Abstract: Systems are provided for multilingual speech data processing. A language identification module is configured to analyze spoken utterances in an audio stream and to detect at least one language corresponding to the spoken language utterances. The language identification module detects that a first language corresponds to the first portion of the audio stream. A first transcription of the first portion of the audio stream in the first language is generated and stored in a cache. A second transcription of a second portion of the audio stream in the first language is also generated and stored. When the second portion of the audio stream corresponds to a second language, a third transcription is generated in the second language using a second speech recognition engine configured to transcribe spoken language utterances in the second language. Then, the second transcription is replaced with the third transcription in the cache and any displayed instances.

Abstract: Examples of the present disclosure describe systems and methods that provide a pipeline to generate personalized queries that are associated with and based on a user's interests determined from a user's past searches, on an Internet search engine, and/or the content the user viewed from the past searches. The suggested queries can be shown in a user interface component associated with the user interface of the search engine and before the user enters anything, such as a new Internet search. This pre-population of searches associated with a user's interests gives an opportunity to the user to try these queries without manually entering in a search string.

Abstract: A method for training one or more AI models for generating audio scores accompanying visual datasets includes obtaining training data comprising a plurality of audiovisual datasets and analyzing each of the plurality of audiovisual datasets to extract multiple visual features, textual features, and audio features. The method also includes correlating the multiple visual features and textual features with the multiple audio features via a machine learning network. Based on the correlations between the visual features, textual features, and audio features, one or more AI models are trained for composing one or more audio scores for accompanying a given dataset.

Abstract: The description relates to input devices that can be secured to and used to control various devices. One example input device can include a housing defining a top surface that includes controls. The housing can define a recess that includes generally opposing upper and lower surfaces that extends at least partially under and parallel to the top surface. A self-adjusting gripping mechanism can extend into the recess through the lower surface and can be biased toward the upper surface. The self-adjusting gripping mechanism can be configured to be contacted by devices having a range of thicknesses and configured to grip individual devices within the range of thicknesses between the self-adjusting gripping mechanism and the upper surface.

Abstract: A computing system includes an illumination light source configured to emit illumination light into an external environment and an orientation sensor configured to estimate an orientation of the illumination light source relative to the external environment. The computing system includes a logic subsystem and a storage subsystem holding instructions executable by the logic subsystem to define a light restriction zone within the external environment. Based at least in part on the orientation of the illumination light source, the illumination light source is dynamically controlled to direct the illumination light toward at least a portion of the external environment outside the light restriction zone, while mitigating emission of the illumination light into the light restriction zone.

Abstract: Technologies for digital content distribution include creating a data set that includes a plurality of entity data records each comprising a plurality of attribute values. By applying a first data security technique to the data set, a first query term and a matching subset of the plurality of entity data records are determined. A second data security technique is applied to the matching subset. When output of the second data security technique satisfies a first noisy threshold, the first query term is added to a set of query terms. A third data security technique is applied to the matching subset. When output of the third data security technique satisfies a second noisy threshold, the set of query terms is expanded to include a second query term. The set of query terms is used to perform the digital content distribution.

Abstract: Recommending one or more relevant skills to a user that are appropriate for the user's current context is provided. A machine learning contextual skills model is trained on data associated with skills that are invoked by the particular user and by various users and context data that defines characteristics of the context of the users, the users' computing devices, and the users' environments in association with the invoked skills. Responsive to a trigger event associated with the user and a computing device, current context data and user data is collected and passed to the contextual skills system. The contextual skills model is applied to the collected data and generates a ranked list of skills. One or more top-ranking skills are selected and provided to the user as one or more skill recommendations that are determined to be relevant to the user and relevant to the user's current context.

Abstract: A host that runs a security component and a container such that code running within the container cannot access data and resources running on the host except through a predetermined location (e.g., a container file system folder) that is visible from within the container. The security component helps to provide this safety by facilitating input of input data to the container, and facilitating output of output data from the container. Such containers and host security components may similarly operate on multiple hosts. An overlay network may be present over multiple of such containers to thereby allow the code within containers to communicate with each other, even if the code is running on different computing systems, so long as the code is running within the overlay network.

Abstract: The present disclosure provides a method and an apparatus for providing responses in an event-related session. The event is associated with a predefined domain, and the session comprises an electronic conversational agent and at least one participant. At least one message from the at least one participant may be detected. A set of candidate responses may be retrieved, from an index set being based on the domain, according to the at least one message. The set of candidate responses may be optimized through filtering the set of candidate responses according to predetermined criteria. A response to the at least one message may be selected from the filtered set of candidate responses. The selected response may be provided in the session.

