Abstract: Techniques and systems are provided for performing online adaptation of machine learning model(s). For example, a process may include obtaining features extracted from a image by a machine learning model during inference and determining, by the machine learning model based on the features during inference, a plurality of keypoint estimates in the image and/or a bounding region estimate associated with an object in the image. The process may further include generating pseudo-label(s) based on the plurality of keypoint estimates and/or the bounding region estimate. The process may include determining at least one self-supervised loss based on the plurality of keypoint estimates and/or the bounding region estimate. The process may further include adapting, based on the at least one self-supervised loss, parameter(s) of the machine learning model. The process may include generating, using the machine learning model with the adapted parameter(s), a segmentation mask for the image (or another image).

Abstract: Systems and techniques are provided for processing one or more images. For instance, according to some aspects of the disclosure, a method may include obtaining an unlabeled image and generating at least one transformed image based on the unlabeled image. The method may include processing the unlabeled image using a pre-trained semantic segmentation model to generate a first segmentation output. The method may further include processing the at least one transformed image using the pre-trained semantic segmentation model to generate at least a second segmentation output. The method may include fine-tuning, based on the first segmentation output and at least the second segmentation output, one or more parameters of the pre-trained semantic segmentation model.

Abstract: Systems and techniques are provided for processing data (e.g., image data). For instance, according to some aspects of the disclosure, a method may include receiving, at a transformer of a machine learning system, learnable queries, keys, and values obtained from a feature map of a segmentation model of the machine learning system. The method may further include learning, via the transformer, a mapping between an unsupervised output and a supervised output of the segmentation model based on the feature map.

Abstract: A method for pruning weights of an artificial neural network based on a learned threshold includes determining a pruning threshold for pruning a first set of pre-trained weights of multiple pre-trained weights based on a function of a classification loss and a regularization loss. Weights are pruned from the first set of pre-trained weights when a first value of the weight is less than the pruning threshold. A second set of pre-trained weights of the multiple pre-trained weights is fine-tuned or adjusted in response to a second value of each pre-trained weight in the second set of pre-trained weights being greater than the pruning threshold.

Abstract: A transmitter may allocate resources across multiple links for a transmission to a receiver. The transmitter may transmit a resource grant to the receiver. The transmission may include a data packet allocated across multiple links, and pilot signals on a number of links. The receiver may use an indicator included with the resource grant to trigger measurement of channel quality for links with pilot signals.

Abstract: Techniques are described for wireless communication. A method for wireless communication at a user equipment (UE) includes identifying, while the UE is in a connected mode with a network, a radio resource configuration of the UE, selecting a dedicated set of resources for the UE or a common set of resources for a plurality of UEs based at least in part on the identified radio resource configuration, and transmitting a pilot signal to the network using the selected set of resources. Methods for wireless communication at a network access device and a network access device controller are also described.

Abstract: Methods, systems, and devices are described for fountain hybrid automatic repeat request (HARQ) for reliable low latency communication. A wireless device may transmit a data block based on a low latency operational mode. The device may then transmit a number of redundancy versions of the data block prior to determining whether an acknowledgement (ACK) has been received. In some examples the ACK may be an augmented ACK, which may be based on the number of redundancy versions received prior to successfully decoding the data block, and which may include an additional resource request. In some examples, the device may select an updated modulation and coding scheme (MCS) based on the augmented ACK. In some examples, the device may increase a number of frequency resources (e.g., component carriers) used for transmission based on the augmented ACK.

Abstract: Techniques are described for wireless communication. A method for wireless communication at a user equipment (UE) includes identifying, while the UE is in a connected mode with a network, a radio resource configuration of the UE, selecting a dedicated set of resources for the UE or a common set of resources for a plurality of UEs based at least in part on the identified radio resource configuration, and transmitting a pilot signal to the network using the selected set of resources. Methods for wireless communication at a network access device and a network access device controller are also described.

Abstract: Methods, systems, and devices are described for fountain hybrid automatic repeat request (HARD) for reliable low latency communication. A wireless device may transmit a data block based on a low latency operational mode. The device may then transmit a number of redundancy versions of the data block prior to determining whether an acknowledgement (ACK) has been received. In some examples the ACK may be an augmented ACK, which may be based on the number of redundancy versions received prior to successfully decoding the data block, and which may include an additional resource request. In some examples, the device may select an updated modulation and coding scheme (MCS) based on the augmented ACK. In some examples, the device may increase a number of frequency resources (e.g., component carriers) used for transmission based on the augmented ACK.

Abstract: Certain aspects of the present disclosure provide techniques for inferencing with a split inference model, including: generating an initial feature vector based on a client-side split inference model component; generating a modified feature vector by modifying a null-space component of the initial feature vector; providing the modified feature vector to a server-side split inference model component on a remote server; and receiving an inference from the remote server.

