Obtaining Biometric Information of a User Based on a Ballistocardiogram Signal Obtained When a Mobile Computing Devie is Held Against the Head of the User
A mobile computing device includes one or more memories to store one or more instructions. an inertial measurement unit. and one or more processors. The one or more processors execute the one or more instructions stored in the one or more memories to: control the inertial measurement unit to detect one or more motion signals generated when the mobile computing device is held against a head of a user of the mobile computing device. determine a ballistocardiogram signal based on the one or more motion signals detected by the inertial measurement unit. obtain, based on the ballistocardiogram signal. biometric information of the user, and output the biometric information of the user.
Disclosed implementations for detecting exploitation of ranking signals used to provide search results. An expected value for a ranking signal is determined based on a plurality of resources responsive to a query. A residual value is determined by aggregating a difference between the expected value and an information retrieval score for the ranking signal across a domain, wherein the domain includes at least one of the plurality of resources. Responsive to determining the residual value is indicative of an exploit, adjust a ranking of a resource associated with the domain in a search result page, the resource responsive to a second query based on the ranking signal.
G06F 16/908 - Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
Methods, systems, devices, and non-transitory computer readable media for training machine-learning models are provided. The disclosed technology can include receiving input samples associated with classification concepts. Based on inputting the input samples into a first plurality of machine-learned models, classification outputs comprising labels and confidence scores can be generated. The first plurality of machine-learned models can comprise one or more multimodal large language models (LLMs) and one or more domain-specific models. Annotated input samples comprising the input samples, the classification outputs, and identifiers that identify each of the first plurality of machine-learned models that generated each of the classification outputs can be generated. Furthermore, based on the annotated input samples, one or more second machine-learned models can be trained. The training can comprise modifying parameters of the one or more second machine-learned models based on the confidence scores.
A battery pack may use an integrated temperature sensor to monitor a temperature of a battery cell during charging and discharging. However, if the integrated temperature sensor fails, a smart home device may continue to charge and discharge the battery outside of its approved temperature range. This may lead to both safety and performance concerns. To identify a failed integrated temperature sensor, the device may leverage any additional temperature sensors that are located in the device. These temperature sensors may be used to externally measure or estimate the battery temperature. If a sufficient deviation between the measurements of these external temperature sensors and the measurements from the integrated temperature sensor is detected, the device may use the comparison of these temperature measurements to determine that the integrated temperature sensor may be malfunctioning. The device may then change its operational state in response to maintain performance and safety.
Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a first device may generate a multi-part neural network based channel state information feedback (CSF) message that comprises: a first part that indicates contents of a second part, and the second part; and transmit the multi-part neural network based CSF to a second device. Numerous other aspects are provided.
The present disclosure provides efficient communication techniques for transmission of model updates within a machine learning framework, such as, for example, a federated learning framework in which a high-quality centralized model is trained on training data distributed overt a large number of clients each with unreliable network connections and low computational power. In an example federated learning setting, in each of a plurality of rounds, each client independently updates the model based on its local data and communicates the updated model back to the server, where all the client-side updates are used to update a global model. The present disclosure provides systems and methods that reduce communication costs. In particular, the present disclosure provides at least: structured update approaches in which the model update is restricted to be small and sketched update approaches in which the model update is compressed before sending to the server.
Methods, systems and apparatus for measuring quantum state purity. In one aspect, a method for determining an average purity of multiple output quantum states, wherein the multiple output quantum states correspond to applications of respective random quantum circuits of a same circuit depth to a same initial quantum state, the method including: obtaining a plurality of data items, wherein each data item corresponds to a respective random quantum circuit of the same circuit depth and represents a probability that application of the respective random quantum circuit to the initial quantum state produces a respective measurement result; calculating a variance of a plurality of data items; determining a Porter-Thomas distribution having a dimension equal to a dimension of each output quantum state; and dividing the calculated variance by a variance of the Porter-Thomas distribution to determine the average purity.
Methods and systems for customized query responses using artificial intelligence are provided. A first request to perform an operation associated with an artificial intelligence (AI) model is received from a first user of a platform. A first adapter model associated with at least one of the first user or the first contextual data pertaining to the first request is identified. A model pipeline associated with the AI model is updated to include the identified first adapter model. A prompt including the first request to perform the operation is provided as input to the first adapter model. An output of the first adapter model is used by the AI model. A first output of the AI model is obtained. A first response to the first request is provided to the user. The first response is based on the first output of the AI model.
Methods and systems for customized query responses using artificial intelligence are provided. A request is received from a client device of a user associated with a client account to perform an operation associated with an artificial intelligence (AI) model. An adapter model associated with the client account is identified. The adapter model is trained to modify parameters of the AI model based on electronic documents having a preferred style or a preferred format of the client account. A prompt including the request to perform the operation as an input to the adapter model. An output of the adapter model is used by the AI model. An output of the AI model is obtained, the output having at least one of the preferred style or the preferred format of the client account. A response to the request is provided using the obtained output of the AI model.
Systems and methods for performing inference for word or wordpiece tokenization are disclosed using a left-to-right longest-match-first greedy process. In some examples, the vocabulary may be organized into a trie structure in which each node includes a precomputed token or token_ID and a fail link, so that the tokenizer can parse the trie in a single pass to generate a list of only those tokens or token_IDs that correspond to the longest matching vocabulary entries in the sample string, without the need for backtracking. In some examples, the vocabulary may be organized into a trie in which each node has a fail link, and any node that would share token(s) or token_ID(s) of a preceding node is instead given a prev_match link that points back to a chain of nodes with those token(s) or token_ID(s).
Aspects of self-adjusting aware thermal control of a semiconductor device are disclosed. For example, a central unit may be coupled with an element of the semiconductor device and one or more temperature controllers configured to sequentially apply throttling steps to thermally control the element. The throttling steps are sequentially applied based on individual throttling tables. The central unit has access to the individual throttling tables and may access a current performance state of the element. The central unit may command one or more of the temperature controllers to throttle the element based on the current performance state of the element. The central unit may command one or more of the temperature controllers to apply a throttling step to the element based on throttling steps previously applied to the element. The temperature controllers may include memory to store a current throttling status of the element communicated by the central unit.
Techniques are disclosed that enable generating jointly probable output by processing input using a multi-stream recurrent neural network transducer (MS RNN-T) model. Various implementations include generating a first output sequence and a second output sequence by processing a single input sequence using the MS RNN-T, where the first output sequence is jointly probable with the second output sequence. Additional or alternative techniques are disclosed that enable generating output by processing multiple input sequences using the MS RNN-T. Various implementations include processing a first input sequence and a second input sequence using the MS RNN-T to generate output. In some implementations, the MS RNN-T can be used to process two or more input sequences to generate two or more jointly probable output sequences.
A method includes identifying, by a processing device of a content sharing platform, a comment associated with a media item on the content sharing platform. A prompt is provided as input to an artificial intelligence (AI) model to cause the AI model to generate a reply to the comment. An output of the artificial intelligence (AI) model is received. Based on the output, a reply window is pre-filled with a reply associated with the comment.
