In an embodiment, PhotoPlethysmoGraphy (PPG) signals are processed by detecting peaks and valleys in the PPG signal, segmenting the PPG signal to provide a time series of PPG waveforms located between two subsequent valleys in the PPG signal, applying to the waveforms in the time series pattern recognition with respect to a reference PPG waveform pattern produced based on a mathematical model of the PPG signal by assigning to the waveforms in the time series a recognition score. A resulting PPG signal is produced by retaining the waveforms in the time series having an assigned recognition score reaching a recognition threshold, and discarding the waveforms in the time series having an assigned recognition score failing to reach the recognition threshold.

A matrix-multiplying-matrix operation method and a processing device for performing the same are provided. The matrix-multiplying-matrix method includes distributing, by a main processing circuit, basic data blocks of one matrix and broadcasting the other matrix to a plurality of the basic processing circuits. That way, the basic processing circuits can perform inner-product operations between the basic data blocks and the broadcasted matrix in parallel. The results are then provided back to main processing circuit for combining. The technical solutions proposed by the present disclosure provide short operation time and low energy consumption.

There is provided a method for determining presence of a person comprising a) receiving IR sensor data (50) during a first time period from a thermopile and using the IR sensor data to determine an IR background signal baseline (51) for the time period, and determining a variability of the IR sensor data (50), b) using the IR background signal baseline (51) and the variability of the IR background signal level to determine a threshold (52) with a value higher than the background signal baseline (51), and in such a way so that greater variability in the IR background signal (50) results in a higher threshold (52), then c) receiving further IR sensor data (50) during a second time period, which is after the first time period, and using the further IR sensor data (50), and the threshold (52) determined in step b) to determine that a person is present when the further IR sensor data (50) comprises a value that is higher than the threshold.

Product demand forecasting accuracy utilizes partitional clustering of time series data with dynamic time warping. The product demand forecasting disclosed herein is particularly suited to forecasting product demand for products with limited sales data. Time-series sales data of a producs (or group of products) with limited sales data (e.g. a sparse or no time series of sales data) are dynamically time warped with sales data of products, or groups of products, having extensive sales data (e.g., an extensive time series of sales data) to determine a clustering model with an optimal number of clusters and a prototype time series for each cluster in the model. The prototype time series for the cluster in which the product (or group of products) with limited sales data lies is utilized as its product demand forecast.

An electrocardiogram waveform signal method includes obtaining a filtered waveform signal, marking the waveform signal as K signal line segments based on monotonicity, extracting line segment data of each signal line segment, and determining a line segment matching template of the waveform signal based on the line segment data of each signal line segment. The extracting of the line segment includes extracting a line segment length Xi and a line segment width Yi of each signal line segment i of the K signal line segments, performing difference extension on the line segment length Xi and the line segment width Yi based on a preset length and a preset width, respectively, to obtain a normalized signal line segment j, and extracting fourth line segment data of the normalized signal line segment j.

In an example, the present invention provides a method for processing rf backscattered signals. The method includes generating a plurality of rf signals numbered from 1 to N, where N is an integer greater than 1, from, respectively, a plurality of rf sources numbered from 1 to N. In an example, each of the rf sources is an antenna. In an example, the method includes transferring the plurality of rf signals to a predetermined region of space. The method includes receiving a stream of back scattered signals derived from each of the the rf signals numbered from 1 to N from the predetermined space, each stream of back scattered signals being one of a plurality of backscattered signals numbered 1 to N corresponding, respectively, to the plurality of rf sources numbered from 1 to N. The method includes processing each stream of the backscattered signals, using a digital signal processor, at a predetermined time to normalize the stream of backscattered signals to form a normalized signal corresponding to the stream of the backscattered signals and outputting a plurality of normalized signals numbered from 1 to N corresponding, respectively, to the plurality of back scattered signals, numbered from 1 to N.

The present disclosure relates to a user recognition method and a user recognition apparatus using a two-dimensional (2D) electromyogram (EMG) spectrogram image. The user recognition method using a 2D EMG spectrogram image may include (a) acquiring a one-dimensional EMG signal for a user, (b) converting the acquired one-dimensional EMG signal to a 2D EMG spectrogram image including a temporal feature and a frequency feature, and (c) recognizing the user based on the 2D EMG spectrogram image.

A method of processing an input data stream including at least one data peak (2), comprising: detecting at least one peak (2) in the input data stream having an apex with an amplitude above a predetermined threshold (4); and extrapolating (30) the segment of the peak which has an amplitude above the predetermined threshold (7, 8), based on a shape characteristic of the peak (2), to estimate the amplitude of the segments of the peak which have an amplitude less than said threshold (15, 16).

A fluorescent single molecule emitter simultaneously transmits its identity, location, and cellular context through its emission patterns. A deep neural network (DNN) performs multiplexed single-molecule analysis to enable retrieving such information with high accuracy. The DNN can extract three-dimensional molecule location, orientation, and wavefront distortion with precision approaching the theoretical limit of information content of the image which will allow multiplexed measurements through the emission patterns of a single molecule.

A method for imaging a downhole formation. The method includes combining the captured images to generate a partial image of the formation, wherein the partial image includes captured images separated by gaps representing portions of the formation not captured with sensors what were disposed downhole. The method includes locating dips in the formation within the partial image and interpolating the partial image using the located dips within the partial image.