Abstract: A method for use with a topological quantum computing device is provided. The method may include setting a plurality of device parameters for a qubit architecture including a plurality of Majorana zero modes (MZMs). The method may further include calibrating the plurality of device parameters at least in part by determining whether the plurality of MZMs exhibit ground state degeneracy. When the plurality of MZMs are determined to not exhibit ground state degeneracy, calibrating the plurality of device parameters may further include modifying one or more device parameters of the plurality of device parameters. When the plurality of MZMs are determined to exhibit ground state degeneracy, the method may further include modifying one or more parameters of a measurement device coupled to the qubit architecture.

Abstract: An FPC cable assembly is provided that includes a first ground layer, a second ground layer, and at least one signal line sandwiched by the first and second ground layers. The FPC cable assembly further includes an electromagnetic shielding structure including a first magnetic layer at least partially covering and electrically grounded to the first ground layer, a second magnetic layer at least partially covering and electrically grounded to the second ground layer, and a plurality of magnetic rings magnetically engaged with and electrically contacting the first magnetic layer and the second magnetic layer so as to surround the first and second ground layers, the at least one signal line, and the first and second magnetic layers, thereby providing electromagnetic shielding of the at least one signal line.

Abstract: A dual-screen computing device includes two separate displays that are coupled to an interconnecting hinge. A hinge detector detects movement or position of the hinge, and the positions of the displays may be determined based on the hinge movement or position. The positions of the displays relative to each other may then be used to determine which mode of operation the dual-screen computing device is operating (e.g., tent mode, open, closed, etc.). Additionally, the dual-screen computing device may include various sensors that detect different environmental, orientation, location, and device-specific information. Applications are configured to operate differently based on the mode of operation and, optionally, the sensor data detected by the sensors.

Abstract: Techniques for adapting previously-annotated training examples into updated training examples for training a machine learning model are disclosed. One example includes a computer program that recognizes a find expression, a replacement expression, and a filtering constraint in which the filtering constraint distinguishes a subset of previously-annotated training examples from others of the previously-annotated training examples. An instance of the find expression is identified by the computer program within the subset of the previously-annotated training examples that were identified among the previously-annotated training examples based on the filtering constraint. The instance of the find expression identified within the subset of the previously-annotated training examples is replaced by the computer program with an instance of the replacement expression to obtain an updated subset of training examples.

Abstract: Some embodiments transfer a software development working context between development environments. The environments may be on the same machine, or on different machines. The working context includes ephemeral state information (ESI), such as edits not yet committed to a repository, development tool user interface state, command history, build history, debug history, execution history, tool window content, hot exit files, or search filters, among others. The ESI is computationally gathered from RAM and other locations in a source environment, written to an edit session artifact (ESA), and then read from the ESA and installed at a target environment. The transfer is accomplished without committing source code to any source code version control system, e.g., repository. The transfer may be tailored for efficiency or efficacy by determining whether a particular software development capability is present in one or both of the environments.

Abstract: Techniques for determining a user's IPD are described. A first stimulus is displayed on a first display, and a second stimulus is displayed on a second display. A stimulus separation distance is a distance that exists between the first and second stimuli. The stimulus separation distance is progressively increased by progressively moving, in opposing directions relative to one another, the first and second stimuli. While that distance is being progressively increased, at least one of the user's eyes is tracked. While the distance is being progressively increased, a change in a rate of eye movement for the user's eye is detected. When the change is detected, a value for the stimulus separation distance is recorded. The recorded value is set as a baseline for the user's IPD.

Abstract: A computer-implemented technique is described herein for assisting a user in advancing a task objective. The technique uses a suggestion-generating system (SGS) to provide one or more suggestions to a user in response to at least a last-submitted query provided by the user. The SGS may correspond to a classification-type or generative-type neural network. The SGS uses a machine-trained model that is trained using a multi-task training framework based on plural groups of training examples, which, in turn, are produced using different respective example-generating methods. One such example-generating method constructs a training example from queries in a search session. It operates by identifying the task-related intent the queries, and then identifying at least one sequence of queries in the search session that exhibits a coherent task-related intent. A training example is constructed based on queries in such a sequence.

