Michail Zak

TL;DR: In this paper, a new type of attractors for an addressable memory in neural networks operating in continuous time is introduced, called terminal attractors, which represent singular solutions of the dynamical system and intersect the families of regular solutions while each regular solution approaches the terminal attractor in a finite time period.

TL;DR: It will be shown that terminal attractors can be incorporated into neural networks such that any desired set of these attractors with prescribed basins is provided by an appropriate selection of the synaptic weights.

TL;DR: In this paper, a general method of anomaly detection from time-correlated sensor data is disclosed, which is applicable to a broad class of problems and is designed to respond to any departure from normal operation, including faults or events that lie outside the training envelope.

TL;DR: A powerful neural learning formalism is introduced for addressing a large class of nonlinear mapping problems, including redundant manipulator inverse kinematics, commonly encountered during the design of real-time adaptive control mechanisms.

TL;DR: A generalized formalism for diagnostics (206, 208) and prognostics (214) in an instrumented system which can provide sensor data and discrete system variable takes into consideration all standard forms of data, both time-varying such as sensor or extracted feature, quantities and other autonmy-enabling components such as planners and schedulers.