H04N 21/4788 - Supplemental services, e.g. displaying phone caller identification or shopping application communicating with other users, e.g. chatting
Techniques for maintaining a knowledge graph of a user, such as user log, are presented herein. The system can include a database storing the knowledge graph having a plurality of structured data entries. The system can include a machine-learned model configured to determine an insight. The system can obtain, from a user device, a first user entry. Additionally, the system can process, using the machine-learned model, the first user entry with a prompt to generate a first structured data entry. Moreover, the system can process, using the machine-learned model, the first user entry to determine a layout of a graphical user interface that is presented on the user device. Furthermore, the system can process the first structured data entry with the knowledge graph to determine the first insight. Subsequently, the system can cause a presentation of the first insight on the graphical user interface of the user device.
Methods, systems, and media comprising: one or more non-scan elements, wherein each non-scan element has respective fanout logic comprising one or more scan flops; and a testing control module configured to block unknown data propagating from the one or more non-scan elements to the scan flops in the fanout logic by providing a scan-enable signal to the scan flops during a capture phase of a testing process.
Techniques and devices for determining a central node for reporting sensor data are described for an electronic device that inserts ranges between nodes in the wireless network into a Euclidean distance matrix (EDM) and decodes the EDM to generate a global topology for the nodes in the wireless network. The electronic device sums, for each node in the wireless network, events detected by each node during a predetermined time period and performs a kernel density filtering of the sums of the detected events over a two-dimensional space of the global topology. The electronic device calculates a product of Gaussian distributions calculated during the kernel density filtering and selects the node that is spatially closest to a peak of the product of Gaussian distributions as the central node for event reporting.
Smart home devices may use a technique known as “power stealing” in order to steal power from an external environmental system. For example, thermostats may steal power from an HVAC system. Different algorithms and techniques may be used for efficiently stealing power from the HVAC system, each of which may provide different levels of power to the thermostat at different times. The smart home device may test an external system to determine which power stealing methods are compatible and calibrate various power stealing parameters. A calibration routine may sample at a plurality of discrete intervals while increasing a test load to determine a maximum current limit and an optimal power stealing method.
Feature sets of smart home devices may be activated based on whether they can be supported by a selected power sourcing method. For example, thermostats may use a technique known as “power stealing” in order to steal power from the HVAC system. Different algorithms and techniques may be used for efficiently stealing power from the HVAC system, each of which may provide different levels of power to the thermostat at different times. The smart home device may test an external system to determine which power stealing methods are compatible, then select predetermined feature sets that are compatible with the available power stealing methods.
Implementations relate to storing historical queries processed using a generative model in association with intermediate model outputs generated using the generative model for each of the historical queries. Implementations further relate to receiving a user query processable using the generative model. In response to receiving the user query, the user query can be compared to the historical queries to identify a particular historical query (e.g., having a similarity score satisfying a similarity threshold) that matches the user query. Particular intermediate model output associated with the particular historical query can be selected from all intermediate model outputs stored in association with the particular historical query, and a response to the user query can be generated based at least on the selected particular intermediate model output associated with the particular historical query.
A radio access network. RAN, participating in a multicast and broadcast services. MBS, session can implement a method for managing MBS communication. The method includes: transmitting (702), to a first UE operating in a connected state, a first multicast configuration for receiving MBS data in the inactive state: transmitting (703), to a second UE operating in the connected state, a second multicast configuration for receiving the MBS data in the connected state; and transmitting (708) the MBS data to the first UE operating in the inactive state, according to the first multicast configuration and the second UE operating in the connected state, according to the second multicast configuration.
The present disclosure provides computer-implemented methods, systems, and devices for enabling search in an augmented reality interface. A computing device generates an interface depicting an AR view including image data of at least a portion of a physical real-world environment for display by the computing device. The computing device displays one or more filter elements within the interface, a respective filter element being associated with a point of interest type. The computing device accesses, from a database of geographic locations, data describing a plurality of points of interest within the physical real-world environment. The computing device receives a selection of one the displayed filter elements. The computing device provides, for display in the AR view, a set of augmented reality elements associated with a set of points of interest, wherein the set of augmented reality elements represents a filter-based set of points of interest associated with the filter element.
Various arrangements for reducing auditory spatial aliasing for a user are detailed herein. A first delay filter may be set that delays output of a first audio signal by a first duration to a speaker of a device compared to a second speaker. A second delay filter may also be set that delays output of a second audio signal by a second duration. The first and second audio signals can be output by the speakers.
Implementations described herein relate to modifying depth of field in images using machine learning. In some implementations, a computer-implemented method for training a machine learning model includes generating an input training image that is a composition of multiple images captured in focus stacks at different lens focus positions and camera distances. A corresponding ground truth image is generated from merged images in particular focus stacks. A convolutional neural network (CNN) machine learning (ML) model receives the input training image and outputs an output image that adjusts blurriness in the input training image to simulate a target depth of field. The CNN ML model is updated based on comparison of the output image and the ground truth image. The CNN ML model can include a depth CNN that performs an implicit depth estimation for features of the input image, and a deconvolution CNN that adjusts the blurriness.
A computing system receives a transcript for a video and an input indicative of a request to adjust a playback position of the video, in which the request does not specify a timestamp of the video to which to adjust the playback position. The computing system applies, based on the request to adjust the playback position, a first machine learning model to the transcript and a current timestamp of the video to identify one or more noncurrent time stamps. The computing system applies a second machine learning model to the transcript, the current timestamp, and the one or more noncurrent time stamps to rank, based on user data, the one or more noncurrent time stamps. The computing system then adjusts, based on the ranking of the one or more noncurrent timestamps, the playback position to a noncurrent timestamp from the one or more noncurrent timestamps.
Techniques and apparatuses are described that implement two-stage thermal throttling. In some examples, two-stage thermal throttling of a mobile device is achieved using a main controller and an auxiliary controller. The auxiliary controller can be a proportional controller that monitors a temperature and a rate of change of the temperature of the mobile device during operations. When a first temperature threshold and a threshold rate of increase of the temperature are exceeded by the mobile device, the auxiliary controller can throttle a metric of the device to slow down the rate of increase of the temperature. After the temperature has exceeded a second temperature threshold, the auxiliary controller can hand off control to the main controller, which can further throttle the metric of the device or one or more additional metrics of the device.
According to at least one implementation, a method includes identifying a command from a user of a device. In response to the command, the method further includes identifying an image associated with a gaze of the user and identifying an action based on an application of a language model to the command and the image, the application of the language model including an identification of an object for the command in the image.
Various implementations include fine-tuning a multilingual large language model (ML-LLM). Many implementations include converting a base instance of natural language (NL) input text into a revised instance of NL input text, where the base instance of NL input text is in a first language and includes a portion corresponding to a first geographic location, and where the revised instance of NL input text is in a second language and includes a portion corresponding to a second geographic location.
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using memory-optimized contrastive learning to train image encoder and text encoder neural networks.
G06V 10/764 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
A computing device may activate an application that is operable to output a user interface in a second interface orientation and is not operable to output the user interface in a first interface orientation. The computing device may determine, for the application, a re-oriented user interface in the first interface orientation. The computing device may output the re-oriented user interface for display at the display device in the first interface orientation.
