An overview of reservoir computing: theory, applications and implementations

By Benjamin Schrauwen, David Verstraeten, J. Van Campenhout. 2007

528 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.

1. Yoshua Bengio
2. Sergey Levine
3. Mohit Bansal
4. Zhe Gan
5. Trevor Darrell
6. Ali Farhadi
7. Quoc V. Le
8. Dacheng Tao

Citations in the corpus

• Deep learning in neural networks: An overview (2014, background): “Certain SL RNNs with fixed weights for all connections except those to output units (Jaeger, 2001; Maass et al., 2002; Jaeger, 2004; Schrauwen et al., 2007) have a maximal problem depth of 1, because only the final links in the corresponding CAPs are modifiable.”
• Supermasks in Superposition (2020, background): “With random features, training a linear head on fixed features is considered in the literature of reservoir computing [42], and more [1].”
• Making AI Forget You: Data Deletion in Machine Learning (2019, background): “Linearity should be most effective in domains in which randomized [70], reservoir [89, 76], domain-specific [54], or pre-trained feature spaces elucidate linear relationships in the data.”
• Reservoir computing approaches to recurrent neural network training (2009, background): “A brief and broad overview of reservoir computing is presented in [60], with an emphasis on”
• Neural networks and neuroscience-inspired computer vision. (2014, background): “For instance, a conceptually simple framework known as ‘reservoir’ computing [91] (Figure 3A), which uses unstructured, randomly connected recurrent networks, paired with a simple linear read-out, has been applied to a variety of temporal recognition problems outside of vision with a surprising…”
• Training BatchNorm and Only BatchNorm: On the Expressive Power of Random Features in CNNs (2020, background): “To fit sequential data, work on reservoir computing (Schrauwen et al., 2007), also known as echo state networks (Jaeger, 2003) or liquid state machines (Maass et al., 2002), learns the linear output layer of a recurrent neural network of random features; such models are capable of behaving in a…”
• Recent Advances in Physical Reservoir Computing: A Review (2018, background): “They are different from conventional RNNs in that the weights on the recurrent connections in the reservoir are not trained but only the weights in the readout are trained (Lukoševičius & Jaeger, 2009; Schrauwen et al., 2007).”
• Brain-Inspired Classroom Occupancy Monitoring on a Low-Power Mobile Platform (2014, background): “Important academic research lines focusing on computation (as opposed to the simulation of the brain) include liquid state machines [20] and reservoir computing [23][22].”
• Introduction to spiking neural networks: Information processing, learning and applications. (2011, background): “…Reservoir computing – is a computational concept that takes advantage of properties of recurrent networks, while avoiding difficulties associated with their training (Atiya and Parlos 2000, Dominey and Ramus 2000, Jaeger 2001, Maas et al. 2002a, Schrauwen et al. 2007, Lukosevicius and Jaeger 2009).”
• Tensor Programs I: Wide Feedforward or Recurrent Neural Networks of Any Architecture are Gaussian Processes (2019, methodology): “Reservoir Computing In reservoir computing [30, 39, 51], sequence processing is typically done by a randomly initialized recurrent neural network.”
• Minimum Complexity Echo State Network (2011, background): “A class of approaches referred to as reservoir computing tries to avoid this problem by fixing the state-transition function—only the readout is fitted to the data [2], [32].”
• Investigation of financial market prediction by recurrent neural network (2011, background): “Investigated network reached good learning and still makes very accurate predictions [1], [10-12].”
• Modeling Activity-Dependent Plasticity in BCM Spiking Neural Networks With Application to Human Behavior Recognition (2011, unknown): “Together with the echo state networks [4], LSMs have now become one of the two main streams in reservoir computing [5].”
• Model learning for robot control: a survey (2011, background): “However, there are attempts to develop neural networks that can change their structures dynamically, such as in reservoir computing (Schrauwen et al. 2007; Lukocevicius and Jaeger 2009; Reinhart and Steil 2009) and echo state neural networks (Steil 2007; Joshi and Maass 2005; Rolf et al.”