Abstract: Domain-specific parameters may be used for tuning speech processing. A pre-trained transformer-based language model may train domain-specific parameters using domain-specific unlabeled text data. This domain-specific parameters can then be appended to candidate texts produced by a speech model on received speech data and input to the transformer-based language model to score the candidate texts. The scores of the candidate texts determined using the pre-trained transformer-based language model can then be used to select a candidate text for further speech processing.

Abstract: Systems and techniques are provided for processing video data. For example, an apparatus may obtain one or more first sets of collocated motion vector data and one or more second sets of collocated motion vector data, associated with a respective first and second block of video data included in a current frame of video data. The apparatus may project the one or more first sets of collocated motion vector data into a first projected motion field associated with a first buffer and project the one or more second sets of collocated motion vector data into the first projected motion field associated with the first buffer. Based on projecting the one or more first sets and one or more second sets of collocated motion vector data, the apparatus may decode the first block of video data based on the first projected motion field associated with the first buffer.

Abstract: Video decoding systems and techniques are described. The decoder applies a deblocking (DB) filter to the plurality of sub-blocks of a block of video data to generate a DB-filtered plurality of sub-blocks. The decoder applies the DB filter to one or more lines (e.g., columns) of pixels in an additional sub-block of the block to generate a DB-filtered portion of the additional sub-block. The one or more lines of pixels in the additional sub-block are filtered without filtering an entirety of the additional sub-block using the DB filter. The additional sub-block is adjacent to at least one of the plurality of sub-blocks. The decoder applies a constrained directional enhancement filter (CDEF) to the DB-filtered plurality of sub-blocks and the DB-filtered portion of the additional sub-block to generate a CDEF-filtered plurality of sub-blocks.

Abstract: A method can include training a machine learning algorithm to determine a duration of a new work order, based on (a) historical input data for the machine learning algorithm and (b) historical output data for the machine learning algorithm. The method can also include determining one or more work orders for a service provider comprising: determining, by the machine learning algorithm, as trained, one or more durations of the one or more work orders; and determining an optimized service route for the one or more work orders. The method can further include updating a work schedule for the service provider based on the optimized service route. Other embodiments are also provided.

Abstract: Embodiments disclosed herein include a motherboard. In an embodiment, the motherboard comprises a first layer with a first trace with a shape. In an embodiment, an insulating layer is provided over the first layer. In an embodiment, a second layer with a second trace with the shape is over the insulating layer. In an embodiment, the second trace is provided directly over the first trace.

Abstract: Systems and methods for efficient presentation of payment options at point of sale devices are disclosed. A method may include: identifying a plurality of stored or provisioned payment cards in customer wallets for a merchant; (2) pre-qualifying each of the customers associated with the stored/provisioned payment cards for one or more installment payment options for a purchase from the merchant; and (3) communicating a list of token reference identifiers for stored/provisioned payment cards that have been pre-qualified and the one or more pre-qualified installment payment options to a merchant computer program executed by a merchant backend, wherein the merchant computer program receives a token reference identifier for a stored or provisioned payment card in one of the customer wallets, determines that the token reference identifier is on the list of pre-qualified token reference identifiers, and presents the one or more installment payment options in a checkout user interface.

Abstract: This disclosure provides methods, apparatuses, and computer-readable mediums for visualizing coverage area deficiencies. In an aspect, a method comprises retrieving, from a data storage, network performance data from a plurality of user wireless mobile devices; filtering the network performance data to narrow the dataset to data samples with accurate locations and dates; counting the number of data samples indicating a coverage deficiency in an area; comparing the number of data samples indicating the coverage deficiency in the area with a predefined threshold; rendering a map including the area; and displaying on the map a visual indicator of the coverage deficiency based on the number of data samples indicating the coverage deficiency in the area exceeding the predefined threshold.

Abstract: A respiratory gas flow coupling includes a device connector having a female receptacle or male nipple with a side-oriented leak path. A patient accessory connector has a male nipple or female receptacle adapted to connect with the female receptacle or male nipple of the device connector. The patient accessory connector seals the side-oriented leak path of the female receptacle or male nipple of the device connector when the patient accessory connector is connected with the device connector. The side-oriented leak path is disposed at a position such that a shorter male nipple or female receptacle of an incompatible patient accessory connector which is shorter than the male nipple of the patient accessory connector does not seal the side-oriented leak path when the incompatible patient accessory connector with the shorter male nipple is connected with the device connector.

Abstract: A method including determining, via a machine learning model, a predicted energy load profile for a facility based at least in part on weather forecast data and sensor data for the facility. The sensor data can be received from one or more energy monitoring sensors for one or more devices in the facility. The method further can include determining one or more demand shedding time slots based at least in part on peak periods and the predicted energy load profile. Moreover, the method can include determining one or more demand shedding events for the one or more devices to be scheduled during the one or more demand shedding time slots. The method additionally can include causing a respective performance of each of the one or more demand shedding events by the one or more devices during the one or more demand shedding time slots. Other embodiments are disclosed.

Abstract: Systems and techniques are described herein for processing video data. For instance, a process can include obtaining a block of luma information for a block of video data. The process can further include applying a sample adaptive offset (SAO) filter to the luma information to generate SAO filtered luma information, inputting the SAO filtered luma information to a cross-component adaptive loop filter (CC-ALF) without storing the SAO filtered luma information to an external memory, applying the CC-ALF to the SAO filtered luma information to generate a first chroma CC-ALF output and a second chroma CC-ALF output, and storing the first chroma CC-ALF output and the second chroma CC-ALF output in the external memory.