Abstract: Methods, systems, apparatuses, and computer-readable storage mediums described herein enable executable code of a hardware security platform (HSP) circuit to communicate with a hypervisor in a separate processor. The hypervisor generates and manages virtual machines. The HSP code comprises trusted platform module (TPM) logic, that processes TPM commands received via the hypervisor, and in response to the processing, communicates security information (e.g., measurements, keys, authorization data) with the virtual machines via the hypervisor. The TPM logic receives security information related to a virtual machine from the hypervisor and stores the security information in non-volatile memory of the HSP circuit, where security information from a particular VM is distinguishable from security information from another VM in the HSP memory.

Abstract: A tinting material may be generated and applied to the backdrop of an application presented on a user interface. The user interface is presented on a display and includes a background. The background comprises a first color, which includes a luminosity value, a hue value, and a saturation value. A tint color, having a luminosity value, a hue value, a saturation value, and an opacity value, is received. The luminosity value of the first color is modified to generate a second color, and the hue value and saturation value of the generated second color are modified to generate a third color. A tinting material color is generated that includes the modified luminosity value of the second color, the modified hue value of the third color, and the modified saturation value of the third color, and an application is presented on the user interface that includes the tinting material color.

Abstract: A method of memory deallocation across a trust boundary between a first software component and a second software component is described. Some memory is shared between the first and second software components. An in-memory message passing facility is implemented using the shared memory. The first software component is used to deallocate memory from the shared memory which has been allocated by the second software component. The deallocation is done by: taking at least one allocation to be freed from the message passing facility; and freeing the at least one allocation using a local deallocation mechanism while validating that memory access to memory owned by data structures related to memory allocation within the shared memory are within the shared memory.

Abstract: Force sensing touchpads and related computing devices are disclosed. In one example, a force sensing touchpad comprises a printed circuit board (PCB) comprising a plurality of PCB electrodes. A base plate is spaced from the PCB and defines an aperture. A spring is mounted on the base plate overlying the aperture in the base plate. The spring comprises a plurality of spring electrodes, a first beam extending from a first internal wall to a first side of a central portion that overlies the aperture, and a second beam extending from a second internal wall that is opposite the first internal wall to a second side of the central portion that is opposite to the first side. A spacer located between the PCB and the central portion of the spring. Each of the PCB electrodes is spaced from and overlies one of the spring electrodes of the spring.

Abstract: Techniques are described for providing real-time audio and/or video feedback during a conference call. Audio feedback can be provided during a conference call in response to a participant unmuting the participant's microphone. For example, a feedback period can be initiated upon receiving an unmute indication. During the feedback period, the participant's microphone audio can be sent, in real-time, back to the participant for playback in addition to sending to the other participants. After the feedback period is over, the participant can be removed from receiving their microphone audio. Video feedback can be provided during a conference call in response to screen sharing. For example, during a feedback period, the participant can receive, in real-time, a screen content thumbnail of their shared screen content. After the feedback period is over, the screen content thumbnail can stop being sent.

Abstract: Innovations in encoding or decoding when switching color spaces are presented. For example, some of the innovations relate to signaling of control information for adaptive color space transformation (“ACT”). Other innovations relate to ACT operations. These innovations can improve coding efficiency when switching between color spaces during encoding and decoding.

Abstract: A method for allocating a plurality of network resources to a plurality of network-access demands of a plurality of network guests comprises (a) receiving the plurality of network-access demands; (b) for each of the plurality of network-access demands (i) dynamically computing, from among the plurality of network resources, a resorted order of resources associated with the network-access demand, and (ii) for each network resource associated with the network-access demand, increasing, in the re-sorted order, an allocation of the network resource to the network-access demand until the network-access demand is saturated, and freezing the allocation of each of the plurality of network resources to the saturated demand; and (c) outputting the frozen allocation of each of the plurality of network resources for each of the plurality of network-access demands.

Abstract: The present disclosure concerns methods and systems for registering dynamically created packaged applications with an operating system. A dynamically created packaged application may not include a way to authenticate the packaged application but may declare a dependency on a host package. The host package may include a verified signature and be registered with the operating system. The host package may include a host runtime that references an executable included in the host package. The packaged application may not include executable files but may reference the host runtime. The operating system may register the packaged application based in part on its dependency on the host package. When the packaged application is activated, the packaged application may invoke the host runtime and the executable but do so with an application identity of the packaged application. The application identity may be different from a host runtime identity of the host runtime.

