TO FEW-SHOT IMITATION:
DEMONSTRATION-CONDITIONED REINFORCEMENT LEARNING
INTERSPEECH 2021, hybrid event, Brno, Czech Republic
International Conference on Machine Learning (ICML) 2021
International ACM SIGIR Conference on Research and Development in Information Retrieval 2021
LeBenchmark: a reproducible framework for assessing self-supervised representation learning from speech
A new approach to learning few-shot imitation agents whereby you simply feed demonstrations of a new test task to the learned policy called DCRL. This new approach has several advantages.
A new sparse bi-encoder BERT-based model for effective and efficient first-stage ranking. The first to rival dense models.
Using domain randomization and meta-learning, computer vision models forget less when exposed to training samples from new domains. Remembering is a crucial element in the deployment of self-driving cars and robots which interact in dynamic environments.
GLOBAL AI R&D BELT
ACADEMIA – EU/GOVT – ENTREPRENEURS
Our partnerships range from long-term fundamental research to investment in products and services.