Abstract: The disclosure provides for an apparatus for wireless communications using carrier aggregation comprised of multiple carrier components. The apparatus can include a processor configured to generate one or more instances of a codeword from a data payload. In an aspect, the apparatus also includes a modulator configured to modulate the one or more instances of the codeword onto the multiple carrier components for transmission. In an aspect, the apparatus also includes a resource manager configured to provide the processor with a virtual carrier space comprising a logical carrier having a contiguous bandwidth equivalent to the aggregated bandwidth of the multiple carrier components. In an aspect, the process may be further configured to interleave at least one of the codeword instances across the multiple carrier components. In an aspect, the modulator may be configured to modulate the codeword instance onto the multiple carrier components in accordance to the interleaving.

Abstract: Techniques are described for wireless communication. A method for wireless communication at a user equipment (UE) includes identifying, while the UE is in a connected mode with a network, a radio resource configuration of the UE, selecting a dedicated set of resources for the UE or a common set of resources for a plurality of UEs based at least in part on the identified radio resource configuration, and transmitting a pilot signal to the network using the selected set of resources. Methods for wireless communication at a network access device and a network access device controller are also described.

Abstract: Techniques are described for wireless communication. A method for wireless communication at a user equipment (UE) includes identifying, while the UE is in a connected mode with a network, a radio resource configuration of the UE, selecting a dedicated set of resources for the UE or a common set of resources for a plurality of UEs based at least in part on the identified radio resource configuration, and transmitting a pilot signal to the network using the selected set of resources. Methods for wireless communication at a network access device and a network access device controller are also described.

Abstract: A method for pruning weights of an artificial neural network based on a learned threshold includes designating a group of pre-trained weights of an artificial neural network to be evaluated for pruning. The method also includes determining a norm of the group of pre-trained weights, and performing a process based on the norm to determine whether to prune the entire group of pre-trained weights.

Abstract: Methods, systems, and devices for wireless communication are described. A method may include measuring channel quality of a reference signal communicated via a control channel, determining a code rate for a control channel based on the measured reference signal, generating feedback data for the control channel based on the code rate, and transmitting the feedback data. Another method may include transmitting a reference signal in a control channel, receiving channel quality feedback data for the control channel in response to the reference signal, and transmitting a control channel transmission in the control channel using a modulation and coding scheme selected based on the feedback data.

Abstract: Various aspects described herein relate to communicating hybrid automatic repeat/request (HARQ) feedback. A HARQ communication can be received over a set of one or more links based on a first scheduling grant. HARQ feedback for the HARQ communication including at least one of HARQ feedback for one or more of the set of one or more links can be transmitted.

Abstract: A method for pruning weights of an artificial neural network based on a learned threshold includes determining a pruning threshold for pruning a first set of pre-trained weights of multiple pre-trained weights based on a function of a classification loss and a regularization loss. The first set of pre-trained weights is pruned in response to a first value of each pretrained weight in the first set of pre-trained weights being greater than the pruning threshold. A second set of pre-trained weights of the multiple pre-trained weights is fine-tuned or adjusted in response to a second value of each pre-trained weight in the second set of pre-trained weights being greater than the pruning threshold.

Abstract: Techniques are described for wireless communication. A first method includes measuring, by a first device, a condition of a wireless channel; and generating at least one channel side information feedback message based on the measured condition of the wireless channel. The at least one channel side information feedback message provides information on a relationship of a set of parameters, including a data rate parameter, an error probability parameter, and at least one of a deadline parameter or a transmission link parameter. A second method includes measuring, by a first device, interference on a wireless channel; identifying an interfering device for the wireless channel based on the measurement; and generating a channel side information feedback message based on the measured interference on the wireless channel. The channel side information feedback message indicates the interfering device for the wireless channel and a correlation of interference from the interfering device with time or frequency.

Abstract: Aspects described herein relate to measuring cross-link interference in wireless communications. A portion of a slot that includes a downlink reference signal transmitted by a base station and an uplink reference signal transmitted by a user equipment (UE) can be determined, where the uplink reference signal and the downlink reference signal at least partially overlap within the portion of the slot. The downlink reference signal can be received from the base station and the uplink reference signal can be received from the UE in the portion of the slot. Cross-link interference between the downlink reference signal and the uplink reference signal can be measured.

Abstract: Aspects directed towards a cell-specific configuration for sounding transmissions and measurements are disclosed. In one example, scheduled entities served by a scheduling entity are identified and a cell-specific sounding pattern configuration uniformly applicable to each of the scheduled entities is generated such that the cell-specific sounding pattern configuration has a different pattern than a cell-specific sounding pattern configuration associated with at least one other cell. The cell-specific sounding pattern configuration is then transmitted to each of the scheduled entities to facilitate an identification of cell-specific resources to use for at least one of a sounding transmission or a sounding measurement. In another example, a cell-specific sounding pattern configuration is received and cell-specific resources associated with the cell-specific sounding pattern configuration are identified.