Publications
This article presents our approach embracing Personalized Federated Learning (PFL) by using a unique model interpolation approach based on hierarchical clustering, achieving a significant increase in performance.
This article presents the application of the proprietary technology of dynamic graph CNNs into heterogeneous, multiprocessing systems.
This article presents a new, radical CNN dynamic pruning approach targeting to the parsimonious inference by learning to exploit and dynamically remove the redundant capacity of a CNN architecture.
This article presents the application of the process of neural plasticity to deep learning.
A summary of metrics for a CPU-only implementation of a deep-learning network on various mobile embedded systems
This article presents a new radical method able to make every CNN model many times faster!
