Robust non-linear control through neuroevolution

By Faustino J. Gomez, Risto Miikkulainen. 2003

182 citations recorded by the source corpus.

How researchers see this paper

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

Citations in the corpus

On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models (2015, methodology): “1 RL through Direct and Indirect Search in RNN Program Space It is possible to train small RNNs with a few 100 or 1000 weights using evolutionary algorithms [200, 255, 105, 56, 68] to search the space of NN weights [165, 307, 44, 321, 180, 259, 320, 164, 173, 69, 71, 187, 121, 313, 66, 270, 269, 305”
One Big Net For Everything (2018, background): “Without a teacher, reward-maximizing programs of an RNN must be learned through repeated trial and error, e.g., through artiﬁcial evolution [39, 107, 83, 40, 16, 18, 103, 14, 89, 88] [76, Sec.”
Real-time neuroevolution in the NERO video game (2005, methodology): “While methods for partially observable Markov processes exist, significant challenges remain in scaling them up to real-world tasks [32].”
Co-evolving recurrent neurons learn deep memory POMDPs (2005, methodology): “H-ESP is a new neuroevolution algorithm based on Enforced SubPopulations (ESP; [4]).”
Reinforcement Learning Upside Down: Don't Predict Rewards - Just Map Them to Actions (2019, background): “RL is also very different from traditional black box optimization (BBO) [36, 57, 19, 9] such as neuroevolution [29, 59, 14, 12] which can be used to solve complex RL problems in partially observable environments [15] through iterative discovery of better and better parameters of an adaptive…”
Intrinsically motivated neuroevolution for vision-based reinforcement learning (2011, background): “INTRODUCTION Neuroevolution, the artificial evolution of neural net works [2,27], has advantages over classical, single-agent learn ing methods for continuous reinforcement learning tasks that require memory [6,22].”
Evolino for recurrent support vector machines (2005, methodology): “In this study, Evoke is instantiated using Enforced SubPopulations (ESP; [10]) to evolve Long Short-Term Memory (LSTM; [11]) networks.”
Evolving Soccer Keepaway Players Through Task Decomposition (2005, methodology): “Neuroevolution has repeatedly been shown to be an effective reinforcement learning method for non-linear control tasks (Gruau et al., 1996; Moriarty and Miikkulainen, 1996; Stanley and Miikkulainen, 2004; Gomez, 2003).”
Driven by Compression Progress: A Simple Principle Explains Essential Aspects of Subjective Beauty, Novelty, Surprise, Interestingness, Attention, Curiosity, Creativity, Art, Science, Music, Jokes (2008, methodology): “Many approaches to evolving RNNs have been proposed [34, 111, 110, 37, 33, 93, 35, 15 ].”
Evolutionary Bilevel Optimization for Complex Control Tasks (2015, background): “However, many modern control problems involve a large number of state of variables that interact in a nonlinear manner, making it difficult to apply classical methods to them [2].”
Cartesian genetic programming encoded artificial neural networks: a comparison using three benchmarks (2013, unknown): “Using these parameters ninety-two of Table 1: Comparison of Results for the Double Pole Balancing Benchmark Method Evaluations Std. Dev Avg ed Over EP [45] 307200 - - CE [8] 34000 - >30 CNE [7] 22100 - 50 EuSAIN [35] .”
Active Guidance for a Finless Rocket Using Neuroevolution (2003, methodology): “Enforced Subpopulations1 [5,6] is a neuroevolution method that extends the Symbiotic, Adaptive Neuroevolution algorithm (SANE; [7]).”
Accelerated Neural Evolution through Cooperatively Coevolved Synapses (2008, background): “…(Perez- Bergquist, 2001; Lubberts and Miikkulainen, 2001; Greer et al., 2002; Whiteson et al., 2003; Bryant and Miikkulainen, 2003; Gomez et al., 2001; Grasemann and Miikkulainen, 2005), including the real world reinforcement learning task of finless rocket control (Gomez and Miikkulainen, 2003).”