Talks

Note: a global program view can be found here. The program is still subject to minor modifications.

Wednesday 22 - Morning (9:00-13:00)

Location: Sala d'Actes - Edifici Vèrtex - Campus Nord UPC - Plaça Eusebi Güell 6, 08034 Barcelona [GMaps]

Opening:

Chair: Cristian Canton (Meta)

9:00 - Opening remarks


Session 1: 3D

Chairs: Coloma Ballester (UPF) & Albert Pumarola (Meta)

9:15 - Javier Romero (Meta) - Learning realistic human reposing using cyclic self-supervision with 3D shape, pose, and appearance consistency (CVPR 2021)

9:30 - Dan Casas (URJC) - Self-supervised collision handling via generative 3D garment models for virtual try-on (CVPR 2021)

9:45 - Enric Corona (IRI) - SMPLicit: topology-aware generative model for clothed people (CVPR 2021)

10:00 - Eduard Ramon (Crisalix/UPC) - H3D-Net: few-shot high-fidelity 3D head reconstruction (ICCV 2021)

10:15 - Guillem Brasó (TUM) - The center of attention: center-keypoint grouping via attention for multi-person pose estimation (ICCV 2021)


Session 2: Natural Language & Multi-modal

Chairs: Àgata Lapedriza (UOC) & Marta R. Costa-jussà (UPC)

11:30 - Xixian Liao (UPF) - Does referent predictability affect the choice of referential form? A computational approach using masked coreference resolution (CoNLL 2021)

11:45 - Jordi Armengol (BSC) - Are multilingual models the best choice for moderately under-resourced languages? A comprehensive assessment for Catalan (Findings-ACL 2021)

12:00 - Gemma Boleda (UPF) - Object naming: The ManyNames dataset and what it tells us about language and Computer Vision models (related to this COLING 2021)

12:15 - Amaia Salvador (Amazon) - Revamping cross-modal recipe retrieval With hierarchical transformers and self-supervised learning (CVPR 2021)

12:30 - Lucas Ventura (UPC) - How2Sign: a large-scale multimodal dataset for continuous american sign language (CVPR 2021)

12:45 - Núria Marzo (MIT) - Detecting natural disasters, damage, and incidents in the wild (ECCV 2020)

Wednesday 22 - Afternoon (14:30-19:00)

Location: Sala d'Actes - Edifici Vèrtex - Campus Nord UPC - Plaça Eusebi Güell 6, 08034 Barcelona [GMaps]

Session 3: Computer Vision & Audio

Chairs: Javier Vázquez & Dimosthenis Karatzas (UAB-CVC)

14:30 - Alejandro Hernandez Ruiz (IRI/UPC) - Neural manifolds and dynamic networks (CVPR 2021)

14:45 - Luís Herranz (CVC-UAB) - Slimmable compressive autoencoders for practical neural image compression (CVPR 2021)

15:00 - Òscar Mañas (Element AI) - Seasonal contrast: unsupervised pre-training from uncurated remote sensing data (ICCV 2021)

15:15 - Dídac Surís (Columbia) - Learning the predictability of the future (CVPR 2021)

15:30 - Jordi Pons (Dolby Labs) - Upsampling artifacts in neural audio synthesis (ICASSP 2021)

15:45 - Santiago Pascual (Dolby Labs) - Adversarial auto-encoding for packet loss concealment (WASPAA 2021)


Keynote 1 (online)

Chair: Xavi Giró (UPC-IDEAI & IRI)

17:00 - Joan Bruna (NYU) - Geometric Deep Learning: Prospects and Challenges (2021)


Careers

18:00 - Meta AI

Thursday 23 - Morning (9:30-13:00)

Thursday 23 - Afternoon (14:30-18:15)

Location: Auditori Vèrtex - Edifici Vèrtex - Campus Nord UPC - Plaça Eusebi Güell 6, 08034 Barcelona [GMaps]

Keynote 2

Chair: Gemma Boleda

14:30 - Joan Serrà (Dolby Labs) - Universal, generative speech enhancement.


Session 5: Machine Learning from NeurIPS 2021

Chair: Àgata Lapedriza (UOC)

16:30 - Roberto Dessì (Meta-UPF) - Interpretable agent communication from scratch (with a generic visual processor emerging on the side) (NeurIPS 2021)

16:45 - Pau Rodríguez (Element AI) - Continual Learning with Local Module Composition (NeurIPS 2021).

17:00 - Manel Baradad (MIT) - Learning to see by looking at noise (NeurIPS 2021)

17:15 - Víctor Garcia (UoA) - E(n) Equivariant normalizing flows (NeurIPS 2021)

17:30 - José M. Álvarez (NVidia) - Deep learning at scale (NeurIPS 2021)

17:45 - Ferran Alet (MIT) - Learning to encode and discover inductive biases (NeurIPS 2021)


Closing

Chair: Xavi Giró (UPC IDEAI & IRI)

18:00 - Closing remarks