Institution:
Distributed Computing Library (dislib)
A classical machine learning library developed on top of PyCOMPSs.
Research Group:
Workflows and Distributed Computing
Researcher/s:
Rosa Maria Badia
Website:
https://dislib.readthedocs.io/en/release-0.9/
Description:
A classical machine learning library developed on top of PyCOMPSs.
Type of asset:
Service
Category:
Computer Sciences
Problem:
Making easier the development of applications in parallel and distributed platforms.
Solution:
dislib is a machine learning library parallelized with PyCOMPSs that offers a simple fit-predict interface but it is able to execute the methods in parallel. Based on a distributed array and on an out-of-core methodology, supports the execution for large-data sets not fitting in the memory of the system.
Aplication areas:
dislib has been applied in use cases of astrophysics, molecular dynamic workflows. It is currently being applied in use cases for urgent computing for natural hazards, in digital twins for manufacturing and in distributed training of neural networks.
Novelty:
The main novelty is the parallelization of the traditional methods in a transparent way to the application developer.
Protection:
Open source (apache v2).
Target market:
Any interested in using our technologies.
Keywords:
Workflows, parallel computing, distributed computing, HPC+AI convergence, machine learning
TRL: 7
CRL: N/A
BRL: N/A
IPRL: N/A
TmRL: N/A
FRL: N/A
More information
if you want to know more about this project do not hesitate to contact us