G06F 3/04845 - Interaction techniques based on graphical user interfaces [GUI] for the control of specific functions or operations, e.g. selecting or manipulating an object, an image or a displayed text element, setting a parameter value or selecting a range for image manipulation, e.g. dragging, rotation, expansion or change of colour
G06F 1/16 - Constructional details or arrangements
30.
INTERLEAVING COMMANDS AND DATA WRITES TO A PROCESSING-IN-MEMORY ARCHITECTURE TO OPTIMIZE EXECUTION OF IN-MEMORY COMPUTATIONS
Methods and systems for hardware and software techniques for interleaving compute commands and input data write operations to a PiM module to optimize execution of in memory computations. A PiM compute module executes the interleaved commands to perform multiply and accumulate (MAC) operations using parameters read from a memory array and input data read from an input queue. A memory controller interleaves the commands and input data writes to the PIM module based on the duration of the MAC operations and/or a timing parameter for issuing write commands or for consuming data from the input queue. The memory controller can interleave different types of commands concurrent with interleaving the commands with the input data write operations. The interleaving operations are timed to maintain a threshold quantity of input data in the input queue to minimize (or prevent) data underflow or overflow during execution of the MAC operations.
An example method includes generating, at a heat map generation frequency, one or more saliency heat maps associated with a video of a scene being captured by an image capturing device. The method also includes detecting, based on the one or more saliency heat maps, a salient object in the scene. The method additionally includes responsive to the detecting of the salient obj ect: initiating a tracking of a region of interest (ROI) associated with the salient object in subsequently captured video of the scene, and reducing the heat map generation frequency for generation of saliency heat maps for the subsequently captured video of the scene. The method also includes adjusting, based on the tracked ROI, an automatic image capture setting of the image capturing device.
H04N 23/61 - Control of cameras or camera modules based on recognised objects
G06V 10/46 - Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]Salient regional features
G06V 10/70 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning
32.
SYSTEMS AND METHODS FOR EXTENDING A DEPTH-OF-FIELD BASED ON FOCUS STACK FUSION
An example method includes determining that a portion of a scene in an image frame being captured at a first focal length is out of focus. The method also includes capturing one or more first image frames at the first focal length and one or more additional image frames at a second focal length to focus on the portion of the scene. The method additionally includes providing the one or more first image frames and the one or more additional image frames as input to a machine learning (ML) model, the ML model having been trained to merge one or more focused regions in a plurality of input images to predict an output image with an extended depth of field (DoF). The method also includes receiving the predicted image from the ML model.
H04N 23/951 - Computational photography systems, e.g. light-field imaging systems by using two or more images to influence resolution, frame rate or aspect ratio
33.
CROSS TECHNOLOGY SPECIFIC ABSORPTION RATE MANAGEMENT TECHNIQUES
A user equipment (UE) in a mobile cellular network implements one or more techniques to perform cross-technology specific absorption rate (SAR) budget allocation. The UE computes an individual SAR budget for each radio access technology (RAT) module of a plurality of RAT modules at the UE based on operational context data associated with the UE. The UE sends each individual SAR budget to a corresponding RAT module of the plurality of RAT modules. The UE then adjusts, at one or more RAT modules of the plurality of RAT modules, transmission parameters based on the corresponding individual SAR budget.
Aspects of the disclosure are directed to managing files in data lakehouses using rewrite-free loading. Rewrite-free loading includes keeping track of information from data files that could be missing when imported to the data lakehouses without having to perform full-copy loading. Rewrite-free loading can store this information in table metadata or augment headers and/or footers of the data files with this information when importing to a data lakehouse. Rewrite-free loading allows for more accurate management of data lakehouses with lower computational costs.
A method includes generating, using an AI model, a first object embedding of a first threat intelligence (TI) data object that includes first one or more cybersecurity attributes of a business entity. The method includes obtaining one or more second object embeddings that each represents a respective second TI data object that includes second one or more cybersecurity attributes of a cybersecurity threat. The method includes, for each second object embedding, generating a respective similarity value reflecting a similarity between the first object embedding and the respective second object embedding. The method includes ranking, based on the similarity values, the one or more second TI data objects. The method includes identifying, based on the ranking, a subset of the one or more second TI data objects that are relevant to the first TI data object.
Methods, systems, devices, and non-transitory computer readable media for generating embeddings are provided. The disclosed technology can include receiving multimodal input samples associated with data modalities and labels. The multimodal input samples can comprise topics associated with topics of multimodal input samples. Based on inputting multimodal input samples into modality-specific machine-learned models configured to process data modalities, modality-specific embeddings can be generated. Each multimodal input sample of the multimodal input samples can be inputted into a modality-specific model that is configured to process the data modality associated with the multimodal input sample. The modality-specific embeddings can comprise topic embeddings based on the topics. Based on the plurality of modality-specific embeddings, multimodal machine-learned models can be trained to generate a plurality of common embeddings. Based on inputting the multimodal input samples into the multimodal machine-learned models, the common embeddings can be generated.
A system includes a housing, a pivotable arm pivotably attached to the housing; a driver coupled to the pivotable arm, a biasing element coupled between the housing and the pivotable arm; a cam rotatably coupled with the driver; a cam follower associated with the cam; and a pusher coupled to the cam follower such that the pusher is configured to translate when the cam is rotated by the driver.
A system and method for fitting a head mounted wearable device for a user based on a single two-dimensional image is provided. The image may include the face/head of the user, captured by an image sensor of a computing device, via an application executing on the computing device. A sellion node, of a plurality of nodes of a reference mesh, may be mapped to a sellion node, of a plurality of nodes, of a user mesh. The reference mesh may represent a general head mesh based on data collected from a large pool of users. The user mesh may be generated from the two-dimensional image. A positioning of a virtual frame on the two-dimensional image of the user may be adjusted based on a difference in position of the sellion node of the reference mesh and the sellion node of the user mesh.
Offline calibration of an inertial measurement unit (IMU) can determine biases in the motions measured by the IMU while it is not in use. The offline calibration uses an expected motion measurement based on a motionless IMU as a reference from which the biases can be computed for a temperature. The bias and the temperature can be stored in a thermal table that can be updated and expanded over multiple calibration sessions to include the biases for a range of temperatures. A model relating the biases to temperature may be created based on the thermal table. For example, a curve-fit equation relating the bias as a function of temperature may be computed based on the values in the thermal table.
G01C 21/16 - NavigationNavigational instruments not provided for in groups by using measurement of speed or acceleration executed aboard the object being navigatedDead reckoning by integrating acceleration or speed, i.e. inertial navigation
G01C 25/00 - Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass
A cooling apparatus for an optical module includes a pedestal including a first pedestal surface and a second pedestal surface, wherein the first pedestal surface is attached to an outer surface of an optical module. A first boss and a second boss extends from opposing ends of the second pedestal surface, the first and the second boss define a first and a second opening having respective first and second through-apertures extending from the first and second openings through the pedestal to the first pedestal surface. A first manifold is attached to the first boss and configured to supply a liquid to the first through-aperture. A second manifold attached to the second boss and configured to receive the liquid from the second through-aperture.
