Samsung Electronics Co., Ltd.

Youngmin Oh

Sangho Lee

Gajin Song

Kibeom Kim

Yeona Hong

Jun Haeng Lee

Yongwoo Lee

Changwook Jeong

Hyojin Choi

In Huh

Minkyoung Cho

Younsik Park

Jaecheol Lee

Hyun Sun Park

Seungju Kim

Won-jo Lee

A neural network pruning method includes acquiring a first task accuracy of an inference task processed by a pretrained neural network, pruning, based on a channel unit, the neural network by adjusting weights between nodes of channels based on a preset learning weight and based on a channel-by-channel pruning parameter corresponding to a channel of each of a plurality of layers of the pretrained neural network, updating the learning weight based on the first task accuracy and a task accuracy of the pruned neural network, updating the channel-by-channel pruning parameter based on the updated learning weight and the task accuracy of the pruned neural network, and repruning, based on the channel unit, the pruned neural network based on the updated learning weight and based on the updated channel-by-channel pruning parameter.

Method and apparatus with neural network pruning

A system verification method includes generating a first verification vector as a result of a first action of an agent, the first verification vector referring to an observation corresponding to at least one state already covered, from among states of elements of a target system, identifying a first coverage corresponding to at least one state covered by the first verification vector, from among the states of the elements, updating the observation by reflecting the first coverage in the observation, and generating a second verification vector through a second action of the agent, the second verification vector referring to the updated observation.

Method and apparatus with system verification based on reinforcement learning

Provided are an apparatus and method of compressing an artificial neural network. According to the method and the apparatus, an optimal compression rate and an optimal operation accuracy are determined by compressing an artificial neural network, determining a task accuracy of a compressed artificial neural network, and automatically calculating a compression rate and a compression ratio based on the determined task accuracy. The method includes obtaining an initial value of a task accuracy for a task processed by the artificial neural network, compressing the artificial neural network by adjusting weights of connections among layers of the artificial neural network included in information regarding the connections, determining a compression rate for the compressed artificial neural network based on the initial value of the task accuracy and a task accuracy of the compressed artificial neural network, and re-compressing the compressed artificial neural network according to the compression rate.

Apparatus and method of compressing neural network

Provided is an electronic device. The electronic device may include: a user interface; a processor operatively connected to the user interface; and a memory operatively connected to the processor, wherein the memory may store instructions that, when executed, cause the processor to control the electronic device to: receive an input via the user interface; determine a task including plural actions based on the input; execute a first action among the plural actions of the determined task; obtain context information related to the task while executing the first action; determine at least one first threshold associated with the first action based at least in part on the obtained context information; and determine the result of the first action based on the execution of the first action being completed based on the at least one first threshold.

Electronic device and method for determining task including plural actions

An electronic device which schedules a plurality of tasks, and an operating method thereof. The electronic device includes a processor and a memory operatively connected to the processor, and when being executed, the memory stores instructions that cause the processor to: detect occurrence of an interrupt requesting performance of a second task while performing a first task; obtain reference values according to a time of the first task, and reference values according to a time of the second task; schedule the first task and the second task based on a reference value of the first task and a reference value of the second task which correspond to a time at which the interrupt occurs; and process the first task and the second task based on a result of the scheduling. Other embodiments are possible.

Electronic device for scheduling a plurality of tasks and operating method thereof

Growing community of inventors

Suwon-si, South Korea

Youngmin Oh