An analog Hopfield type neural network is given, that identifies the K largest components of a list of d of N real numbers. The neurons are identical, with a tanh characteristic, and the weightâ€¦ (More)

A new dynamical system and computational circuit is described and analyzed. The dynamics permits the construction of a Lyapunov function that ensures global convergence to a unique stableâ€¦ (More)

In a previous work, the authors proposed an analog Hopfield-type neural network that identified the K largest components of a list of real numbers. In this work, we identify computable restrictionsâ€¦ (More)

The classical continuous time recurrent (Hopfield) network is considered and adapted to <i>K</i> -winner-take-all operation. The neurons are of sigmoidal type with a controllable gain <i>G</i>, anâ€¦ (More)

A continuous time neural network built with nonlinear amplifiers which selects the largest item of a list (WTA) is considered. The network receives and processes lists admitted one by one. If theâ€¦ (More)

2012 IEEE 55th International Midwest Symposium onâ€¦

2012

A well known circuit - due to Lazzaro - for WTA neural applications is considered. It is a current mode low voltage network based on subthreshold MOS devices. The authors show that the circuit modelâ€¦ (More)

663 points per period still fails to capture the full bandwidth needed for an accurate solution estimate. Interpolation of the Gear-2 solution was used to produce a 60-point starting estimate withâ€¦ (More)

The 2011 International Joint Conference on Neuralâ€¦

2011

The paper shows rigorously how to build a KWTA selector from a classical neural Hopfield network in continuous time. The analytical relations between parameters result in a step-by-step accurate andâ€¦ (More)