Abstract: An arithmetic operation device causes a convolution arithmetic unit to perform a convolution arithmetic operation between a filter and target data corresponding to a size of the filter in each of a plurality of convolution layers constituting a neural network. The arithmetic operation device includes: a bit reduction unit that reduces a bit string corresponding to a first bit number from a least significant bit of the target data and reduces a bit string corresponding to a second bit number from a least significant bit of a weight that is an element of the filter for each convolution layer; and a bit addition unit that adds a bit string corresponding to a third bit number obtained by adding the first bit number and the second bit number to a least significant bit of a convolution arithmetic operation result output from the convolution arithmetic unit by inputting the target data and the weight after being reduced by the bit reduction unit to the convolution arithmetic unit.

Abstract: The state monitoring system includes: a 3D camera configured to acquire an image in a work area; and information processing apparatus connected to the 3D camera and including a processing unit and a storage unit. The processing unit calculates a camera setting parameter that determines an imaging condition of the 3D camera with respect to a monitoring target as a target to be monitored in the work area and stores the camera setting parameter in the storage unit, determines the imaging condition of the 3D camera by applying the camera setting parameter corresponding to the monitoring target with reference to the camera setting parameter stored in the storage unit, acquires an image of the monitoring target from the 3D camera configured to image the monitoring target in the determined imaging condition, and determines a state of the monitoring target based on the acquired image of the monitoring target.

Abstract: The time-series signal of the sensor is transformed to the spectral intensity by fast Fourier transform (FFT) or the like, and the one-dimensional data of the spectral intensity is generated. A pseudo image is generated, for example, by repeatedly arranging the one-dimensional data in the vertical direction, or by arranging the one-dimensional data for a plurality of sensors in the vertical direction. The state of the facility is identified by analyzing the pseudo image with an image analysis unit such as a convolutional neural network.

Abstract: Provided is a new technique that enables authentication of sources as well as storage and distribution conditions of food.

Abstract: An information processing device executes a DNN computation by a neural network including a plurality of layers. The information processing device executes a computation process corresponding to a given layer in the neural network, on a first area and on a second area different from the first area, the first and second areas being included in a feature map inputted to the neural network. The information processing device synthesizes a result of the computation process on the first area and a result of the computation process on the second area, and outputs the synthesized computation process results, as a result of the computation process on the feature map.

Abstract: The present invention aims to reduce power consumption in the operation based upon a recognition device (1000) including, in a neural network of multiple layers that output type of an object and existing coordinates based on external environment information, a selector (103) that selects input data to convolution operation units (107-1 to L) from external environment information; convolution operation units (107-1 to L) configured by a plurality of layers connected in cascade; and a parameter storage unit (109) that stores a weight parameter of each layer, a cumulative addition count of each layer, and an omitting bit number of each layer. The recognition device includes operation stop signal generation units (116-1 to L) that transmit one or more stop signals for stopping some or all of the computing units of the convolution operation units (107-1 to L) to the convolution operation units (107-1 to L) for each layer.

Abstract: Provided is a training recognition device that implements training of a DNN for article recognition that does not require manual annotation for an image for training and can reduce power consumption, time, and hardware amount required for training.

Abstract: An information processing apparatus having an input device for receiving data, an operation unit for constituting a convolutional neural network for processing data, a storage area for storing data to be used by the operation unit and an output device for outputting a result of the processing. The convolutional neural network is provided with a first intermediate layer for performing a first processing including a first inner product operation and a second intermediate layer for performing a second processing including a second inner product operation, and is configured so that the bit width of first filter data for the first inner product operation and the bit width of second filter data for the second inner product operation are different from each other.

Abstract: A computing system includes a server, and an on-vehicle device mounted on a vehicle. The server includes: a server storage part containing a learned model; a simplification part which generates a contraction information for determination of calculation precision by using the learned model and an object of inference; and a server communication part for transmitting the contraction information to the on-vehicle device. The on-vehicle device includes: a signal input part to which an output signal from a sensor mounted on the vehicle is inputted; an on-vehicle communication part for receiving the contraction information; an inference part for making an inference on the output signal and a reconfigurable logic circuit; and a reconfiguration part for configuring the inference part in the logic circuit based on the contraction information.

Abstract: An arithmetic operation device causes a convolution arithmetic unit to perform a convolution arithmetic operation between a filter and target data corresponding to a size of the filter in each of a plurality of convolution layers constituting a neural network. The arithmetic operation device includes: a bit reduction unit that reduces a bit string corresponding to a first bit number from a least significant bit of the target data and reduces a bit string corresponding to a second bit number from a least significant bit of a weight that is an element of the filter for each convolution layer; and a bit addition unit that adds a bit string corresponding to a third bit number obtained by adding the first bit number and the second bit number to a least significant bit of a convolution arithmetic operation result output from the convolution arithmetic unit by inputting the target data and the weight after being reduced by the bit reduction unit to the convolution arithmetic unit.

