Nonlinear transient computation
Nonlinear transient computation (NTC) is a reservoir computing theory developed by Nigel T. Crook at Oxford Brookes University in 2007.[1] NTC is based on the exploitation of chaotic sensitivity upon initial conditions,[2] such that similar initial inputs to a NTC Machine (NTCM)[1] lead to similar responses from the machine and different initial inputs lead to different responses. This is called the Separation Property (SP), which is the first necessary and sufficient condition that allows the machine to perform computation on time-varying signals, as in a liquid state machine (LSM).[3] Since the responses from the machine reflect their inputs, then, the Approximation Property (AP) condition, which is the second necessary and sufficient condition that allows the machine to classify time varying signals as in LSM[3], takes effect and ensures that the training of the responses could be achieved. In this regard, inputs that are reflected by their responses, will be correctly classified to their desired classes. Besides those two conditions, a NTCM requires a third condition upon which its core, considered as the reservoir, would function properly. In fact, the fundamental block (i.e. the core) of a NTCM should rely on spiking neurons that embed chaotic behaviour in their dynamics.
A variety of RC architectures that implement NTC theory have been elaborated, such as the Lorentz transient computing machine[4] and the chaotic liquid state machine.[5]
References[edit]
- ↑ 1.0 1.1 Crook, Nigel (2007). "Nonlinear Transient Computation". Neurocomputing. 70: 1167-1176. doi:10.1016/j.neucom.2006.10.148.
- ↑ "Chaos Theory". Wikipedia. Wikipedia.
- ↑ 3.0 3.1 Maass, Wolfgang (2002). "Real-Time Computing Without Stable States: A New Framework for Neural Computation Based on Perturbations". Neural computation. 14 11: 2531-60. doi:10.1162/089976602760407955.
- ↑ "Lorentz Transient Computing Machine" (PDF).
- ↑ Aoun, Mario Antoine (2015). "Chaotic Liquid State Machine". International Journal of Cognitive Informatics and Natural Intelligence. 9 (4): 1-20. doi:10.4018/IJCINI.2015100101.
This article "Nonlinear transient computation" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Nonlinear transient computation. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
