DIPVIEW
Reducing the energy cost/carbon footprint of training deep learning models
Benefits
- Automation of the training parallelization of CNNs
- Significant acceleration of the training process
- Decrease in AI frugality
Key words
- Deep learning
- Distributed system
- Applied mathematics
- Medical applications
Intellectual Property
- 1 patent
- 1 software
Laboratory
- LIG
Institutions
- CNRS
- GRENOBLE INP-UGA
- INRIA
- UGA
Linksium Continuum
- Maturation
Context
The objective of the project is the creation of a start-up that aims to offer products and services to optimise and accelerate deep learning on parallel and heterogeneous architectures. This will be achieved by implementing a recommender that reduces the energy cost/carbon footprint of training deep learning models.
Technology
The Dipview project consists of two main components:
- The Auto-CNNp (Automatic CNN parallelization) framework based on software components that automate the process of training convolutional neural networks (CNNs).
- A multi-criteria recommender for the ideal configuration of resources for deep learning.
Advantages
- Automation of the training parallelization of CNNs
- Significant acceleration of the training process
- Decrease in AI frugality
State of progress
A proof of concept (POC) of the Auto-CNNp solution on medical applications (COVID,...) has been implemented, evaluated and validated over TensorFlow and Horovod.
Applications
- Medical for the diagnosis of various pathologies (brain tumours)
- Industrial Robotics
- Autonomous cars
- Training and retraining in deep learning
Featured articles
1, 2 and 3 startups for Nora Dempsey, National Medal of Innovation 2021
Nora Dempsey is one of the four winners of National Medal for Innovation 2021 awarded by CNRS. Nora Dempsey, born in Ireland, has gained an international recognition for her research in Grenoble on...
Read the articleAmiral Technologies, la startup IA grenobloise spécialisée dans les solutions de maintenance prédictive, intègre Station F
An ambitious promise: to invent a zero-defect and zero-downtime future for industry. Amiral Technologies, a startup co-founded and boosted by Linksium, based on the technological prowess of the Gre...
Read the article