Abstract: A method of operation of a display device for controlling brightness adaptation comprises (a) receiving a position signal that varies in dependence on a position of a fovea of an eye; (b) receiving a brightness signal that varies in dependence on brightness external to the display device; and (c) projecting a color image into the eye over a range of angles that varies in dependence on the position signal and on the brightness signal, such that when the brightness is below a predetermined threshold the color image is confined to the retinal target area.

Abstract: This document relates to employing physiological feedback to aid a user in targeting objects using a computing device. One example relates to a method or technique that can include receiving a user attention signal conveying where a user directs attention. The method or technique can also include, based on the user attention signal, identifying a predicted object that the user intends to target with a targeting mechanism and outputting a visual identification of the predicted object. The method or technique can also include receiving a user reaction signal conveying a physiological reaction of the user to the visual identification of the predicted object. The method or technique can also include, in an instance when the physiological reaction of the user indicates an error, identifying another predicted object that the user intends to target with the targeting mechanism.

Abstract: The discussion relates to thermal management. One example can include a circuit board including inner, intermediate, and outer generally concentric zones and a cryogenically cooled chip located in the inner zone as well as non-cryogenic electronic components positioned in the outer zone. In this example, the intermediate zone can have a skeletonized configuration that slows thermal energy movement from the outer zone to the inner zone.

Abstract: A technique leverages a machine-trained pattern-completion engine to successively extract items-of-interest from unstructured data, categorize the items-of-interest, and identify relations in the unstructured data. The technique then generates a structured database based on the information it has identified. In some cases, the items-of-interest represent facts expressed by the unstructured data. The technique also leverages the structured database to perform various application tasks. In one approach, in the course of answering a query, the technique extracts supplemental information from the structured database. The technique then feeds the query and the supplemental information to the pattern-completion engine, and, in response thereto, receives output information that addresses the query. In some cases, the query is part of lengthy prompt information.

Abstract: A system for cropping an image is disclosed, which performs receiving a source image and user intention data; determining a target feature based on the user intention data; identifying a plurality of visual features within the source image; determining a contextual relevance between the target feature and each identified visual feature of the source image; identifying, based on the determined contextual relevance between the target feature and each identified visual feature of the source image, one or more cropping candidate portions within the source image; cropping, based on the one or more cropping candidate portions, the source image to generate a plurality of cropped images; and causing the plurality of cropped images to be displayed on a display.

Abstract: A method of fabricating a semiconductor-superconductor hybrid device comprises providing a workpiece comprising a semiconductor component, a layer of a first superconductor material on the semiconductor component, and a layer of a second superconductor material on the first superconductor material, the second superconductor material being different from the first superconductor material; etching the layer of the second superconductor material to expose a portion of the first superconductor material; and oxidising the portion of the first superconductor material to form a passivating layer on the semiconductor. The first superconductor provides energy coupling between the semiconductor and the second superconductor, and the passivating layer protects the semiconductor while allowing electrostatic access thereto. Also provided are a hybrid device, and a method of etching.

Abstract: Lightweight mechanisms provide a way to assert provenance when live streaming media content and establish provenance upon playback. For example, a provenance claim generator generates a key pair including a live-stream private key and live-stream public key. The claim generator signs, with a long-term private key reliably associated with a sender, manifest metadata including the live-stream public key, thereby producing a manifest signature. During live streaming, the claim generator signs respective portions of media content with the live-stream private key, producing portion signatures for the respective portions. A provenance claim validator receives the manifest signature and manifest metadata. The claim validator verifies the manifest metadata using a long-term public key (reliably associated with the sender) and the manifest signature.

Abstract: The disclosure provides for, at a host computing device, a method of interacting with a target computing device. The method includes receiving one or more sonic waveforms from the target computing device at a microphone subsystem of the host computing device. The host computing device determines an identity of the target computing device via processing of the one or more sonic waveforms. In response to said identity determination, the method includes (1) initializing a wireless data link for communication between the host computing device and the target computing device, and (2) entering a mode of host-target interactivity in which the host computing device and the target computing device interact via the wireless data link.

Abstract: A technique performs the task of knowledge-graph completion in a manner that is both scalable and resource efficient. In some implementations, the technique identifies a source entity having a source-target relation that connects the source entity to a yet-to-be-determined target entity. The technique also identifies a source-entity data item that provides a passage of source-entity text pertaining to the source entity. The technique uses a machine-trained encoder model to map the source-entity data item to source-entity encoded information. The technique then predicts an identity of the target entity based on the source-entity encoded information, and based on predicate encoded information that encodes the source-target relation. In some implementations, the technique also predicts the target entity based on a consideration of one or more neighboring entities that are connected to the source entity and their respective source-to-neighbor relations.

