NeuroEvolution Method

Neuroevolution is a powerful technique that uses evolutionary algorithms to automatically design and optimize neural network architectures, addressing the challenges of manual design and hyperparameter tuning. Current research focuses on applying neuroevolution to diverse problems, including anomaly detection (using models like AD-NEv++), semi-supervised learning (leveraging neuron coverage metrics), and side-channel analysis (with NASCTY-CNNs), often incorporating techniques like subspace evolution and ensemble learning to improve efficiency and performance. These advancements are significant because they offer a more automated and potentially more effective approach to building neural networks across various domains, leading to improved model accuracy and robustness in applications ranging from cybersecurity to environmental monitoring.

Papers

March 25, 2024

AD-NEv++ : The multi-architecture neuroevolution-based multivariate anomaly detection framework
Marcin Pietroń, Dominik Żurek, Kamil Faber, Roberto Corizzo
NeuroEvolution Method NeuroEvolution Algorithm

March 5, 2023

Neuroevolutionary algorithms driven by neuron coverage metrics for semi-supervised classification
Roberto Santana, Ivan Hidalgo-Cenalmor, Unai Garciarena, Alexander Mendiburu, Jose Antonio Lozano
Semi Supervised Semi Supervised Classification Neuron Coverage NeuroEvolution Method

February 12, 2023

Sparse Mutation Decompositions: Fine Tuning Deep Neural Networks with Subspace Evolution
Tim Whitaker, Darrell Whitley
Deep Neural Network Fine Tuning NeuroEvolution Method

January 25, 2023

NASCTY: Neuroevolution to Attack Side-channel Leakages Yielding Convolutional Neural Networks
Fiske Schijlen, Lichao Wu, Luca Mariot
Side Channel Evolutionary Operator NeuroEvolution Method

October 6, 2022

Neuroevolution is a Competitive Alternative to Reinforcement Learning for Skill Discovery
Felix Chalumeau, Raphael Boige, Bryan Lim, Valentin Macé, Maxime Allard, Arthur Flajolet, Antoine Cully, Thomas Pierrot
Reinforcement Learning Deep Reinforcement Learning Skill Discovery Hierarchical Planning Efficient Alternative NeuroEvolution Method

August 23, 2022

Neuroevolution-based Classifiers for Deforestation Detection in Tropical Forests
Guilherme A. Pimenta, Fernanda B. J. R. Dallaqua, Alvaro Fazenda, Fabio A. Faria
Deforestation Area Deforestation Detection NeuroEvolution Method

August 4, 2022

Modular Grammatical Evolution for the Generation of Artificial Neural Networks
Khabat Soltanian, Ali Ebnenasir, Mohsen Afsharchi
Neural Network Faithful Generation Structured Neural Network NeuroEvolution Method Grammatical Evolution

July 12, 2022

Adversarial Robustness Assessment of NeuroEvolution Approaches
Inês Valentim, Nuno Lourenço, Nuno Antunes
Adversarial Attack Adversarial Example Adversarial Robustness NeuroEvolution Method

December 10, 2021

Fast and scalable neuroevolution deep learning architecture search for multivariate anomaly detection
M. Pietroń, D. Żurek, K. Faber, R. Corizzo
Anomaly Detection Evolutionary Algorithm Architecture Search Model Evolution NeuroEvolution Method Hardware Accelerated NeuroEvolution

NeuroEvolution Method

Papers

AD-NEv++ : The multi-architecture neuroevolution-based multivariate anomaly detection framework

Neuroevolutionary algorithms driven by neuron coverage metrics for semi-supervised classification

Sparse Mutation Decompositions: Fine Tuning Deep Neural Networks with Subspace Evolution

NASCTY: Neuroevolution to Attack Side-channel Leakages Yielding Convolutional Neural Networks

Neuroevolution is a Competitive Alternative to Reinforcement Learning for Skill Discovery

Neuroevolution-based Classifiers for Deforestation Detection in Tropical Forests

Modular Grammatical Evolution for the Generation of Artificial Neural Networks

Adversarial Robustness Assessment of NeuroEvolution Approaches

Fast and scalable neuroevolution deep learning architecture search for multivariate anomaly detection