Leveraging unlabeled examples is a crucial issue for boosting performances in semi-supervised learning. In this work, we introduce the SAMOSA framework based on semantic augmentation for mixing semantic components from labeled examples and
The last few years have demonstrated that neural networks are remarkably good at processing and drawing conclusions from raw data like images or text. An unfortunate side effect of this is that higher level knowledge on problems and inputs
Trajectory forecasting (and planning) techniques require a good understanding of the world around them to predict their next steps. To this end, we usually have to rely on previously acquired and curated HDMaps that contain all the informat…
Federated Learning studies how models can be trained without a centralized dataset through the cooperation of multiple clients that each posess their own private dataset. To this end, the models usually train for a few epochs on their own d…
Multi-Input Multi-Output training leads to the emergence of smaller subnetworks within the base network. This has a number of benefits such a better utilization of network parameters, better training and strong regularization of learned fea…
While Mixing Samples Data Augmentation (MSDA) offer a powerful tool for regularization, they offer very little control on the content mixed. I study how manipulating the contents mixed into the final augmented image can improve the training…
Mixing Samples Data Augmentation (MSDA) help regularize models by mixing the contents of multiple images during training. I believe they provide a good opportunity to teach neural algorithms to reconcile information mixed from different sam…
Multi-input multi-output training improves network performance by optimizing multiple subnetworks simultaneously. In this paper, we propose MixViT, the first MIMO framework for vision transformers that takes advantage of ViTs’ innate mechan…
Deep architecture have proven capable of solving many tasks provided a sufficient amount of labeled data. In fact, the amount of available labeled data has become the principal bottleneck in low label settings such as Semi-Supervised Learni…
Multi-input multi-output architectures propose to train multiple subnetworks within one base network and then average the subnetwork predictions to benefit from ensembling for free. Despite some relative success, these architectures are was…
Online High Definition Map (HDMap) estimation from sensors offers a low-cost alternative to manually acquired HDMaps. As such, it promises to lighten costs for already HDMap-reliant Autonomous Driving systems, and potentially even spread th…
Multi-input multi-output models have proven capable of providing ensembling for free in convolutional neural networks by training independent subnetworks that can be ensembled. How these architectures translate to vision transformer archite…