Abstract: Embodiments of the present disclosure relate to fine-tuning a neural network model using a weight-decomposed low-rank adaptation (DoRA). DoRA reduces the number of parameters that are fine-tuned, thereby reducing memory and the time needed to fine-tune the parameters. the Pre-trained weights are decomposed into two components, magnitude and direction, which are separately fine-tuned. The magnitude components are fine-tuned while the direction components remain unchanged (frozen). Then low-rank adaptation (LoRA) is used to fine-tune the direction components, efficiently minimizing the number of trainable parameters. Compared with using LoRA to fine-tune the weights directly, using DoRA exhibits a closer resemblance to full fine-tuning's learning behavior and improves upon LoRA in commonsense reasoning and visual instruction tuning tasks. By employing DoRA, both the learning capacity and training stability of LoRA is enhanced.

Abstract: According to an example, an electronic device comprises a device housing having an outer surface, a pivot element disposed within an indentation having a set of detents, and a support structure rotatable about the pivot element. The support structure may comprise a projecting element to engage with each detent of the set of detents. Upon a force being applied at a first portion of the support structure, the support structure is to rotate such that a second portion of the support structure is to protrude from the outer surface of the device housing.

Abstract: Semantic segmentation generally refers to a machine learning process that associates a label or category with every pixel in an image. This can be used to recognize a collection of pixels that form distinct categories of objects, which may have applications in autonomous driving for example where the vehicle needs to identify other vehicles, pedestrians, traffic signs, pavement, and other road features from captured images of a surrounding environment. While using pixel-level annotations may be ideal for fully-supervised training of the semantic segmentation model, collecting such annotations is time-consuming and expensive, and therefore limits the scalability and practicality of fully-supervised training methods. The present disclosure enables weakly supervised training of a semantic segmentation model aided by learned prompts embedded with semantic knowledge.

Abstract: Various solutions for determining channel state information (CSI) measurement window and CSI reporting window with respect to user equipment and network apparatus in mobile communications are described. An apparatus may transmit a user equipment (UE) capability parameter related to a maximum supportable time span of a CSI reporting window. The apparatus may receive a time span of a CSI reporting window configured according to the UE capability parameter.

Abstract: A method of downlink channel state information (DL CSI) computation and reporting is proposed to support high velocity in new radio (NR) systems. In a first novel aspect, two CSI reference slots are defined. CSI reference slot for CSI measurement is defined for determining which CSI-RS/SSB occasion(s) is used for computing CSI. CSI reference slot for CSI computation is defined for determining the slot where UE assumes the CSI computation should be based upon the channel at that time onwards. In a second novel aspect, UE can be configured with a CSI computation period consisting of one or N slots and can be divided into multiple sub-periods without overlap. UE can be configured to compute and report wideband CSI and subband CSI for the whole CSI computation period and/or for each sub-period.

Abstract: A method for handling a call using Sessions Initiation Protocol (SIP) includes sending an INVITE or a first media type by a network side to user equipment, sending an UPDATE of a second media type by the user equipment to the network side, receiving an SIP global failure response for the UPDATE by the user equipment from the network side, sending an SIP success response for the INVITE of the first media type by the user equipment to the network side, receiving an ACK of the first media type by the user equipment from the network side, and establishing a call session of the first media type between the user equipment and the network side.

Abstract: A voice call management method is provided. The voice call management method may be applied to an apparatus. The voice call management method may include the following steps. The apparatus may determine the current scenario associated with the operation environment of the apparatus. Then, the apparatus may determine to perform a voice call through a modem (MD) voice engine or through an application processor (AP) voice engine according to the current scenario.

Abstract: A receive (Rx) UE performs blind detection to decode a first-stage Sidelink Control Information (SCI) in a received signal. The first-stage SCI contains control information for the Rx UE to locate time-and-frequency resources used by a transmit (Tx) UE. The time-and-frequency resources are used to transmit the first-stage SCI and a second-stage SCI via a Physical Sidelink Control Channel (PSCCH) and to transmit data via a Physical Sidelink Shared Channel (PSSCH) associated with the PSCCH. The Rx UE locates the second-stage SCI in the time-and-frequency resources based on the first-stage SCI. The second-stage SCI contains additional control information from the Tx UE to the Rx UE for the sidelink V2X communication. The Rx UE decodes the second-stage SCI using at least in part an identifier which identifies the Rx UE as a destination of the received signal.

Abstract: A method of downlink channel state information (DL CSI) computation and reporting is proposed to support high velocity in new radio (NR) systems. In a first novel aspect, two CSI reference slots are defined. CSI reference slot for CSI measurement is defined for determining which CSI-RS/SSB occasion(s) is used for computing CSI. CSI reference slot for CSI computation is defined for determining the slot where UE assumes the CSI computation should be based upon the channel at that time onwards. In a second novel aspect, UE can be configured with a CSI computation period consisting of one or N slots and can be divided into multiple sub-periods without overlap. UE can be configured to compute and report wideband CSI and subband CSI for the whole CSI computation period and/or for each sub-period.

Abstract: A method of downlink channel state information (DL CSI) computation and reporting is proposed to support high velocity in new radio (NR) systems. In a first novel aspect, two CSI reference slots are defined. CSI reference slot for CSI measurement is defined for determining which CSI-RS/SSB occasion(s) is used for computing CSI. CSI reference slot for CSI computation is defined for determining the slot where UE assumes the CSI computation should be based upon the channel at that time onwards. In a second novel aspect, UE can be configured with a CSI computation period consisting of one or N slots and can be divided into multiple sub-periods without overlap. UE can be configured to compute and report wideband CSI and subband CSI for the whole CSI computation period and/or for each sub-period.

