Sequential Constant Size Compressors for Reinforcement Learning

By Linus Gisslén, Matthew D. Luciw, Vincent Graziano, J. Urgen Schmidhuber. 2011

18 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Cognitive Neuroscience

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

Citations in the corpus

Deep learning in neural networks: An overview (2014, methodology): “To deal with POMDPs (Section 6.3) involving high-dimensional inputs, RBM-based RL was used (Otsuka, 2010), and a RAAM (Pollack, 1988) (Section 5.7) was employed as a deep unsupervised sequence encoder for RL (Gisslen et al., 2011).”
On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models (2015, methodology): “A RAAM RNN [193] was employed as a deep unsupervised sequence encoder for RL [65].”
Evolving deep unsupervised convolutional networks for vision-based reinforcement learning (2014, background): “), is to combine action learning with an unsupervised learning (UL) preprocessor or “compressor” which provides a lower-dimensional feature vector that the agent receives as input instead of the raw observation [5, 8, 11, 15, 17, 18, 19].”
Online Evolution of Deep Convolutional Network for Vision-Based Reinforcement Learning (2014, background): “), is to combine action learning with an unsupervised learning (UL) preprocessor or “compressor” which provides a lower-dimensional feature vector that the agent receives as input instead of the raw observation [5, 8, 12, 17, 19–21].”
Incremental Slow Feature Analysis: Adaptive Low-Complexity Slow Feature Updating from High-Dimensional Input Streams (2012, background): “Other UL systems that both learn and apply features to sequences (Schmidhuber, 1992a,c,b; Lindstädt, 1993; KlapperRybicka et al., 2001; Jenkins and Matarić, 2004; Lee et al., 2010; Gisslen et al., 2011), thus assuming that the state of the system itself can depend on past information.”
Intrinsically motivated neuroevolution for vision-based reinforcement learning (2011, background): “TD, policy gradients, etc), where the agent simultaneously learns both the mapping from the compressed features to actions, and the features themselves from the observations provoked by the actions [1,4,11,13,14,16].”
Learning States Representations in POMDP (2013, background): “For instance, (Gisslén et al., 2011) propose a model based on a recurrent auto-associative memory with history of arbitrary depth, while (Duell et al.”
Model-based Utility Functions (2011, methodology): “There is very active research on algorithms for learning MDPs and DBNs (Baum et al, 1970; Ghahramani 1997; Sutton and Barto, 1998; Alamino and Nestor, 2006; Hutter 2009a; Hutter 2009b; Gisslén et al, 2011).”
Artificial curiosity with planning for autonomous perceptual and cognitive development (2011, background): “There are still few works where unsupervised learning (UL) methods have been com bined with RL [2], [9], [ 4 ].”