1042 results about "Non linear functions" patented technology

New and efficient protocols are provided for privacy-preserving machine learning training (e.g., for linear regression, logistic regression and neural network using the stochastic gradient descent method). A protocols can use the two-server model, where data owners distribute their private data among two non-colluding servers, which train various models on the joint data using secure two-party computation (2PC). New techniques support secure arithmetic operations on shared decimal numbers, and propose MPC-friendly alternatives to non-linear functions, such as sigmoid and softmax.

A method for predicting future states of a vehicle including the steps of selecting a model having n states reflecting dynamic features of the vehicle; inputting noisy sensor measurements representing a current state of the vehicle to generate (2n+1) sigma points Xi where i=0, . . . . 2n, each of the sigma points having n states; performing (2n+1) integrations, each integration includes propagating the n-states of the respective sigma points Xi through the non-linear function Yi=f(Xi); and combining the propagated sigma points to generate the predicted future states of the vehicle.

Provided is a method for estimating carrier wave or doppler frequency offset based on an approximated quadratic non-linear function in fine search, and a receiver employing the method. The method for estimating frequency offset includes the steps of: a) extracting more than three discrete samples in an order where the discrete samples are close to a maximum value of a power spectrum density function; b) inducing a quadratic non-linear function by using the extracted discrete samples; and c) estimating the frequency offset from a maximum point of the quadratic non-linear function. The technology of the present research makes it possible to exactly estimate frequency offset without increasing hardware complexity in estimation of carrier wave or doppler frequency offset.

A method and system for full-field fringe-projection for 3-D surface-geometry measurement, referred to as “triangular-pattern phase-shifting” is disclosed. A triangular grey-scale-level-coded fringe pattern is computer generated, projected along a first direction onto an object or scene surface and distorted according to the surface geometry. The 3-D coordinates of points on the surface are calculated by triangulation from distorted triangular fringe-pattern images acquired by a CCD camera along a second direction and a triangular-shape intensity-ratio distribution is obtained from calculation of the captured distorted triangular fringe-pattern images. Removal of the triangular shape of the intensity ratio over each pattern pitch generates a wrapped intensity-ratio distribution obtained by removing the discontinuity of the wrapped image with a modified unwrapping method. Intensity ratio-to-height conversion is used to reconstruct the 3-D surface coordinates of the object. Intensity-ratio error compensation involves estimating intensity-ratio error in a simulation of the measurement process with both real and ideal captured triangular-pattern images obtained from real and ideal gamma non-linearity functions. A look-up table relating the measure intensity-ratio to the corresponding intensity-ratio error is constructed and used for intensity-ratio error compensation. The inventive system is based on two-step phase-shifting but can be extended for multiple-step phase-shifting.

Signals are separated by introducing a function having a monotonously increasing characteristic like an exponential type function as a cost function, and applying an adaptive algorithm that minimizes that cost function in terms of a signal separation matrix. That is, there is provided a signal processing apparatus that separates and outputs an original signal from the observed signal x(t), in which multiple multidimensional signals are mixed, wherein the nonlinear function 21 is operated on an input observed signal x(t) and an estimated separation matrix W(t−1) estimated at a previous cycle. Then, an error signal e(t) is calculated 22 based on y(t) formed by this nonlinear function 21, the estimated separation matrix W(t−1) estimated at the previous cycle, and the observed signal x(t) at that time. Then, based on the calculated error signal e(t), the update of the separation matrix W(t) at that time is performed 23 such that consideration weight is increased when estimation errors are large using the cost function having a monotonously increasing characteristic.

Methods and systems are provided for coarse phase estimation for highly-spectrally efficient communications. An example method may include, equalizing, in a receiver, a received inter-symbol correlated (ISC) signal to generate an equalized ISC signal. A phase adjustment signal may be generated based on an ISC feedback signal. The ISC feedback signal may be generated using a sequence estimation process and a non-linearity function. A phase of the equalized ISC signal may be adjusted using the generated phase adjustment signal, to generate a phase adjusted partial response signal. The phase adjustment signal may be generated based on a phase difference between the equalized ISC signal and the partial response feedback signal. At least one ISC vector may be generated by buffering samples of the phase adjusted ISC signal.