recent years is alarming. It requires proactive and automated mechanisms to help in detecting and dealing with them. In this context, it is important to analyze the topics raised in social …
Category: Seminars
Sep 17
Regularization of the embedding extractor for robust language identification
Speaker: Raphaël Duroselle Date and place: September 17, 2020 at 10:30 -VISIO-CONFERENCE Abstract: Language identification systems achieve impressive performance in matched conditions, when the training data corresponds to the testing conditions. However, in the presence of an important domain shift, performance drops drastically. The main focus of this work is to address this issue to …
Sep 10
Expressive speech synthesis using deep learning
Speaker: Ajinkya Kulkarni Date and place: September 10, 2020 at 10:30 -VISIO-CONFERENCE Abstract: At present the speaking style of the synthesized speech signal is neutral, as a result of the type of speech data used for training text-to-speech systems. Multi-speaker expressive speech synthesis is still an open problem due to the limited availability of expressive …
Jun 25
Impact of weak labels for ambient sound analysis
Speaker: Nicolas Turpault Date and place: June 25, 2020 at 10:30 -VISIO-CONFERENCE Abstract: In the domain of ambient sound analysis, many applications are dealing with unlabeled or weakly labeled data as opposed to strongly labeled data. Weak labels indicate which sound events happened in an audio clip. Strong labels indicate which sound events happened in …
Mar 12
Language and communication difficulties in children: issues and challenges
Speaker: Agnès Piquard Date and place: March 12, 2020 at 10:30 – C005 Abstract: Some language and communication difficulties such as dyslexia, developmental language disorder (DLD), deafness, can be encountered by children. Those multifaceted difficulties entail severe consequences, mostly in the school system. Because a lot of children cannot overcome their reading difficulties, their opportunities …
Jan 16
Verbal Multi Word Expressions identification on spoken transcription
Speaker: Nicolas Zampieri Date and Room: January 16, 2020 at 10:30 – C005 Abstract: Recent initiatives such as the PARSEME shared task have allowed the rapid development of Multi Word Eexpressions (MWE) identification systems. Many of those are based on recent NLP advances, using neural sequence models that take continuous word representations as input. We …
Oct 24
Use of Transfer Learning for Automatic Dietary Monitoring through Throat Microphone Recordings
Speaker: M. A. Tugtekin Turan Date: October 24, 2019 at 10:30 – B013 Abstract: Wearable devices and technologies in healthcare have been accelerating the development and integration of modern engineering approaches. Dietary monitoring is one challenging application among other healthcare services and typically performed over personal recordings. However, manual logging is highly biased and unreliable …
Oct 10
DeepLearn Lecture Series II
This second episode of the DeepLearn lecture series will feature talks by Sandipana Dowerah on “Speech Recognition and Machine Translation: From Statistical Decison Theory to Machine Learning and DeepNeural Networks” and by Adrien Dufraux on “Maths and Deep Learning”. Date and Room: October 10, 2019 at 10:30 – C005
Sep 26
DeepLearn Lecture Series I
This first instance of our lecture series from the DeepLearn 2019 summer school will feature Nicolas Furnon presenting on “Compressing Neural Networks” and Raphaël Duroselle presenting on “Representation Learning in Limited Data Settings”. Date and Room: September 26, 2019 at 10:30 – C005
Sep 17
Lead2Gold: Towards exploiting the full potential of noisy transcriptions for speech recognition
Speaker: Adrien Dufraux Paper by Adrien Dufraux, Emmanuel Vincent, Awni Hannun, Armelle Brun, Matthijs Douze, submitted to ASRU 2019 Date: Sep 05, 2019 at 10:30 – C005 tl;dr: Learn an ASR model from noisy transcriptions. At training time, we search better transcriptions by incorporating a noise model into a differentiable beam search algorithm. Abstract: The transcriptions …
