Abstract: A processor-implemented neural network method includes: determining a reference sample among sequential input samples to be processed by a neural network, the neural network comprising an input layer, one or more hidden layers, and an output layer; performing an inference process of obtaining an output activation of the output layer based on operations in the hidden layers corresponding to the reference sample input to the input layer; determining layer contraction parameters for determining an affine transformation relationship between the input layer and the output layer, for approximation of the inference process; and performing inference on one or more other sequential input samples among the sequential input samples using affine transformation based on the layer contraction parameters determined with respect to the reference sample.

Abstract: A method for operating an artificial neuron and an apparatus for performing the method are provided. The artificial neuron may calculate a change amount of an activation based on an input signal received via an input synapse, determine whether an event occurs in response to the calculated change amount of the activation, and transmit, to an output synapse, an output signal that corresponds to the event in response to an occurrence of the event.

Abstract: A method for operating an artificial neuron and an apparatus for performing the method are provided. The artificial neuron may calculate a change amount of an activation based on an input signal received via an input synapse, determine whether an event occurs in response to the calculated change amount of the activation, and transmit, to an output synapse, an output signal that corresponds to the event in response to an occurrence of the event.

Abstract: Disclosed is a neural network-based inference method and apparatus. The neural network-based inference method includes compressing a matrix comprising processing elements corresponding to an operation of a neural network, balancing workloads related to the operation by reordering the compressed matrix based on the workloads, and performing inference based on the reordered matrix.

Abstract: A method for operating an artificial neuron and an apparatus for performing the method are provided. The artificial neuron may calculate a change amount of an activation based on an input signal received via an input synapse, determine whether an event occurs in response to the calculated change amount of the activation, and transmit, to an output synapse, an output signal that corresponds to the event in response to an occurrence of the event.

Abstract: A method for operating an artificial neuron and an apparatus for performing the method are provided. The artificial neuron may calculate a change amount of an activation based on an input signal received via an input synapse, determine whether an event occurs in response to the calculated change amount of the activation, and transmit, to an output synapse, an output signal that corresponds to the event in response to an occurrence of the event.

Abstract: A method for operating an artificial neuron and an apparatus for performing the method are provided. The artificial neuron may calculate a change amount of an activation based on an input signal received via an input synapse, determine whether an event occurs in response to the calculated change amount of the activation, and transmit, to an output synapse, an output signal that corresponds to the event in response to an occurrence of the event.

Abstract: An operation method of a neural network, a training method, and a signal processing apparatus are provided. The operation method includes receiving an output signal from a first neural network, and converting a first feature included in the output signal to a second feature configured to be input to a second neural network, based on a conversion rule controlling conversion between a feature to be output from the first neural network and a feature to be input to the second neural network. The operation method further includes generating an input signal to be input to the second neural network, based on the second feature, and transmitting the input signal to the second neural network.

Abstract: An operation method of a neural network, a training method, and a signal processing apparatus are provided. The operation method includes receiving an output signal from a first neural network, and converting a first feature included in the output signal to a second feature configured to be input to a second neural network, based on a conversion rule controlling conversion between a feature to be output from the first neural network and a feature to be input to the second neural network. The operation method further includes generating an input signal to be input to the second neural network, based on the second feature, and transmitting the input signal to the second neural network.

Abstract: A processor-implemented neural network method includes: determining a reference sample among sequential input samples to be processed by a neural network, the neural network comprising an input layer, one or more hidden layers, and an output layer; performing an inference process of obtaining an output activation of the output layer based on operations in the hidden layers corresponding to the reference sample input to the input layer; determining layer contraction parameters for determining an affine transformation relationship between the input layer and the output layer, for approximation of the inference process; and performing inference on one or more other sequential input samples among the sequential input samples using affine transformation based on the layer contraction parameters determined with respect to the reference sample.

Abstract: An event-based sensor is provided and may include a sensor configured to generate an event signal that includes identification information that relates to an active pixel that detects an event from among a plurality of sensing pixels, a determiner configured to determine whether the event signal is to be filtered based on a predetermined condition, and an outputter configured to output the event signal based on a result of the determination.