A system and method for implementing user and entity behavioral analytics (UEBA) in a cybersecurity analytics platform. An example method includes receiving, by one or more processing devices of a security analytics platform, security data associated with a specified entity; generating, based on at least a subset of the security data, one or more security signals associated with the specified entity and occurring within a specified time window; computing, for each security signal of the one or more security signals, a respective risk score associated with the specified time window; computing, by aggregating risk scores associated with the one or more security signals, a risk score associated with the specified entity for the specified time window; and modifying, based on an attribute of a security watchlist associated with the specified entity, the risk score of the specified entity.
Methods, systems, and apparatuses, including computer programs encoded on computer storage media, for processing a video to generate guided content. Then presenting the guided content during video playback along with responses to user queries. In particular, the described techniques use multi-modal neural networks to process the video to generate summaries, question prompts, responses to question prompts, and responses to user queries that take into account video context, previous user queries, or both. As a result, the described techniques increase video playback efficiency by presenting engaging guided content that enhance user video playback experience and by presenting responses to user queries that are maximally relevant to the user in real-time.
The present disclosure is directed to systems and methods for generating synthetic training data using augmented reality (AR) techniques. For example, images of a scene can be used to generate a three-dimensional mapping of the scene. The three-dimensional mapping may be associated with the images to indicate locations for positioning a virtual object. Using an AR rendering engine, implementations can generate an augmented image depicting the virtual object within the scene at a position and orientation. The augmented image can then be stored in a machine learning dataset and associated with a label based on aspects of the virtual object.
A computing system for detecting objects in an image can perform operations including generating an image pyramid that includes a first level corresponding with the image at a first resolution and a second level corresponding with the image at a second resolution. The operations can include tiling the first level and the second level by dividing the first level into a first plurality of tiles and the second level into a second plurality of tiles; inputting the first plurality of tiles and the second plurality of tiles into a machine-learned object detection model; receiving, as an output of the machine-learned object detection model, object detection data that includes bounding boxes respectively defined with respect to individual ones of the first plurality of tiles and the second plurality of tiles; and generating image object detection output by mapping the object detection data onto an image space of the image.
G06V 10/764 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
G06V 10/75 - Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video featuresCoarse-fine approaches, e.g. multi-scale approachesImage or video pattern matchingProximity measures in feature spaces using context analysisSelection of dictionaries
G06V 10/778 - Active pattern-learning, e.g. online learning of image or video features
A browser-based tool is disclosed for providing context-based assistance during web browsing. An example method involves receiving a prompt pertaining to main content displayed in a first display area, extracting content from the main content, receiving generated content based on the extracted content, and displaying the generated content in a second display area while the main content remains displayed in the first display area. This innovative approach streamlines the search process by providing users with relevant generated content based on the content they are currently viewing, thereby improving efficiency in navigating online information.
Systems and methods for performing captioning for image or video data are described herein. The method can include receiving unlabeled multimedia data, and outputting, from a machine learning model, one or more captions for the multimedia data. Training the machine learning model to create these outputs can include inputting a subset of video frames and a first utterance into the machine learning model, using the machine learning model to predict a predicted utterance based on the subset of video frames and the first utterance, and updating one or more parameters of the machine learning model based on a loss function that compares the predicted utterance with the second utterance.
Systems, methods and apparatuses include techniques for a user equipment (UE) (130) to select antenna ports for determining phase offset PO in multiple transmission and reception point (multi-TRP) operation, The UE may receive, from a network entity (120), at least one channel state information (CSI) report configuration (104) including at least one of:one or more CSI reference signal (CSI-RS) resources for channel measurement, and one or more sounding reference signal (SRS) parameters associated with antenna port selection for phase offset (PO) reporting. The UE may receive, from the network entity, the one or more CSI-RS resources. The UE may transmit, to the network entity, at least one PO report (116) based on the at least one CSI report configuration and the one or more CSI-RS resources.
An example foldable display assembly includes a first housing. The foldable display assembly further includes a second housing. The foldable display assembly further includes a hinge mechanism coupled to the first housing and the second housing, the hinge mechanism including one or more tabs. The foldable display assembly further includes a continuous display connected to the first housing and the second housing, tire continuous display spanning the hinge mechanism. The foldable display assembly further includes a hinge cover including one or more snap engagement points that are positioned to engage with the one or more tabs.
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating updated output sequences using a language model neural network. One of the methods includes maintaining a data store that stores a plurality of prompt vectors, wherein the plurality of prompt vectors correspond to historic prompts that have been received by a language model neural network; obtaining a new document; a query vector for the new document; performing a search in the data store for one or more most similar prompt vectors to the query vector according to a similarity measure; an input sequence based on (i) the one or more historic prompts that correspond to the one or more most similar prompt vectors and (ii) the new document; and processing, using the language model neural network, the input sequence to generate an updated output sequence.
Techniques are described herein for progressive training of a machine learning network for name embedding. Embodiments seek to refine embedding models so that automated name detection, automated attention handling, and other similar features can be applied to active noise control systems in a manner that is robust to noise and competing speech. Embodiments begin by training a foundation model based on a name detection paradigm. Progressive training is used, based initially on the foundation model, to teach progressive machine learning networks to generate unified embeddings for each of multiple linguistic classes robustly in the presence of noise and/or competing speech. Those networks are ultimately used to train a robust name embedding (RNE) model to produce target outputs (e.g., classifications, acoustic segments, etc.) according to the name detection paradigm.
Machine learning models that operate on images can exhibit significant increases in their cost to execute as the size (e.g., resolution, number of pixels) of the images increase. While it is possible to downsample input images and perform some or all of the model image processing in a lower-resolution space, followed by upsampling, the results of such operation have previously been poor. Embodiments are provided that overcome these limitations, resulting in decreased computational cost without decreasing output image quality. These benefits are obtained, in part, by combining pixels of an input image (e.g., by concatenation) into an effectively lower-resolution image space, performing computations thereon, and then de-concatenating or otherwise separating the combined pixels to generate an output image. TTiis can allow individual processor elements of a TPU to efficiently implement a machine learning model, thereby improving output image quality while limiting computational costs to those available even on resource-constrained platforms.
A distributed unit (DU) of a distributed base station operating in a radio access network (RAN) can implement a method for managing discontinuous reception (DRX) for a multicast and broadcast services (MBS) session. The method includes: (i) receiving, by processing hardware and from a CU of the distributed base station, information related to configuring DRX for the MBS session; (ii) generating, by the processing hardware and based at least on the information related to configuring DRX, a DRX configuration for a user equipment (UE) participating in the MBS session; and (iii) transmitting, by the processing hardware, the DRX configuration to the UE.
A central unit (CU) of a distributed base station transmits, to a core network (CN), a distribution setup request message: receives, from the CN, a distribution setup response message including a first MBS QoS flow configuration; and transmits, to a distributed unit (DU) of the distributed base station, a multicast context setup request message including a second MBS QOS flow configuration based on the first MBS QoS flow configuration, to establish a multicast context for an MBS session.
