[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
[science-iwg] DIANNE - a deep learning framework on Java/OSGi
|
Hi all,
My name is Tim Verbelen, I am researcher at Ghent University and
committer of the Eclipse Concierge project. At EclipseCon Europe last
week I learned about the Eclipse Science community and within the
university I am working on a project that might be worth sharing.
DIANNE is a modular framework designed for easy and dynamic distribution
of deep learning models and procedures. It is built on OSGi, and uses
native bindings to the CPU/GPU to achieve state of the art performance.
DIANNE features:
- A web-based visual user interface for designing neural networks.
- A service model for neural networks, datasets, training procedures, etc.
- Distributed deployment of neural networks on various platforms such as
Raspberry Pi, nVidia Jetson up to GPU servers.
- Support for deep reinforcement learning, with benchmark environments
such as the Arcade Learning Environment.
More info at the website: http://dianne.intec.ugent.be/ or our github:
https://github.com/ibcn-cloudlet/dianne
Best regards,
Tim
--
Tim Verbelen
Ghent University - imec
IDLab
iGent Tower - Department of Information Technology
Technologiepark-Zwijnaarde 15, B-9052 Ghent, Belgium
T: +32 9 33 14905 ; T Secr: +32 9 33 14900
E: tim.verbelen@xxxxxxxxxxxxxx
W: IDLab.UGent.be ; W: IDLab.technology