Evolving large-scale neural networks for vision-based reinforcement learning

By Jan Koutník, Giuseppe Cuccu, J. Urgen Schmidhuber, Faustino J. Gomez. 2013

172 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Cognitive Neuroscience, Computer Vision and Pattern Recognition, Reinforcement Learning

How researchers see this paper

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

In Artificial Intelligence

In Cognitive Neuroscience

In Computer Vision and Pattern Recognition

In Reinforcement Learning

Citations in the corpus

End-to-End Training of Deep Visuomotor Policies (2015, methodology): “CNNs have also been trained to play video games with Q-learning, Monte Carlo tree search, and stochastic search (Mnih et al., 2013; Koutńık et al., 2013; Guo et al., 2014), and have been applied to simple simulated control tasks (Watter et al.”
From Pixels to Torques: Policy Learning with Deep Dynamical Models (2015, background): “r based on raw images, employing deep architectures for ﬁnding compact low-dimensional representations. Other examples of deep learning in the context of RL on image data include (Cuccu et al., 2011; Koutnik et al., 2013). These approaches have in common that they try to estimate the value function ”
Recurrent World Models Facilitate Policy Evolution (2018, background): “Evolution-based algorithms have even been able to solve difficult RL tasks from high dimensional pixel inputs [1, 31, 45, 63].”
Deep learning in neural networks: An overview (2014, background): “, 2010), and an earlier, raw videobased RL NN for computer games (Koutnı́k et al., 2013) trained by Indirect Policy Search (Sec.”
RUDDER: Return Decomposition for Delayed Rewards (2018, background): “ods provide solutions to one or more of these phases. Most prominent are Deep Q-Networks (DQNs) [81, 82], which combine Q-learning with convolutional neural networks for visual reinforcement learning [69]. The success of DQNs is attributed to experience replay [74], which stores observed statereward”
A Survey of Autonomous Driving: Common Practices and Emerging Technologies (2019, methodology): “Furthermore, these design philosophies are realized with two alternative approaches: modular [15], [47], [51]–[58] or endto-end driving [59]–[67].”
Multiagent cooperation and competition with deep reinforcement learning (2015, background): “Recently, however, the Swiss AI Lab IDSIA [8] and Google DeepMind [7, 9] have produced spectacular results in applying reinforcement learning to very high-dimensional and complex environments such as video games.”
Learning to Navigate in Complex Environments (2016, background): “Recurrent networks have also been successfully incorporated to enable state disambiguation in partially observable environments (Koutnk et al., 2013; Hausknecht & Stone, 2015; Mnih et al., 2016; Narasimhan et al., 2015).”
Evolution Strategies as a Scalable Alternative to Reinforcement Learning (2017, methodology): “Koutník et al. [2013, 2010] and Srivastava et al. [2012] have similarly applied an an ES method to RL problems with visual inputs, but where the policy was compressed in a number of different ways.”
On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models (2015, methodology): “In very general early work, we used universal assembler-like languages to encode NNs [235], later coefficients of a Discrete Cosine Transform (DCT) [132].”
A Brief Survey of Deep Reinforcement Learning (2017, neutral): “eschewed the commonly used backpropagation algorithm in favour of evolutionary algorithms [32, 19, 53], which are gradient-free policy search algorithms.”
Analyzing Classifiers: Fisher Vectors and Deep Neural Networks (2015, background): “While much of research is devoted to extending the applicability of DNNs to more domains [9, 11, 12, 13, 31], we focus here on a different question, namely the impact of context, and the ability to use context.”
One Big Net For Everything (2018, methodology): “The search space can often be reduced dramatically by evolving compact encodings of RNNs, e.g., [67, 87, 32, 94][76, Sec.”
Query-Efficient Imitation Learning for End-to-End Simulated Driving (2017, methodology): “We empirically evaluate the proposed SafeDAgger using TORCS (tor accessed May 12 2016), a racing car simulator, which has been used for vision-based autonomous driving research in recent years (Koutník et al. 2013; Chen et al. 2015).”
Query-Efficient Imitation Learning for End-to-End Autonomous Driving (2016, methodology): “We empirically evaluate the proposed SafeDAgger using TORCS [1], a racing car simulator, which has been used for vision-based autonomous driving research in recent years [9, 6].”