The paper is co-authored by Jianyuan Zhong, @santty128, Pawel Swietojanski, Joao Monteiro, Jan Trmal, Yoshua Bengio. The work was started at JSALT 2019 (organized by @jhuclsp)
Prikaži ovu nit
I'm happy to announce our latest work on self-supervised learning for #speech. PASE+ is based on a multi-task approach useful for #speech recognition. It will be presented at #ICASSP2020.
paper: https://arxiv.org/abs/2001.09239
code: https://github.com/santi-pdp/pase
@Mila #deeplearning #AIpic.twitter.com/ynWReEHGn4
-
-
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
PASE+ is an improved version of PASE for robust speech recognition. We employ an online speech distortion module, a revised encoder that better learns short- and long-term speech dynamics, and a refined the set of workers that encourage better cooperation.
#ICASSP2020#ASRPrikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
I'm a big fan of self-supervised deep learning. I think it will be widely adopted in the future in many fields, including speech. Maybe this will contribute to "make pre-training great again"! hahaha
#ICASSP2020#DeepLearning#AI#machinelearning#speech#audio#NeuralNetworksPrikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
Awesome - we have quite a few papers from our team as well. See you at
#ICASSP2020 -
Thank you Vivek! See you in Barcellona
Kraj razgovora
Novi razgovor -
-
-
Can we use it for cochlear implant processing? I am working on it
-
I'm not familiar with this task. The idea is to learn a general speech representation that is useful for many applications, but I don't know if cochlear implant processing could be one of these tasks.
- Još 1 odgovor
Novi razgovor -
-
-
Nice, more minions are added!! I am curious if self-sup. lear. can be used as a method for robustifying the encoder against adversarial attacks? The input can be distorted by adv pert. by fooling one of the workers. Can the other workers correct the infected one?
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
Great work, can’t wait to test it on my projects! And thanks again for sharing the code with the community
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.