Description du projet
Étendre l’apprentissage profond jusqu’au niveau méta
L’apprentissage profond est un sous-ensemble de l’apprentissage automatique qui apprend aux ordinateurs à tirer des conclusions, comme le feraient les humains, en analysant des données de manière continue. Ce domaine a permis d’accomplir des progrès remarquables en matière de vision par ordinateur, de reconnaissance vocale, de traitement du langage naturel et d’apprentissage par renforcement. Le projet DeepLearning 2.0 financé par l’UE, entend faire passer l’apprentissage profond au niveau supérieur. Il proposera des méthodes d’apprentissage de niveau méta qui améliorent les résultats et les performances des algorithmes d’apprentissage. Ces nouvelles méthodes permettront de créer de nouveaux pipelines d’apprentissage profond personnalisés, sans éléments intégrés manuellement, qui seront plus précis, plus simples à utiliser et dont l’entraînement prendra moins de temps. Pour démontrer la viabilité des méthodes proposées, les chercheurs implémenteront les nouveaux pipelines d’apprentissage profond personnalisés dans le décryptage des électroencéphalogrammes et le pliage de l’ARN.
Objectif
Deep learning has revolutionized many fields, such as computer vision, speech recognition, natural language processing, and reinforcement learning. This success is based on replacing domain-specific hand-crafted features with features that are learned for the particular task at hand. The logical step to take deep learning to the next level is to also (meta-)learn other hand-crafted elements of the deep learning pipeline. We therefore propose to develop meta-level learning methods for the creation of novel customized deep learning pipelines, by means of:
1. Hierarchical neural architecture searchfor learning qualitatively new architectures and architectural building blocks from scratch;
2. Learning of optimizers and hyperparameter adaptation policies that adapt totheir context in order to converge faster and more robustly;
3. Learning the data to train on, to remove the need for large sets of labelled data; and
4. Bootstrapping from prior design efforts to increase efficiency and make an integrative design of architectures, optimizers, hyperparameter adaptation policies, and pretraining tasks feasible in practice.
These advances will allow the next generation of deep learning pipelines to achieve higher accuracy, lower training time, and improved ease-of-use (democratization of deeplearning). They will also allow a customization to particular design contexts, including additional objectives next to accuracy (such as robustness, memory requirements, energy consumption, latency, interpretability, training cost, uncertainty estimation, and algorithmic fairness) in order to facilitate trustworthy AI. In order to demonstrate the effectiveness of these methods, we plan to develop:
5. New state-of-the-art customized deep learning pipelines for various applications, including EEG decoding, RNA folding, and improving the reinforcement learning pipeline and deep learning on tabular data.
Champ scientifique
Programme(s)
- HORIZON.1.1 - European Research Council (ERC) Main Programme
Régime de financement
HORIZON-AG - HORIZON Action Grant Budget-BasedInstitution d’accueil
79098 Freiburg
Allemagne