This disclosure provides systems, devices, apparatus, and methods, including computer programs encoded on storage media, for beam indication techniques for ICM. A UE (102) receives (310), from a first network entity (304), an indication of a beam associated with a second network entity (305). The indication of the beam corresponds to at least one of a TCI state associated with non-dedicated signaling from the second network entity (305) or an activation delay time for the beam based on whether the beam corresponds to dedicated signaling from the second network entity (305) to the UE (102) or the non-dedicated signaling from the second network entity (305). The UE (102) attempts to receive a downlink communication from the second network entity (305) based on the indication.
H04B 7/06 - Diversity systemsMulti-antenna systems, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas at the transmitting station
H04L 5/00 - Arrangements affording multiple use of the transmission path
H04W 76/16 - Setup of multiple wireless link connections involving different core network technologies, e.g. a packet-switched [PS] bearer in combination with a circuit-switched [CS] bearer
58.
MICROLED DISPLAY WITH UNIFORM FEATURE THICKNESS AND METHOD FOR MANUFACTURING THE SAME
In a general aspect, a microLED display (600) includes a semiconductor member (605a) having a thickness, where the semiconductor member has a LED side (602a) configured to produce light (670) for displaying an image, and an output side (602b) configured to display the image by outputting the produced light, the output side being opposite the LED side. The display also includes a plurality of microLED mesas (627) included on the LED side of the semiconductor member, and a plurality of etched features (655a, 655b, 655c) defined in the semiconductor member. The plurality of etched features are defined on at least one of the LED side or the output side. The plurality of etched features define un-etched portions in the semiconductor member having respective thicknesses that are less than the thickness of the semiconductor member. The respective thicknesses of the un-etched portions are uniform across the display, with a total thickness variation less than 200 nanometers.
In various implementations. audio data that captures a spoken utterance of a first user is received. The audio data being is generated by one or more microphones of a transcription device and is received while at least one first signal, rendered by a first signaling device responsive to a determination that the first user is speaking. is received by the transcription device. A transcription comprising recognized text from the spoken utterance of the first user is generated based on performance of automatic speech recognition on the audio data, and is annotated to indicate that the recognized text from the spoken utterance of the first user is associated with a first identifier corresponding to the at least one first signal, based at least in part on receiving the audio data while receiving the at least one first signal. The annotated transcription can be provided for output.
Methods and devices are presented for transforming a layout of a densely packed grid of micro-LED light emitters to a layout of a square rectilinear pixel grid to achieve compatibility with hardware and software used in imaging and display technologies. In particular, a pattern of regular hexagonal emitter cells for fabrication on a III-nitride substrate can be transformed to a square pixel array of irregular hexagonal trichrome pixels that are readily addressable. Separation between adjacent trichrome pixels, and between their constituent emitters, can be established for overlay tolerance, while maintaining a cell packing density of about 70% and a pixel pitch of about 4.0 μm. Wavelength and quantum efficiency properties are shown to depend on optical current density, which can be determined by the emitter area specified in the grid layout.
H10H 20/812 - Bodies having quantum effect structures or superlattices, e.g. tunnel junctions within the light-emitting regions, e.g. having quantum confinement structures
H10H 29/14 - Integrated devices comprising at least one light-emitting semiconductor component covered by group comprising multiple light-emitting semiconductor components
61.
USING USER INPUT TO ADAPT SEARCH RESULTS PROVIDED FOR PRESENTATION TO THE USER
Methods, apparatus, and computer readable media related to interaction between a user and an automated assistant during a dialog between the user and the automated assistant. Some implementations are directed to adapting a graphical and/or audible presentation of search results provided by the automated assistant for presentation to the user. The adaptation may be in response to attribute(s), of one or more of the search results, referenced in spoken and/or typed textual input provided by the user during the dialog. Some of those implementations may enable a user to provide textual input to navigate the search results within the dialog and within resource and/or interface constraints associated with the dialog. Some of those implementations may additionally and/or alternatively enable adapting, based on textual input provided by a user to the automated assistant, when and/or whether search results having certain attributes are provided to the user by the automated assistant.
G06F 3/0484 - Interaction techniques based on graphical user interfaces [GUI] for the control of specific functions or operations, e.g. selecting or manipulating an object, an image or a displayed text element, setting a parameter value or selecting a range
Methods and apparatus for detecting hazardous vehicle events and encouraging usage of driving optimized application features to mitigate occurrence of the hazardous vehicle events. The driving optimized application features can address unsafe driving events that are determined to be correlated with certain distracting application features. For example, an application of a computing device can determine that a user is occupying a vehicle and is driving toward a destination. While driving, data available to the application can indicate that an unsafe driving event, such as a hard braking event, has occurred while the user was interacting with another application. Thereafter, and based on this data, the application can render an output characterizing the correlation between the hard braking event and the other application, and/or provide the user with an option to interact with the other application via driving optimized feature(s).
B60K 35/26 - Output arrangements, i.e. from vehicle to user, associated with vehicle functions or specially adapted therefor using acoustic output
B60K 35/28 - Output arrangements, i.e. from vehicle to user, associated with vehicle functions or specially adapted therefor characterised by the type of the output information, e.g. video entertainment or vehicle dynamics informationOutput arrangements, i.e. from vehicle to user, associated with vehicle functions or specially adapted therefor characterised by the purpose of the output information, e.g. for attracting the attention of the driver
G01C 21/36 - Input/output arrangements for on-board computers
G06F 3/0481 - Interaction techniques based on graphical user interfaces [GUI] based on specific properties of the displayed interaction object or a metaphor-based environment, e.g. interaction with desktop elements like windows or icons, or assisted by a cursor's changing behaviour or appearance
G06F 3/04842 - Selection of displayed objects or displayed text elements
A microelectronic system may include a microelectronic component having electrically conductive elements exposed at a first surface thereof, a socket mounted to a first surface of the microelectronic component and including a substrate embedded therein, one or more microelectronic elements each having active semiconductor devices therein and each having element contacts exposed at a front face thereof, and a plurality of socket pins mounted to and extending above the substrate, the socket pins being ground shielded coaxial socket pins. The one or more microelectronic elements may be disposed at least partially within a recess defined within the socket. The socket may have a land grid array comprising top surfaces of the plurality of the socket pins or electrically conductive pads mounted to corresponding ones of the socket pins, and the element contacts of the one or more microelectronic elements may be pressed into contact with the land grid array.
H01L 25/10 - Assemblies consisting of a plurality of individual semiconductor or other solid-state devices all the devices being of a type provided for in a single subclass of subclasses , , , , or , e.g. assemblies of rectifier diodes the devices having separate containers
H01L 23/538 - Arrangements for conducting electric current within the device in operation from one component to another the interconnection structure between a plurality of semiconductor chips being formed on, or in, insulating substrates
H01L 25/00 - Assemblies consisting of a plurality of individual semiconductor or other solid-state devices
Systems and methods are disclosed that address the need for adaptive exposure within high dynamic range (HDR) images. Solutions can leverage recent advances in the use of virtual reality (VR) headsets and Augmented Reality (AR) displays equipped with infrared (IR) eye tracking devices. A gaze vector determined by the eye tracking device identifies one or more fixation points on the image that corresponds to an area where there exists a faulty exposure. The exposure around the fixation point can be adaptively corrected using image processing techniques. Using spatial adaptive exposure, the resulting image, a type of foveated image, can be rendered on a low dynamic range (LDR) display with sufficient detail.
