Abstract: An electronic device and method for performing class-incremental learning are provided. The method includes designating a pre-trained first model for at least one past data class as a first teacher; training a second model; designating the trained second model as a second teacher; performing dual-teacher information distillation by maximizing mutual information at intermediate layers of the first teacher and second teacher; and transferring the information to a combined student model.

Abstract: Apparatuses and methods of manufacturing same, systems, and methods for performing network parameter quantization in deep neural networks are described. In one aspect, diagonals of a second-order partial derivative matrix (a Hessian matrix) of a loss function of network parameters of a neural network are determined and then used to weight (Hessian-weighting) the network parameters as part of quantizing the network parameters. In another aspect, the neural network is trained using first and second moment estimates of gradients of the network parameters and then the second moment estimates are used to weight the network parameters as part of quantizing the network parameters. In yet another aspect, network parameter quantization is performed by using an entropy-constrained scalar quantization (ECSQ) iterative algorithm. In yet another aspect, network parameter quantization is performed by quantizing the network parameters of all layers of a deep neural network together at once.

Abstract: A multilayer capacitor includes a body including dielectric layers and internal electrodes, and external electrodes, wherein the body has first and second surfaces opposing each other in a first direction, third and fourth surfaces opposing each other in a second direction, and fifth and sixth surfaces opposing each other in a third direction perpendicular to the first and second directions. A length of a portion of the plurality of internal electrodes in the third direction in an intermediate region of the body in the first direction is greater than a length of the first surface or the second surface of the body in the third direction. The plurality of internal electrodes have a bottleneck structure between the intermediate region and at least one of the first and second surfaces, and wherein the bottleneck structure has a shape recessed into an inner portion of the body.

Abstract: A method for training a generator, by a generator training system including a processor and memory, includes: extracting training statistical characteristics from a batch normalization layer of a pre-trained model, the training statistical characteristics including a training mean ? and a training variance ?2; initializing a generator configured with generator parameters; generating a batch of synthetic data using the generator; supplying the batch of synthetic data to the pre-trained model; measuring statistical characteristics of activations at the batch normalization layer and at the output of the pre-trained model in response to the batch of synthetic data, the statistical characteristics including a measured mean {circumflex over (?)}? and a measured variance {circumflex over (?)}?2; computing a training loss in accordance with a loss function L? based on ?, ?2, {circumflex over (?)}?, and {circumflex over (?)}?2; and iteratively updating the generator parameters in accordance with the training loss unti

Abstract: A method is provided. The method includes selecting a neural network model, wherein the neural network model includes a plurality of layers, and wherein each of the plurality of layers includes weights and activations; modifying the neural network model by inserting a plurality of quantization layers within the neural network model; associating a cost function with the modified neural network model, wherein the cost function includes a first coefficient corresponding to a first regularization term, and wherein an initial value of the first coefficient is pre-defined; and training the modified neural network model to generate quantized weights for a layer by increasing the first coefficient until all weights are quantized and the first coefficient satisfies a pre-defined threshold, further including optimizing a weight scaling factor for the quantized weights and an activation scaling factor for quantized activations, and wherein the quantized weights are quantized using the optimized weight scaling factor.

Abstract: Methods and apparatuses are provided for continual few-shot learning. A model for a base task is generated with base classification weights for base classes of the base task. A series of novel tasks is sequentially received. Upon receiving each novel task in the series of novel tasks, the model is updated with novel classification weights for novel classes of the respective novel task. The novel classification weights are generated by a weight generator based on one or more of the base classification weights and, when one or more other novel tasks in the series are previously received, one or more other novel classification weights for novel classes of the one or more other novel tasks. Additionally, for each novel task, a first set of samples of the respective novel task are classified into the novel classes using the updated model.

Abstract: The present disclosure provides a server through which a user can increase the usability of a content in a display apparatus by providing the content output from the display apparatus based on upload image data, a display apparatus, and a method of controlling the display apparatus. The display apparatus includes a display; a communication interface configured to communicate with the server; and a controller configured to receive a content generated in the server based on image data uploaded by a user to the server, categories selected by the user, and setting information, and to control the content to be output on the display with the setting defined in the setting information.