Abstract: An information processing device includes a parallel deep neural network configured to input a captured image of an article to deep neural network models respectively corresponding to a plurality of articles and perform inferences about the plurality of articles in parallel using the deep neural network models, a new article determination unit configured to determine whether an article included in the image is an unlearned article based on learned model information about the articles and the image, and a new article learning unit configured to learn a deep neural network model corresponding to the article determined to be unlearned based on the image and initial model configuration information about the deep neural network model when the article included in the image is determined to be an unlearned article. The new article learning unit adds the learned deep neural network model to the deep neural network models.

Abstract: To improve learning accuracy, while avoiding transferring of a dataset from an edge terminal to a cloud server. A management device accessible to a target object to be managed has a processor executing a program, a storage device storing the program, and a communication interface communicable with the target object. The processor executes a reception process for receiving first environmental information representing a first environment of the target object, a first generation process for generating relevant information representing relevancy between the first environmental information and second environmental information representing a second environment of the target object, a second generation process for generating a first learned model for inference by the target object in the first environment based on the relevant information and a second learned model for inference by the target object in the second environment, and a transmission process for transmitting the first learned model to the target object.

Abstract: Learning data of a usage environment can be efficiently collected. A recognition apparatus includes: a first neural network configured to receive input of data; a second neural network configured to receive input of the data, the second neural network having a different structure from a structure of the first neural network; a comparison unit configured to compare a first output result of the first neural network and a second output result of the second neural network; and a communication unit configured to wirelessly transmit the data to a host system configured to learn the data when a comparison result between the first output result and the second output result is different by a predetermined standard or more.

Abstract: An information processing device includes a DNN operation unit that executes a DNN operation by a neural network including plural layers and a weight storage unit that stores a weight used in the DNN operation, and the DNN operation unit specifies data having a value larger than a predetermined threshold as operation target data among data input to a predetermined layer of the neural network, acquires a weight corresponding to the operation target data from the weight storage unit, and executes an operation of the predetermined layer based on the operation target data and the weight acquired from the weight storage unit.

Abstract: Processing time of a neural network is shortened, and the number of operations of the neural network is reduced such that a plurality of calculators can be effectively used. A neural network reduction device (100) that reduces the number of operations of a neural network by an operation device (140) including a plurality of calculators by reducing the neural network, the neural network reduction device including: a calculator allocation unit (102) that sets the number of calculators allocated to calculation processing of the neural network; a number-of-operations setting unit (103) that sets the number of operations of a reduced neural network based on the number of allocated calculators; and a neural network reduction unit (104) that reduces the neural network such that the number of operations of the neural network by the operation device (140) is equal to the number of operations set by the number-of-operations setting unit (103).

Abstract: A computing system includes a server, and an on-vehicle device mounted on a vehicle. The server includes: a server storage part containing a learned model; a simplification part which generates a contraction information for determination of calculation precision by using the learned model and an object of inference; and a server communication part for transmitting the contraction information to the on-vehicle device. The on-vehicle device includes: a signal input part to which an output signal from a sensor mounted on the vehicle is inputted; an on-vehicle communication part for receiving the contraction information; an inference part for making an inference on the output signal and a reconfigurable logic circuit; and a reconfiguration part for configuring the inference part in the logic circuit based on the contraction information.

Abstract: To improve learning accuracy, while avoiding transferring of a dataset from an edge terminal to a cloud server. A management device accessible to a target object to be managed has a processor executing a program, a storage device storing the program, and a communication interface communicable with the target object. The processor executes a reception process for receiving first environmental information representing a first environment of the target object, a first generation process for generating relevant information representing relevancy between the first environmental information and second environmental information representing a second environment of the target object, a second generation process for generating a first learned model for inference by the target object in the first environment based on the relevant information and a second learned model for inference by the target object in the second environment, and a transmission process for transmitting the first learned model to the target object.

Abstract: To achieve, in a load drive device including an H-bridge circuit, miniaturization of a capacitor or/and the constituent elements of the H-bridge circuit (e.g., reduction in volume) with circuit elements, for example, switching elements included in the H-bridge circuit, being inhibited from breaking down or destroying even in a case where a load is overloaded. The invention is disclosed in which, as a solution to the achievement, first and second modes are provided as a switching mode for the switching elements 11, 12, 13, and 14 and switching is appropriately performed between the first and second modes.

Abstract: In a processing method using a convolutional neural network, the neural network includes a convolution calculation unit that performs a convolution calculation by using a matrix vector product and a pooling calculation unit that performs a maximum value sampling calculation. A threshold value is set related to the matrix data for the convolution calculation, the matrix data is divided into a first and second halves based on the threshold value. The convolution calculation unit divides a first half convolution calculation by using the first half of the matrix data and a second half convolution calculation by using the second half of the matrix data into two and executes the calculations. The pooling calculation unit selects vector data to which the matrix vector product convolution calculation is to be performed in the second half convolution calculation, along with the maximum value sampling calculation.

Abstract: An information processing apparatus having an input device for receiving data, an operation unit for constituting a convolutional neural network for processing data, a storage area for storing data to be used by the operation unit and an output device for outputting a result of the processing. The convolutional neural network is provided with a first intermediate layer for performing a first processing including a first inner product operation and a second intermediate layer for performing a second processing including a second inner product operation, and is configured so that the bit width of first filter data for the first inner product operation and the bit width of second filter data for the second inner product operation are different from each other.