M
Mihai Puscas
Researcher at University of Trento
Publications - 15
Citations - 615
Mihai Puscas is an academic researcher from University of Trento. The author has contributed to research in topics: Deep learning & k-nearest neighbors algorithm. The author has an hindex of 9, co-authored 15 publications receiving 375 citations. Previous affiliations of Mihai Puscas include Huawei.
Papers
More filters
Proceedings ArticleDOI
Learning to Remember: A Synaptic Plasticity Driven Framework for Continual Learning
TL;DR: Dynamic Generative Memory (DGM) as mentioned in this paper relies on conditional generative adversarial networks with learnable connection plasticity realized with neural masking and proposes a dynamic network expansion mechanism that ensures sufficient model capacity to accommodate for continually incoming tasks.
Proceedings ArticleDOI
Unsupervised Adversarial Depth Estimation Using Cycled Generative Networks
TL;DR: Pilzer et al. as mentioned in this paper propose a deep generative network that learns to predict the correspondence field (i.e., the disparity map) between two image views in a calibrated stereo camera setting.
Posted Content
Learning to Remember: A Synaptic Plasticity Driven Framework for Continual Learning
TL;DR: DGM relies on conditional generative adversarial networks with learnable connection plasticity realized with neural masking, and a dynamic network expansion mechanism is proposed that ensures sufficient model capacity to accommodate for continually incoming tasks.
Proceedings ArticleDOI
Unsupervised Tube Extraction Using Transductive Learning and Dense Trajectories
TL;DR: The goal is to provide a method for unsupervised collection of samples which can be further used for object detection training without any human intervention, and this approach uses Dense Trajectories in order to robustly match and track candidate boxes over different frames.
Posted Content
Multimodal Prototypical Networks for Few-shot Learning
TL;DR: A generative model is trained that maps text data into the visual feature space to obtain more reliable prototypes and shows that in such cases nearest neighbor classification is a viable approach and outperform state-of-the-art single-modal and multimodal few-shot learning methods on the CUB-200 and Oxford-102 datasets.