Abstract: A device for upscale filtering video data includes one or more processors configured to generate a number of upscaled pixels that is an integer multiple of one half of a size of a filter used to upscale filter the video data. The device may then store the remaining pixels to be used to upscale filter a subsequent, right-neighboring block. In this manner, the device may avoid generating upscaled pixels that will result in being unaligned with the block and that may not be output due to being unaligned.

Abstract: Video decoding systems and techniques are described. A decoder reads first video data from a first block of a video frame. The decoder retrieves neighboring video data from a line buffer. The neighboring video data is from a neighboring block that neighbors the first block in the video frame. The decoder processes the first video data and the retrieved neighboring video data using a constrained directional enhancement filter (CDEF) to generate filtered first video data. The decoder upscales the filtered first video data using an upscaler to generate upscaled filtered first video data. The decoder upscales the retrieved neighboring video data using the upscaler to generate upscaled neighboring video data, for instance after generating the filtered first video data. The decoder processes the upscaled filtered first video data and the upscaled neighboring video data using a loop restoration (LR) filter to generate output video data.

Abstract: Some embodiments help manage cybersecurity insider risk. An authorized user influence pillar value is based on an influence signal representing the user's actual or potential influence in a computing environment. An authorized user access pillar value is based on an access signal representing the user's actual or potential access to resources. An impact risk value is calculated as a weighted combination of the pillar values. In response, an embodiment automatically adjusts a cybersecurity characteristic, such as a security risk score, security group membership, threat detection mechanism, or alert threshold. In some cases, impact risk is also based on a cumulative potential exfiltration anomaly access signal. In some cases, impact risk is based on one or more values which represent user public visibility, user social network influence, brand damage risk, resource mission criticality, access request response speed or success rate, or a known cybersecurity attack.

Abstract: Systems and methods for decoding are described. A decoder system is configured to retrieve, from the grain array stored within the on-chip memory, noise pixel data associated with at least one reconstructed video pixel. The decoder system is configured to apply noise data to the at least one reconstructed video pixel to generate at least one output video pixel. The noise data is determined based on the noise pixel data.

Abstract: Systems and techniques are provided for processing video data. For example, an apparatus may obtain one or more first sets of collocated motion vector data and one or more second sets of collocated motion vector data, associated with a respective first and second block of video data included in a current frame of video data. The apparatus may project the one or more first sets of collocated motion vector data into a first projected motion field associated with a first buffer and project the one or more second sets of collocated motion vector data into the first projected motion field associated with the first buffer. Based on projecting the one or more first sets and one or more second sets of collocated motion vector data, the apparatus may decode the first block of video data based on the first projected motion field associated with the first buffer.

Abstract: Video decoding systems and techniques are described. A decoder reads first video data from a first block of a video frame. The decoder retrieves neighboring video data from a line buffer. The neighboring video data is from a neighboring block that neighbors the first block in the video frame. The decoder processes the first video data and the retrieved neighboring video data using a constrained directional enhancement filter (CDEF) to generate filtered first video data. The decoder upscales the filtered first video data using an upscaler to generate upscaled filtered first video data. The decoder upscales the retrieved neighboring video data using the upscaler to generate upscaled neighboring video data, for instance after generating the filtered first video data. The decoder processes the upscaled filtered first video data and the upscaled neighboring video data using a loop restoration (LR) filter to generate output video data.

Abstract: Video decoding systems and techniques are described. The decoder applies a deblocking (DB) filter to the plurality of sub-blocks of a block of video data to generate a DB-filtered plurality of sub-blocks. The decoder applies the DB filter to one or more lines (e.g., columns) of pixels in an additional sub-block of the block to generate a DB-filtered portion of the additional sub-block. The one or more lines of pixels in the additional sub-block are filtered without filtering an entirety of the additional sub-block using the DB filter. The additional sub-block is adjacent to at least one of the plurality of sub-blocks. The decoder applies a constrained directional enhancement filter (CDEF) to the DB-filtered plurality of sub-blocks and the DB-filtered portion of the additional sub-block to generate a CDEF-filtered plurality of sub-blocks.

Abstract: A computing system includes a processor and memory coupled to the processor and storing instructions that, when executed by the processor provide a user interface module. The user interface module is configured to generate a tracker definition user interface having a threat parameter selection user interface element configured to receive a selection of at least one threat parameter, the tracker definition user interface also having a threat value user interface element configured to receive input specifying a threat value to match for the specified at least one threat parameter. The processor is configured to save a tracker based on the selection of at least one threat parameter and the threat value, and wherein the processor is configured to access a threat data store and execute the tracker against the threat data store to provide a tracker result.

Abstract: In one embodiment, an illustrative method herein may comprise: sending, from a server instrumentation agent configured on a transaction server, instrumented server performance data regarding the transaction server and an associated correlation token to an application programming interface (API) monitoring agent; receiving, at the server instrumentation agent, a returned acknowledgment of the instrumented server performance data regarding the transaction server and the associated correlation token from the API monitoring agent, the acknowledgement also having corresponding API monitored performance data regarding an API of the transaction server; returning, from the server instrumentation agent, a synchronization acknowledgment to the API monitoring agent in response to the returned acknowledgment to indicate complete synchronization; and correlating, by the server instrumentation agent, the instrumented server performance data regarding the transaction server and the API monitored performance data regarding t

Abstract: Time-based partitioning of a data set is applied to capture updates to the data set in a copy of the data set. Items that have been updated in a data set with in a time period are identified. Partitions of the data set that include the updated items are created according to a partitioning scheme. The created partitions are grouped in a storage location for the time period in a file structure that stores a copy of the database. A latest version of the copy of the data set may be accessed according to latest partitions of the data set stored in the different locations of the file structure.