Neuroevolution

By Joel Lehman, Risto Miikkulainen. 2010

73 citations recorded by the source corpus.

Neuroevolution is a method for modifying aspects of neural network design in order to learn a specific task. Evolutionary computation is used to discover designs that maximize a fitness function that measures performance in the task. Compared to other neural network learning methods, neuroevolution is highly general, allowing learning without explicit targets, and modifying both differentiable and nondifferentiable aspects of the design, such as the architecture, weights, activation and loss functions, and learning algorithms. Neu-roevolution thus serves three roles: First, it is a policy search method for reinforcement learning problems, where it is well suited to continuous domains and to domains where the state is only partially observable. Second, it is an automatic method for discovering effective deep learning architectures. Third, especially when combined with learning in individual networks, it is a model of biological adaptation

Subfields: Molecular Biology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

1. Risto Miikkulainen
2. Peter Stone
3. Matthew J. Hausknecht
4. Joel Lehman

Citations in the corpus

• Evolving Deep Neural Networks (2017, methodology): “Neuroevolution techniques have been shown to work well in many tasks in control, robotics, constructing intelligent agents for games, and artificial life [31].”
• A Neuroevolution Approach to General Atari Game Playing (2014, background): “Neuro-evolution algorithms [13, 32] have been successful on a variety of tasks including quadruped locomotion [8, 50], evolution of morphology [6], design of robot morphology and control [29], checkers [14], Unreal Tournament [38], and game playing [5, 47].”
• Conversion rate optimization through evolutionary computation (2017, background): “Neuroevolution [6, 10] can discover optimal such mappings, in effect evolve to discover a dynamic, continuous segmentation of the user space.”
• Sentient Ascend: AI-Based Massively Multivariate Conversion Rate Optimization (2018, background): “Neuroevolution (Lehman and Miikkulainen 2013b; Floreano, D¨urr, and Mat-tiussi 2008) can discover optimal such mappings, in effect evolve to discover a dynamic, continuous segmentation of the user space.”