Adaptive Nonlinear System Identification with Echo State Networks

By Herbert Jaeger. Neural Information Processing Systems, 2002

747 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/24547f720f472dd92870c1a7c4cb8bb450307f27

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

Machine Learning for Wireless Networks with Artificial Intelligence: A Tutorial on Neural Networks (2017, background): “Demonstrations of RLS for ESNs are presented in [96] and [103].”
Randomness in neural networks: an overview (2017, unknown): “To increase stability, it is possible to add a small uniform noise term to the state update before computing the non-linear transformation h(·).(78) In the second step, the current output is updated according to:”
Online Extreme Evolutionary Learning Machines (2014, unknown): “" Specifically, instead of introducing new features at random as Mahmood and Sutton have done, new features are created through a Darwinian process of replication + variation of existing features, which have been dis- 1 Echo State Networks (Becker and Obermayer, 2003) and Liquid State Machines (Maas”
On the computational power of circuits of spiking neurons (2004, methodology): “An essentially equivalent framework for real-time computing has been introduced independently by Jäger (see [17] and [18]) and applied to artificial neural network models.”
A Practical Guide to Applying Echo State Networks (2012, background): “Demonstrations of RLS for ESNs are presented in [11,43].”
Learn more by training less: systematicity in sentence processing by recurrent networks (2006, methodology): “Indeed, independently from each other, Jaeger (2003) and Maass et al. (2002) proposed a new type of recurrent neural network based on the idea that the recurrent, dynamical part of the network does not require training.”
Recent Advances in Physical Reservoir Computing: A Review (2018, background): “Time series forecasting Chaotic time series prediction (Jaeger, 2001a) NARMA time series prediction (Jaeger, 2003)”
Recurrent Network Models of Sequence Generation and Memory. (2016, methodology): “, 1988), rather than being specifically designed for one single task, have beenmodified by training to perform a variety of tasks (Buonomano and Merzenich, 1995; Buonomano, 2005; Williams and Zipser, 1989; Pearlmutter, 1989; Jaeger and Haas, 2004; Maass et al., 2002, 2007; Sussillo and Abbott, 2009;”
Non-commercial Research and Educational Use including without Limitation Use in Instruction at Your Institution, Sending It to Specific Colleagues That You Know, and Providing a Copy to Your Institution's Administrator. All Other Uses, Reproduction and Distribution, including without Limitation Comm (2005, unknown): “Later, the Cosenza group in Italy achieved some notable successes, despite the need for significant operator intervention to avoid the problem of trapping in local minima [10,11].”
Emerging opportunities and challenges for the future of reservoir computing (2024, unknown): “The state-of-the-art of RC currently can reach a word error rate (WER) of 0.4% from memristor chip RC 35 , and 0.2% from electronic RC 94 .”
2007 Special Issue: Optimization and applications of echo state networks with leaky- integrator neurons (2007, unknown): “This simple idea leads to likewise simple offline (Jaeger, 2001) and online (Jaeger, 2003) learning algorithms, sometimes amazingly accurate models (Jaeger & Haas, 2004), and may also be realized in vertebrate brains (Stanley, Li, & Dan, 1999; Mauk & Buonomano, 2004).”
Selection of cortical dynamics for motor behaviour by the basal ganglia (2015, methodology): “This output function differs from the simple tanh function used in classical echo state models (Jaeger 2002) as it takes only its positive part.”
An overview of reservoir computing: theory, applications and implementations (2007, methodology): “The Recursive Least Squares learning rule is better suited as has been shown in [19].”
Strong systematicity through sensorimotor conceptual grounding: an unsupervised, developmental approach to connectionist sentence processing (2012, methodology): “Frank (2006) proposed a number of further simulations, using both the SRN, as well as a related variant, the echo state network (ESN; Jaeger 2003).”