Abstract: A method for operating an artificial neuron and an apparatus for performing the method are provided. The artificial neuron may calculate a change amount of an activation based on an input signal received via an input synapse, determine whether an event occurs in response to the calculated change amount of the activation, and transmit, to an output synapse, an output signal that corresponds to the event in response to an occurrence of the event.

Abstract: An operation method of a neural network, a training method, and a signal processing apparatus are provided. The operation method includes receiving an output signal from a first neural network, and converting a first feature included in the output signal to a second feature configured to be input to a second neural network, based on a conversion rule controlling conversion between a feature to be output from the first neural network and a feature to be input to the second neural network. The operation method further includes generating an input signal to be input to the second neural network, based on the second feature, and transmitting the input signal to the second neural network.

Abstract: An event-based sensor is provided and may include a sensor configured to generate an event signal that includes identification information that relates to an active pixel that detects an event from among a plurality of sensing pixels, a determiner configured to determine whether the event signal is to be filtered based on a predetermined condition, and an outputter configured to output the event signal based on a result of the determination.

Abstract: A motion recognizing apparatus and method are provided. According to an aspect, a motion recognizing apparatus may include: an optical sensor configured to sense at least a portion of a subject where a motion occurs and to output one or more events in response thereto; a motion tracing unit configured to trace a motion locus of the portion where the motion occurs based on the one or more outputted events; and a motion pattern determining unit configured to determine a motion pattern of the portion where the motion occurs based on the traced motion locus.

Abstract: A proximity sensor and proximity sensing method using a change in light quantity of a reflected light are disclosed. The proximity sensor may include a quantity change detection unit which detects a change in a quantity of reflected light which is output light which has been reflected by an object, where an intensity of the output light changes, and a proximity determination unit which determines a proximity of the object to the quantity change detection unit based on a change in the intensity of the output light and the detected change in the quantity of the reflected light.

Abstract: A proximity sensor and proximity sensing method using a change in light quantity of a reflected light are disclosed. The proximity sensor may include a quantity change detection unit which detects a change in a quantity of reflected light which is output light which has been reflected by an object, where an intensity of the output light changes, and a proximity determination unit which determines a proximity of the object to the quantity change detection unit based on a change in the intensity of the output light and the detected change in the quantity of the reflected light.

Abstract: A photoarray includes a one-dimensional or two-dimensional array of cells, each having a photosensor generating a sensor signal dependent on a light intensity at the cell, a first capacitor charged by a time-derivative of a current, at least one threshold detector detecting if a voltage over the first capacitor exceeds a threshold value and generating an output signal if it does, and a discharge device for discharging the first capacitor after occurrence of the output signal. Such a cell generates an event only when the incoming light intensity changes, which reduces the amount of data to be processed from the photoarray.

Abstract: A photoarray includes a one-dimensional or two-dimensional array of cells, each having a photosensor generating a sensor signal dependent on a light intensity at the cell, a first capacitor charged by a time-derivative of a current, at least one threshold detector detecting if a voltage over the first capacitor exceeds a threshold value and generating an output signal if it does, and a discharge device for discharging the first capacitor after occurrence of the output signal. Such a cell generates an event only when the incoming light intensity changes, which reduces the amount of data to be processed from the photoarray.

Abstract: An imaging array having overflow protection and electronic shuttering features is realized without an increase in pixel complexity. Overflow protection is provided by pulsing each row of the imager with a small overflow pulse during the sense amplifier reset phase. An electronic shutter is realized using a modified version of the pixel readout timing. The shutter provides sub-frame exposure by restricting the number of line-times a pixel is allowed to integrate. For a full-frame exposure, each pixel is read out once per frame; during readout of the other rows of the array, the pixel integrates. For subframe exposure, the pixel is continually reset, using a shutter pulse applied to the row lines during sense amplifier reset, until a certain number of rows (line-times) before it is to be read out. The pixel then is allowed to integrate until it is read out normally.