Systems and methods for generating and providing a virtual walkthrough interface can include generating a virtual walkthrough video based on view synthesis renderings generated by neural radiance field model. The neural radiance field model can be trained based on a plurality of images of an environment and may generate the view synthesis renderings based on processing positions along a determined walkthrough path. The generated virtual walkthrough video can then be scrubbed through to provide the virtual walkthrough interface.
G06T 19/00 - Manipulating 3D models or images for computer graphics
G06F 16/787 - Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using geographical or spatial information, e.g. location
To adjust an aspect ratio of an image to match the aspect ratio of a display area for presenting the image, a computing device receives an image having a first aspect ratio, and obtains a second aspect ratio for a display area of a display in which to present the image, where the second aspect ratio is different from the first aspect ratio. The computing device extends the image to include one or more additional features which were not included in the image. Additionally, the computing device automatically crops the extended image around an identified region of interest by selecting a portion of the extended image that has an aspect ratio which matches the second aspect ratio of the display area, and provides the cropped image for presentation within the display area of the display.
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating video anchors for a video. In one aspect, a method includes obtaining key moment identifiers for a video, where each key moment identifier includes a time index value specifying a playback time in the video, and is indicative subject matter of the video that has been determined to meet one or more interest criteria that define salient topics within the video. For each key moment identifier, a video anchor is generated, where each video anchor indicates a playback time for the video, and may include an image of a frame that occurs near the playback time. Upon a selection of the video anchor by the user, an instruction in the video anchor causes a video player to begin playback of the video at the playback time specified by the video anchor.
To provide personalized data for display on a map, a server device obtains location data for a user and identifies locations that are familiar to the user based on the frequency and recency in which the user visits the locations. The server device then provides the familiar locations in search results/suggestions and annotates the familiar locations with a description of a relationship between the familiar location and the user. The service device also includes the familiar locations as landmarks for performing maneuvers in a set of navigation instructions. Furthermore, the server device provides a familiar location as a frame of reference on a map display when a user selects another location nearby the familiar location. Moreover, the server device includes a familiar location as an intermediate destination when the user request navigation directions to a final destination.
Example aspects of the present disclosure are directed to computing systems and methods for hand tracking using a machine-learned system for palm detection and key-point localization of hand landmarks. In particular, example aspects of the present disclosure are directed to a multi-model hand tracking system that performs both palm detection and hand landmark detection. Given a sequence of image frames, for example, the hand tracking system can detect one or more palms depicted in each image frame. For each palm detected within an image frame, the machine-learned system can determine a plurality of hand landmark positions of a hand associated with the palm. The system can perform key-point localization to determine precise three-dimensional coordinates for the hand landmark positions. In this manner, the machine-learned system can accurately track a hand depicted in the sequence of images using the precise three-dimensional coordinates for the hand landmark positions.
This disclosure provides systems, devices, apparatus, and methods, including computer programs encoded on storage media, for a PRACH for multi-cell operation. A UE (102) receives (530), from a network entity (104), a configuration configuring: a plurality of PRACH configurations, and a plurality of SSB configurations, the plurality of PRACH configurations and the plurality of SSB configurations corresponding to different cells associated with multi-cell operation. The UE (102) transmits (540), to the network entity (104) based on the configuration and an SSB-RO mapping rule, a PRACH transmission associated with a valid RO in a cell of the different cells, the valid RO being mapped to an SSB.
A method (500) includes receiving a first request (162) to schedule a first replica pod (124R) on a first node (122) and determining that no snapshots (152) associated with a target workload (190) currently exist. Based on determining that no snapshots currently exist, the method includes initializing the target workload at the first node and receiving a signal (194) indicating that the target workload is in a ready state. The method also includes generating a snapshot of a current state of the target workload based on receiving the signal and receiving a second request (164) to schedule a second replica pod on a second node. The method also includes determining that the snapshot of the current state of the target workload currently exists based on receiving the second request and starting the target workload at the second node using the snapshot of the current state of the target workload.
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for performing tasks using a generative neural network and a cache. One of the methods includes maintaining a cache, the cache storing, for each of a plurality of cached sub-tasks (i) a cached intermediate result generated by using a generative neural network for the cached sub-task in association with (ii) an identifier for the cached sub-task; receiving a prompt a task using the generative neural network; obtaining a plan for performing one or more sub-tasks based on the prompt; obtaining an intermediate result for each sub-task, comprising determining, for each sub-task, based on the sub-task and the identifiers for the plurality of cached sub-tasks, whether to use any of the cached intermediate results to generate the intermediate result for the sub-task; and generating a result for the task based on the intermediate result obtained for each sub-task.
Methods, systems, and apparatus, including computational instructions/programs encoded on a computer-readable medium, are disclosed for implementing adaptive power grid design procedures for designing power grids of a semiconductor device. A system generates a design of a power grid in a semiconductor device at least by identifying: i) a power switch in the design of the power grid, and ii) a first metal wire within a region associated with the power switch. For each end of the first metal wire, the system determines whether the end is extendable based on an extension rule and in response to determining that the end is extendable, the system modifies the design by adding an extension to the end of the first metal wire and adding a first via to connect the extension with another metal layer.
H01L 23/528 - Layout of the interconnection structure
G06F 119/06 - Power analysis or power optimisation
H01L 23/522 - Arrangements for conducting electric current within the device in operation from one component to another including external interconnections consisting of a multilayer structure of conductive and insulating layers inseparably formed on the semiconductor body
74.
AUTOFOCUS SYSTEM AND METHOD FOR IMAGING SENSOR WITH MULTIPLE CONVERSION GAINS
Methods and devices for imaging with autofocus based on phase difference, PD, use three or more PD gain configurations. A dual conversion gain (DCG) imaging sensor, which includes a plurality of PD pixels, receives (802) light and selects (804) one of three or more DCG PD gain configurations to autofocus, AF, a lens associated with the DCG imaging sensor. The three or more DCG PD gain configurations are determined based on a single gain configuration for the plurality of PD pixels. A PD generator module associated with the DCG imaging sensor applies (806) the selected one of the three or more DCG PD gain configurations to the DCG imaging sensor.
H04N 23/67 - Focus control based on electronic image sensor signals
H04N 25/59 - Control of the dynamic range by controlling the amount of charge storable in the pixel, e.g. modification of the charge conversion ratio of the floating node capacitance
H04N 25/13 - Arrangement of colour filter arrays [CFA]Filter mosaics characterised by the spectral characteristics of the filter elements
H04N 25/778 - Pixel circuitry, e.g. memories, A/D converters, pixel amplifiers, shared circuits or shared components comprising amplifiers shared between a plurality of pixels, i.e. at least one part of the amplifier must be on the sensor array itself
75.
