Quantitative measurement of uncertainties in artificialintelligence-guided simulation

CEA-List’s probabilistic deep learning tools can be used to quantitatively measure prediction reliability.

PyRAT wins formal verification competition

CEA-List researchers developed PyRAT, a formal verification tool for neural networks, to respond to growing demand for more reliable AI-based systems.

August 29, 2024 | A safety supervision environment for autonomous systems

CEA-List has developed a runtime safety supervision environment for autonomous systems built using AI.

July 30, 2024 | AI robustness and safety characterization software

CAISAR (Characterizing Artificial Intelligence Safety and Robustness) is an end-to-end open source software environment for AI system specification and verification.

October 18, 2022 | Deep learning for non-specialists now possible with transfer learning

CEA-List recently came up with a method for selecting the most suitable existing neural network for adaptation and reuse for new target applications. This advance makes it possible to train classification neural networks, or CNNs, without subject matter expertise or large datasets.

N2D2

N2D2 is used to optimize neural networks, embed them on components or on dedicated hardware accelerators, and measure their performance.

October 05, 2021 | Safeguarding the data used to train artificial neural networks

June 23, 2021 | Cost-effective training of neural networks