Abstract: A method of downlink channel state information (DL CSI) computation and reporting is proposed to support high velocity in new radio (NR) systems. In a first novel aspect, two CSI reference slots are defined. CSI reference slot for CSI measurement is defined for determining which CSI-RS/SSB occasion(s) is used for computing CSI. CSI reference slot for CSI computation is defined for determining the slot where UE assumes the CSI computation should be based upon the channel at that time onwards. In a second novel aspect, UE can be configured with a CSI computation period consisting of one or N slots and can be divided into multiple sub-periods without overlap. UE can be configured to compute and report wideband CSI and subband CSI for the whole CSI computation period and/or for each sub-period.

Abstract: An image generation method and an image generation device are disclosed. The method includes: reading a first image from a storage circuit; disposing an image mask on the first image, wherein the image mask covers a part of the image areas in the first image; moving, randomly, the image mask to change a covering range of the image mask in the first image; obtaining a second image from the first image according to the moved image mask; and storing the second image into the storage circuit.

Abstract: A method of two-stage sidelink control information (SCI) can include receiving a transport block, and transmitting from a transmission user equipment (Tx UE) to a reception user equipment (Rx UE) the transport block with two-stage SCI including a 1st-stage SCI and a 2nd-stage SCI. The 1st-stage SCI can be transmitted over a physical sidelink control channel (PSCCH). The 2nd-stage SCI and the transport block can share resources of a physical sidelink shared channel (PSSCH). The 2nd-stage SCI and the transport block can share a demodulation reference signal (DMRS).

Abstract: A method of two-stage sidelink control information (SCI) for sidelink transmission is described. In one novel aspect, the SCI is transmitted from a transmission UE (Tx UE) to a reception UE (Rx UE) in two stages. In one embodiment, the 1st-stage SCI carries information related to channel sensing, and is transmitted over a physical sidelink control channel (PSCCH). The 2nd-stage SCI is transmitted over a physical sidelink shared channel (PSSCH) associated with the PSCCH. In one embodiment, the first-stage SCI includes priority information or information of time-frequency resources for the SL transmission. In another embodiment, the first-stage SCI indicates time-frequency location of the second-stage SCI.

Abstract: A noise measuring device is provided. The noise measuring device includes a soundproof box, a sound receiving device, a holding device, and a driving device. The sound receiving device is disposed in the soundproof box. The holding device is disposed in the soundproof box and configured to hold a testing object. The driving device is connected with the soundproof box and configure to drive the soundproof box to rotate.

Abstract: For sidelink vehicle-to-everything (V2X) transmission, a first user equipment (UE) obtains time-and-frequency resources that are bounded by two subchannel boundaries and two slot boundaries. The first UE transmits encoded control information to a second UE in a Physical Sidelink Control Channel (PSCCH) that occupies a first portion of the time-and-frequency resources starting from a lowest resource block (RB) in a lowest subchannel and from an earliest symbol available for sidelink control and data transmission. The first UE further transmits encoded data to the second UE in a Physical Sidelink Shared Channel (PSSCH) that occupies a second portion of the time-frequency resources. The second portion includes (a) a first set of RBs in the same symbols as the PSCCH and not used by the PSCCH, and (b) a second set of RBs in symbols not used by the PSCCH.

Abstract: User equipment (UE) for sounding reference signal (SRS) triggering is provided. The UE may include a radio frequency (RF) signal processing device and a processor. The RF signal processing device receives first slot offset information and second slot offset information from a network node. The first slot offset information is configured through a radio resource control (RRC) information element (IE) to indicate a first time difference between the down link control information (DCI) and a resource reference slot. The second slot offset information indicates the second time difference between the reference slot and an SRS resource. The processor is coupled to the RF signal processing device. When the RF signal processing device receives the DCI, the processor triggers an SRS transmission for the SRS resource according to the first slot offset information and the second slot offset information.

Abstract: A user equipment (UE) in wireless communication receives a configuration of search space (SS) sets that include a first SS set in a first control resource set (CORESET) and a second SS set in a second CORESET. The first CORESET and the second CORESET have different transmission configuration indication (TCI) states. The UE repeatedly receives downlink control information (DCI) via a first physical downlink control channel (PDCCH) and a second PDCCH. The first PDCCH is allocated with a first set of time-and-frequency resources configured for first SS set and the second PDCCH is allocated with a second set of time-and-frequency resources configured for the second SS set. The UE then parses the DCI to obtain control information for the wireless communication. Alternatively, the configuration of SS sets received by the UE is in one CORESET that has at least a TCI state and a second TCI state activated.

Abstract: A control method is provided, which is applied to an electronic apparatus including an input device. The control method includes the following steps: detecting whether a setup program exists or not, wherein the setup program includes a plurality of adjusting information; detecting whether the setup program is executed or not, when the setup program exists; receiving a selecting instruction, when the setup program is executed; generating a control signal according to the selecting instruction and the adjusting information; receiving an input signal generated by the input device; and delaying and processing the input signal according to the control signal.

Abstract: According to an example, an electronic device comprises a device housing having an outer surface, a pivot element disposed within an indentation having a set of detents, and a support structure rotatable about the pivot element. The support structure may comprise a projecting element to engage with each detent of the set of detents. Upon a force being applied at a first portion of the support structure, the support structure is to rotate such that a second portion of the support structure is to protrude from the outer surface of the device housing.