QUANTUM-ALGORITHM-SPECIFIC CALIBRATION AND OPTIMIZATION
Systems and methods for algorithm-specific calibration of quantum systems are provided. In one example, a method may include obtaining, by one or more computing devices, data indicative of one or more quantum computing algorithms comprising a plurality of quantum computing operations. The method may include identifying, by the one or more computing devices, a plurality of respective calibration parameters, wherein each respective calibration parameter is associated with one or more algorithm-dependent interdependencies associated with one or more respective quantum computing operations of the plurality of quantum computing operations. The method may include obtaining, by the one or more computing devices, calibration data for each of the respective calibration parameters. The method may include determining, by the one or more computing devices based at least in part on the calibration data, a plurality of respective calibration values for the plurality of respective calibration parameters.
76.
METHODS AND SYSTEMS FOR MANAGING INITIAL ACCESS AND CONTENTION BASED DATA TRANSMISSION
Devices and methods for managing initial access and data transmission in contention based random access procedure use a contention-based (CB) preconfigured grant configuration received from a radio access network (RAN) node within a system information block. A user equipment (UE) receives (504) the CB preconfigured grant configuration from the RAN node and transmits (508), to the RAN node, a CB physical uplink shared channel (PUSCH) transmission including a radio resource control (RRC) connection request message. The RAN node transmits to the UE an RRC connection setup message in response to the RRC connection request message to establish the connection between the RAN node and the UE.
H04W 72/115 - Grant-free or autonomous transmission
H04W 72/231 - Control channels or signalling for resource management in the downlink direction of a wireless link, i.e. towards a terminal the control data signalling from the layers above the physical layer, e.g. RRC or MAC-CE signalling
77.
METHODS AND SYSTEMS FOR ENABLING AND CONFIGURING CONTENTION-BASED DATA TRANSMISSION
A user equipment (UE) and a base station (BS) perform a contention-based (CB) access procedure without using a random preamble and without using a random access response. The UE receives (504) a CB preconfigured grant configuration from the BS, generates (506) an uplink (UL) packet data unit (PDU) based on the CB preconfigured grant configuration, and then transmits (508) the UL PDU to the BS based on the CB preconfigured grant configuration. The UE uses a CB radio network temporary identification (CB-RNTI) to receive a downlink assignment from the BS.
A user equipment (UE) and a base station (BS) enable a contention-based (CB) data transmissions without using a random preamble and a random access response. The UE receives (704, 703), from the BS, a CB preconfigured grant configuration and a power control parameter. The UE then determines (706) a transmission power based on the power control parameter and transmits (708) the UL PDU to the BS based on the CB preconfigured grant configuration in the cell, using the determined transmission power.
A user equipment (UE) and a base station (BS) enable a contention-based (CB) data transmissions without using a random preamble and a random access response. The UE receives (704) a CB preconfigured grant configuration and an orthogonal cover code. The UE then transmits (708) a CB transmission including uplink data on a CB physical uplink shared channel occasion and using the orthogonal cover code.
A method includes receiving a first query issued by a user and processing the first query to classify the first query as being related to a particular existing topic that corresponds to a respective one of a plurality of topic summaries stored in a topic summary datastore. Each respective topic summary of the plurality of topic summaries stored in the topic summary datastore corresponds to a different respective topic and is associated with a respective summary of past query-response interactions between a user and an assistant interface that are related to the different respective topic. The method also includes retrieving the respective topic summary from the topic summary datastore that corresponds to the particular existing topic, processing the first query conditioned on the respective topic summary retrieved from the topic summary datastore to generate a first response, and providing presentation content based on the first response for output.
42 - Scientific, technological and industrial services, research and design
Goods & Services
Providing temporary use of non-downloadable computer software for implementing a general-purpose computer programming language for use in developing, building, and managing other software.
83.
Preprocessing for Correlated Topological Quantum Error Correction
A computer-implemented method for correcting one or more errors in a quantum computing system can include obtaining, by a computing system comprising one or more computing devices, a plurality of weighted detection graphs, each of the plurality of weighted detection graphs being descriptive of a plurality of error detection measurements and having a plurality of weights, each of the weights respectively determined according to an error probability. The method can include generating, by the computing system, a plurality of reweighted detection graphs based at least in part on a correlation between physical errors in the quantum computing system. The method can include correcting, by the computing system, one or more errors in a quantum computing system based at least in part on a global decoding of the plurality of reweighted detection graphs.
A semiconductor die includes a metalized layer on an upper surface of the semiconductor die and a plurality of metal wires having a defined shape. At least one end of each of the plurality of metal wires is bonded to the metalized layer and an upper portion of each of the plurality of metal wires may extend at least partially in parallel to the metalized layer of the semiconductor die. The plurality of metal wires are arranged in a sequence such that a channel is formed by a space between the metalized layer of the semiconductor die and the upper portion of each of the metal wires that may extend at least partially in parallel to the metalized layer. The upper portion of each of the plurality of metal wires is configured to be flush with an inner surface of a cover. A cooling system including such a semiconductor die is also provided.
A method includes identifying, by a processing device of a content sharing platform, a media item associated with a channel of the content sharing platform and data related to the media item. A prompt is provided as input to an artificial intelligence (AI) model, the prompt is to cause the AI model to identify, from the data related to the media item, one or more mentions of channel memberships associated with the channel. An output is received from the artificial intelligence (AI) model and an action is performed based on the output.
H04N 21/234 - Processing of video elementary streams, e.g. splicing of video streams or manipulating encoded video stream scene graphs
H04N 21/472 - End-user interface for requesting content, additional data or servicesEnd-user interface for interacting with content, e.g. for content reservation or setting reminders, for requesting event notification or for manipulating displayed content
Techniques for improving collaboration in a video conferencing system are described herein. The system can include a projector configured to output a first user input. Additionally, the system can include a projector mirror configured to reflect the first user input outputted by the projector to a physical medium. The physical medium can include a drawing surface and be configured to display the first user input. Moreover, the system can include a first computing device having one or more that cause the first computing system to perform operations. The operations can include receiving, using an optical device, a second user input on the drawing surface. Furthermore, the operations can include generating the collaborative information by integrating the first user input and the second user input. Subsequently, the operations can include causing the projector to output the collaborative information.
To configure analytics or event monitoring in a mobile communication system, a first network function (NF) of a core network (CN) sends (1002), to a second NF, a request related to an operation involving analytics and/or event monitoring, the request indicating that the operation requires user consent; and receives (1004), from the second NF, a response to the request, the response including user consent information for one or more user equipment units (UEs).
H04L 41/16 - Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
Complexity in entropy coding a partition type for a block in image and video coding is reduced by using a cardinality of symbols that is less than a cardinality of available partition types. A bitstream modification uses the block size, and optionally the location of the block relative to the frame boundaries, to select a probability table for entropy coding a variable representing the partition type. By allowing multiple variables to represent the partition types, instead of a single variable, multiple probability tables corresponding to the variables can be used that include fewer symbols.
