Abstract: An embodiment for automatically controlling peripheral devices based on online meeting participant information. The embodiment may detect participants of an online meeting and generate a participant information table. The embodiment may generate a participant group table including one or more preliminary participant groups based on the participant information. The embodiment may generate and send audio through peripheral devices associated with at least one participant in each of the one or more preliminary participant groups to identify the participants in physically shared meeting spaces. The embodiment may update the participant group table to include confirmed participant groups based on the identified participants in the physically shared meeting spaces. The embodiment may determine a presenter for each of the confirmed participant groups and update the participant information.

Abstract: A separate quantization method of forming a combination of 4-bit and 8-bit data of a neural network is disclosed. When a training data set and a validation data set exist, a calibration manner is used to determine a threshold for activations of each of a plurality of layers of a neural network model, so as to determine how many of the activations to perform 8-bit quantization. In a process of weight quantization, the weights of each layer are allocated to 4-bit weights and 8-bit weights according to a predetermined ratio, so as to make the neural network model have a reduced size and a combination of 4-bit and 8-bit weights.

Abstract: A separate quantization method of forming a combination of 4-bit and 8-bit data of a neural network is disclosed. When a training data set and a validation data set exist, a calibration manner is used to determine a threshold for activations of each of a plurality of layers of a neural network model, so as to determine how many of the activations to perform 8-bit quantization. In a process of weight quantization, the weights of each layer are allocated to 4-bit weights and 8-bit weights according to a predetermined ratio, so as to make the neural network model have a reduced size and a combination of 4-bit and 8-bit weights.

Abstract: A height adjustable joystick device including a base, a manual operation body and a height adjustment structure is provided. The manual operation body is located on the base and includes a joystick. The joystick is located on an operating surface of the manual operation body. The height adjustment structure is arranged between the base and the manual operation body, and includes a positioning member, a moving member and a releasing member, wherein the operating surface and the base have a height difference, and the moving member and the positioning member can be engaged with each other. The releasing member is used for releasing the positioning member and the moving member is moved to a second position from a first position to adjust the height difference between the operating surface and the base when the positioning member is moved to a release position from a positioning position.