LioNets on Time Series

LioNets technique applied to the Turbofan Engine Degradation Simulation dataset (time-series data) LioNets on Time Series

Jupyter Notebook

Lio Nets
@ Intelligent Systems Lab - Ioannis Mollas

License
GNU General Public License (GPL) v3

Main Characteristic

In this asset, the code of LioNets (v2.0) is supporting two python notebooks, which present two ways to interpret a binary classifier and a remaining useful lifetime predictor on an engine degradation dataset from NASA.

Research areas
Explainable AI
Technical Categories
Machine learning Robotics and automation
Business Categories
Manufacturing
Last updated
31.05.2021 - 08:27
Detailed Description

Additional information: LioNets is a methodology on providing explanations for a neural network's decisions, in a local scope, through a process that actively takes into consideration the neural network's architecture on creating an instance's neighbourhood, that assures the adjacency among the generated neighbours and the instance. LioNets (v1.0) is available on this platform.

In the current asset, the code of LioNets (v1.5) is used to support two python notebooks, which present two ways to interpret a binary classifier and a remaining useful lifetime predictor on the Turbofan Engine Degradation Simulation Dataset [1]. This work is an essential setup, to be prepared for the AI4Robotics dataset. When this dataset is to be published, it will also be adapted to the exact same methodology presented in this asset.

External links:

Trustworthy AI
The interpretation technique implemented in LioNets is intended to provide explanations to neural-based classifier, towards a trustworthy machine learning component. LioNets actively takes into consideration the internal structure of the model it tries to explain, providing that way better adjacency among the generated neighbours and the instance.
GDPR Requirements
LioNets component is GDPR compliant (Articles 13–15) because it is providing a way to interpret the decision of a neural network in a form of feature importances. This approach addresses the "explicability" requirement of the GDPR, where a requirement is fixed for automated decision processes that have an impact on humans.