H04N 19/91 - Entropy coding, e.g. variable length coding [VLC] or arithmetic coding
H04N 19/119 - Adaptive subdivision aspects e.g. subdivision of a picture into rectangular or non-rectangular coding blocks
H04N 19/176 - Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
A head-worn device may be configured with a curved window-element that can generate distortion in images captured by a camera of the head-worn device. Window extrinsics describing the shape, orientation, and/or position of the curved window-element may be used as a calibration to reduce the distortion. An online calibration process may be run at times during use so that the window extrinsics can be updated to accurately represent the curved window-element after changes in the shape, orientation, and/or position of the curved window-element occur.
According to an aspect, a method includes receiving, by a head-mounted display device, an authentication code associated with multi-factor authentication, receiving image data from an image camera on the head-mounted display device, detecting, by the head-mounted display device, that the image data includes an interface for receiving the authentication code, and displaying, by the head-mounted display device, the authentication code at a location that corresponds to the interface.
An example technique for image analysis is provided. An example image analysis method includes obtaining an instructive sequence descriptive of an instructive query, an instructive response, and an instructive trace of intermediate states from the instructive query to the instructive response. The example image analysis method includes inputting, to a machine-learned model, the instructive sequence and an operative image processing query that comprises image data, wherein the machine-learned model is configured to process the operative query with attention over the instructive sequence. The example method can include generating, using the machine-learned model and responsive to the operative query, an operative image processing response that comprises an analysis of the image data.
A system for cooling a plurality of in-line memory modules includes sliding thermal interface material (“TIM”) pads and a heatsink thermally coupled to the in-line memory modules through the sliding TIM pads. The heatsink further includes a base, a plurality of thermally conductive fins, and a plurality of pedestals. The base extends in a plane. The plurality of thermally conductive fins extend in a first direction away from the base. The plurality of pedestals extend in a second direction away from the base and opposite the first direction. The sliding TIM pads are positioned between each of the plurality of pedestals and an adjacent in-line memory module. The plurality of pedestals further include a first leg and a second leg. The first and second legs are configured to move between a first position and a second position,
Systems and methods for training a restoration model can leverage training for two sub-tasks to train the restoration model to generate realistic and identity-preserved outputs. The systems and methods can balance the training of the generation task and the reconstruction task to ensure the generated outputs preserve the identity of the original subject while generating realistic outputs. The systems and methods can further leverage a feature quantization model and skip connections to improve the model output and overall training.
Systems and methods are disclosed herein for applying different visual transformations to sensitive user input based on input-device type. The described techniques distinguish between input sources and apply different, user-configurable, visual transformations based on the source of the input. In this way, user input from a hardware input device, which may provide tactile feedback, can be handled differently from input from a software input device, which may benefit from visual feedback. Such techniques can thereby improve security and usability when entering sensitive information, such as a password or an account number.
G06F 21/84 - Protecting input, output or interconnection devices output devices, e.g. displays or monitors
G06F 3/04886 - Interaction techniques based on graphical user interfaces [GUI] using specific features provided by the input device, e.g. functions controlled by the rotation of a mouse with dual sensing arrangements, or of the nature of the input device, e.g. tap gestures based on pressure sensed by a digitiser using a touch-screen or digitiser, e.g. input of commands through traced gestures by partitioning the display area of the touch-screen or the surface of the digitising tablet into independently controllable areas, e.g. virtual keyboards or menus
G06F 3/0489 - Interaction techniques based on graphical user interfaces [GUI] using specific features provided by the input device, e.g. functions controlled by the rotation of a mouse with dual sensing arrangements, or of the nature of the input device, e.g. tap gestures based on pressure sensed by a digitiser using dedicated keyboard keys or combinations thereof
Various implementations are directed towards fine-tuning a large language model (LLM) using search engine feedback (e.g., responsive content generated based on a reference source material such as a set of search engine results). Additionally or alternatively, a supervision signal can be generated based on comparing search engine conditioned LLM output with unconditioned LLM output. In many implementations, the supervision signal(s) can be used in training a reward model using reinforcement learning, where the trained reward model can be used in fine-tuning the LLM.
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating an output using a generative neural network. One of the methods include: obtaining a prompt input; determining, for each of a plurality objectives that correspond to different aspects of an output to be generated by a generative neural network, a weight to assign to the objective; generating a prefix input that defines the weight assigned to each of the plurality of objectives; and processing, using the generative neural network, the prefix input and the prompt input to generate the output.
This specification relates to methods, systems, and apparatus for multiply-accumulate (MAC) units having a unified architecture for performing both floating point MAC operations and integer MAC operations. An example method for performing a MAC operation with a MAC cell having a first unified adder and a second unified adder includes receiving floating point input operands including a first operand, a second operand, a third operand, and a fourth operand. The method further includes performing a pre-multiplication alignment process that aligns a mantissa of one or more of the floating point input operands based on comparing 1) a first sum of exponents of the first operand and the second operand, and 2) a second sum of exponents of the third operand and the fourth operand. The method further includes performing a first multiplication between aligned mantissas of the first operand and the second operand to generate a first mantissa product.
G06F 7/544 - Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation using non-contact-making devices, e.g. tube, solid state deviceMethods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation using unspecified devices for evaluating functions by calculation
G06F 7/483 - Computations with numbers represented by a non-linear combination of denominational numbers, e.g. rational numbers, logarithmic number system or floating-point numbers
98.
ARCHITECTURE AND NETWORK TOPOLOGY FOR ACOUSTIC SEGMENTATION OF SPEECH
Machine learning network topologies, and training systems and methods therefor, are described for implementing acoustic segmentation, such as for automated name detection. Such automated name detection can support automated attention handling in wearable audio components with active noise control (ANC) to suppress ambient sound. One technique for automated attention handling is based on acoustic segmentation, by which spoken audio of a class (i.e., a word) is converted into a sequence of acoustic segments representing the acoustic information of the class without speaker-specific suprasegmental features. Embodiments of network topologies for such acoustic segmentation include a Mel-frequency cepstral coefficients (MFCC) converter, a conformer-based encoder, an embedding layer, and a conformer-based decoder.
A UE (102) receives (306), from a network entity (104), a configuration configuring a first set of RSs associated with an ML prediction module. The UE receives (310) a second set of RSs different from the first set of RSs. The UE transmits (316) a prediction report based on a measurement of the second set of RSs being used in an input to the ML prediction module. A UE (102) receives (806), from a network entity (104), a configuration configuring a plurality of RSs associated with a plurality of ML prediction modules. The UE receives (808) an indication indicating an RS associated with an ML prediction module. The ML prediction module is being executed at the UE. The UE transmits (816) a prediction report output from the ML prediction module based on a measurement of the RS being used as an input to the ML prediction module.
A compound prediction block for a current block is obtained based on a weight mask, a first motion vector, and a second motion vector. The current block is partitioned into units based on a motion vector granularity. For each of the units, the first motion vector, the second motion vector, or both of the first motion vector and the second motion vector is stored in association with the each unit based on a respective portion of the weight mask that is co-extensive with each unit.
H04N 19/119 - Adaptive subdivision aspects e.g. subdivision of a picture into rectangular or non-rectangular coding blocks
H04N 19/176 - Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
H04N 19/52 - Processing of motion vectors by encoding by predictive encoding