Abstract: A convolutional neural network (CNN) system for generating a classification for an input image is presented. The CNN system comprises circuitry running on clock cycles and configured to compute a product of two received values, and at least one non-transitory computer-readable medium that stores instructions for the circuitry to derive a feature map based on at least the input image; puncture at least one selection among the feature map and a kernel by setting the value of an element at an index of the at least one selection to zero and cyclic shifting a puncture pattern to achieve a 1/d reduction in number of clock cycles, where d is an integer and puncture interval value>1. The feature map is convolved with the kernel to generate an output, and a classification of the input image is generated based on the output.

Abstract: Disclosed herein is convolutional neural network (CNN) system for generating a classification for an input image. According to an embodiment, the CNN system comprises a sequence of neural network layers configured to: derive a feature map based on at least the input image; puncture at least one selection among the feature map and a kernel by setting the value of one or more elements of a row of the at least one selection to zero according to a pattern and cyclic shifting the pattern by a predetermined interval per row to set the value of one or more elements of the rest of the rows of the at least one selection according to the cyclic shifted pattern; convolve the feature map with the kernel to generate a first convolved output; and generate the classification for the input image based on at least the first convolved output.

Abstract: A system and method for operating a neural network. In some embodiments, the neural network includes a variational autoencoder, and the training of the neural network includes training the variational autoencoder with a plurality of samples of a first random variable; and a plurality of samples of a second random variable, the plurality of samples of the first random variable and the plurality of samples of the second random variable being unpaired, the training of the neural network including updating weights in the neural network based on a first loss function, the first loss function being based on a measure of deviation from consistency between: a conditional generation path from the first random variable to the second random variable, and a conditional generation path from the second random variable to the first random variable.

Abstract: A method and system are provided. The method includes receiving, at a generator, a random input, producing, at the generator, a synthetic output of the received random input, receiving, at a teacher network, the synthetic output, receiving, at a student network, the synthetic output, minimizing a maximum of a distance between an output of the teacher network and an output of the student network, and constraining the generator.

Abstract: A nonvolatile memory device and method for fabricating the same are provided. The nonvolatile memory device comprising: a substrate; a mold structure including a first insulating pattern and a plurality of gate electrodes alternately stacked in a first direction on the substrate; and a word line cut region which extends in a second direction different from the first direction and cuts the mold structure, wherein the word line cut region includes a common source line, and the common source line includes a second insulating pattern extending in the second direction, and a conductive pattern extending in the second direction and being in contact with the second insulating pattern and a cross-section in the second direction.

Abstract: A system and method for characterizing an interference demodulation reference signal (DMRS) in a piece of user equipment (UE), e.g., a mobile device. The UE determines whether the serving signal is transmitted in a DMRS-based transmission mode; if it is, the UE cancels the serving DMRS from the received signal; otherwise the UE cancels the serving data signal from the received signal. The remaining signal is then analyzed for the amount of power it has in each of four interference DMRS candidates, and hypothesis testing is performed to determine whether interference DMRS is present in the signal, and, if so, to determine the rank of the interference DMRS, and the port and scrambling identity of each of the interference DMRS layers.

Abstract: According to the present disclosure, when drip coffee is extracted, it may learn drip coffee recipe information of a barista to be imitated and extract the drip coffee based on the learned reference acidity and reference concentration. The extracted drip coffee may be compared with the learned reference acidity and the reference concentration and evaluated, and when the acidity and concentration of the drip coffee are matched with the reference acidity and the reference concentration, the drip coffee having the same acidity and concentration as the drip coffee may be extracted. Alternatively, when the acidity and concentration of the drip coffee are not matched with the reference acidity and the reference concentration, drip coffee having the same or/and similar acidity and concentration as or/and to the reference acidity and the reference concentration may be extracted by changing extraction conditions of the drip coffee through reinforcement learning.