Quick search
Advanced search in
Field to searchTerm
Add rule
Filters (7)
Download Results
24 research outcomes, page 1 of 3
  • publication . Conference object . Preprint . 2020
    Open Access
    Authors:
    Antoine Miech; Jean-Baptiste Alayrac; Lucas Smaira; Ivan Laptev; Josef Sivic; Andrew Zisserman;
    Persistent Identifiers
    Publisher: IEEE
    Project: EC | ACTIVIA (307574), EC | LEAP (336845)

    Annotating videos is cumbersome, expensive and not scalable. Yet, many strong video models still rely on manually annotated data. With the recent introduction of the HowTo100M dataset, narrated videos now offer the possibility of learning video representations without m...

    Add to ORCIDorcid
  • publication . Conference object . Preprint . 2019
    Open Access
    Authors:
    Julia Peyre; Josef Sivic; Ivan Laptev; Cordelia Schmid;
    Persistent Identifiers
    Publisher: IEEE
    Project: EC | LEAP (336845), EC | ALLEGRO (320559), EC | ACTIVIA (307574)

    International audience; We seek to detect visual relations in images of the form of triplets t = (subject, predicate, object), such as “person riding dog”, where training examples of the individual entities are available but their combinations are unseen at training. Th...

    Add to ORCIDorcid
  • publication . Conference object . Preprint . 2019
    Open Access
    Authors:
    Dimitri Zhukov; Jean-Baptiste Alayrac; Ramazan Gokberk Cinbis; David F. Fouhey; Ivan Laptev; Josef Sivic;
    Persistent Identifiers
    Publisher: Aperta
    Project: EC | ACTIVIA (307574), EC | LEAP (336845)

    In this paper we investigate learning visual models for the steps of ordinary tasks using weak supervision via instructional narrations and an ordered list of steps instead of strong supervision via temporal annotations. At the heart of our approach is the observation t...

    Add to ORCIDorcid
  • publication . Preprint . Conference object . 2018
    Open Access English
    Authors:
    Rachid Riad; Corentin Dancette; Julien Karadayi; Neil Zeghidour; Thomas Schatz; Emmanuel Dupoux;
    Persistent Identifiers
    Project: EC | BOOTPHON (295810), ANR | IEC (ANR-10-LABX-0087)

    Recent studies have investigated siamese network architectures for learning invariant speech representations using same-different side information at the word level. Here we investigate systematically an often ignored component of siamese networks: the sampling procedur...

    Add to ORCIDorcid
  • publication . Conference object . 2018
    French
    Authors:
    Hérisson, David; Paris, Clément; Deneuve, Émeline;
    Publisher: HAL CCSD
    Project: EC | ARIADNE (313193)

    National audience; Le quartier de Renancourt, situé à l’ouest de la ville d’Amiens, est connu dans la littérature archéologique depuis le début du XXe siècle par les travaux de V. Commont. Jusqu’à une date récente, ce gisement de plein air est resté l’un des rares témoi...

  • publication . Contribution for newspaper or weekly magazine . Preprint . Conference object . 2017
    Open Access English
    Authors:
    Héctor Martínez Alonso; Željko Agić; Barbara Plank; Anders Søgaard;
    Publisher: Association for Computational Linguistics
    Project: EC | LOWLANDS (313695)

    We propose UDP, the first training-free parser for Universal Dependencies (UD). Our algorithm is based on PageRank and a small set of head attachment rules. It features two-step decoding to guarantee that function words are attached as leaf nodes. The parser requires no...

    Add to ORCIDorcid
  • publication . Conference object . Preprint . 2017
    Open Access English
    Authors:
    Rahma Chaabouni; Ewan Dunbar; Neil Zeghidour; Emmanuel Dupoux;
    Persistent Identifiers
    Publisher: HAL CCSD
    Project: EC | BOOTPHON (295810), ANR | IEC (ANR-10-LABX-0087)

    Recent works have explored deep architectures for learning multimodal speech representation (e.g. audio and images, articulation and audio) in a supervised way. Here we investigate the role of combining different speech modalities, i.e. audio and visual information repr...

    Add to ORCIDorcid
  • des debats faite d’interactions dans de nombreuses arenes, faite aussi de tensions entre des debats institues et des debats ou controverses lies aux dynamiques de mobilisation des groupes concernes. Le besoin de recherches en sciences humaines et sociales est ici eviden...

    Add to ORCIDorcid
  • publication . Conference object . 2016
    Open Access English
    Authors:
    Chloé Braud; Pascal Denis;
    Persistent Identifiers
    Publisher: HAL CCSD
    Project: EC | LOWLANDS (313695), EC | STAC (269427)

    International audience; We introduce a simple semi-supervised approach to improve implicitdiscourse relation identification. This approach harnesses largeamounts of automatically extracted discourse connectives along withtheir arguments to construct new distributional w...

    Add to ORCIDorcid
  • publication . Conference object . 2016
    Open Access English
    Authors:
    Xavier Hinaut; Johannes Twiefel; Stefan Wermter;
    Persistent Identifiers
    Publisher: HAL CCSD
    Project: EC | ECHOROB (627156)

    International audience; We present a Recurrent Neural Network (RNN), namely an Echo State Network (ESN), that performs sentence comprehension and can be used for Human-Robot Interaction (HRI). The RNN is trained to map sentence structures to meanings (i.e. predicates). ...

    Add to ORCIDorcid
24 research outcomes, page 1 of 3