Abstract: A system and method for providing a collaboration template including a brainstorming canvas to a display device of each of a plurality of participants coupled to the system, wherein the template includes a selection element configured to activate an artificial intelligence (AI) chat interface to receive a natural language command from at least one of the participants. The natural language command is received from the participant, combined with context prompts generated by a context prompt generator system to form a combined AI request, and transmitted to an AI system. In response to the natural language commands and context prompts transmitted to the AI system, a response is received from the AI system and instructions are provided to the client device of the participant to display the AI response on the brainstorming canvas of the template.

Abstract: The present disclosure concerns methods and systems for registering dynamically created packaged applications with an operating system. A dynamically created packaged application may not include a way to authenticate the packaged application but may declare a dependency on a host package. The host package may include a verified signature and be registered with the operating system. The host package may include a host runtime that references an executable included in the host package. The packaged application may not include executable files but may reference the host runtime. The operating system may register the packaged application based in part on its dependency on the host package. When the packaged application is activated, the packaged application may invoke the host runtime and the executable but do so with an application identity of the packaged application. The application identity may be different from a host runtime identity of the host runtime.

Abstract: A method for scheduling a coordinated transfer of data among a plurality of processor nodes on a network comprises operating a multi-commodity flow model subject to a plurality of predetermined constraints. The model is configured to (a) receive as input a set of demands defining, for each of the plurality of processor nodes, an amount of data to be transferred to that processor node, (b) assign a plurality of paths linking the plurality of processor nodes, and (c) emit a schedule for transfer of the data along the plurality of paths so as to minimize a predetermined cost function, wherein the schedule comprises at least one store-and-forward operation and at least one copy operation.

Abstract: Examples are provided for a method of providing access to data of a data center. In one aspect, the method comprises storing a unit of data to each of a plurality of data nodes of a data center, designating a first node of the data center as a primary access node for the unit of data, the primary access node being configured to service access requests to the unit of data using one or more of the plurality of data nodes, determining that the first node is not available, and performing a failover process by reconfiguring a second node of the data center as the primary access node for the unit of data.

Abstract: A data processing system includes a processor; and a memory comprising programming instructions for execution by the processor to manage data files stored using both local and cloud data storage. The instructions, when executed by the processor, cause the processor to obtain a handle with an oplock on a selected file; set a timer; and dehydrate the selected file when the timer elapses provided that the oplock remains unbroken when the file is dehydrated.

Abstract: Innovations in use of chroma quantization parameter (“QP”) offsets when determining a control parameter for deblock filtering. For example, as part of encoding, an encoder sets a picture-level chroma QP offset and slice-level chroma QP offset for encoding of a slice of a picture. The encoder also performs deblock filtering of at least part of the slice, where derivation of a control parameter considers only the picture-level chroma QP offset. The encoder outputs at least part of a bitstream including the encoded content. As part of decoding, a corresponding decoder sets a picture-level chroma QP offset and a slice-level chroma QP offset for decoding of a slice of a picture, but derivation of a control parameter for deblock filtering considers only the picture-level chroma QP offset.

Abstract: Examples of the present disclosure describe systems and methods for multichannel audio speech classification. In examples, an audio signal comprising multiple audio channels is received at a processing device. Each of the audio channels in the audio signal is transcoded to a predefined audio format. For each of the transcoded audio channels, an average power value is calculated for one or more data windows in the audio signal. A correlation value is calculated between the average power value for each audio channel and the combined average power value of the other audio channels in the audio signal. Each of the correlation values (or an aggregated correlation value for the audio channels) is then compared against a threshold value to determine whether the audio signal is to be classified as a speech-based communication. Based on the classification, an action associated with the audio signal may be performed.

Abstract: A data processing system includes: a processor; a network interface; and a memory comprising programming instructions for execution by the processor to: access an electronic workspace; determine a list of changes to the workspace within a period of time; structure a query to a Large Language Model (LLM) or Artificial Intelligence (AI) engine, the query to generate for a user a summary of the changes; obtain a corresponding summary from the LLM or AI engine; and in a user interface, associate a display of the workspace with the corresponding summary to update a user as to a current state of the workspace.

Abstract: Examples are disclosed related to analog to digital conversions. One example provides a conversion circuit comprising a first analog-to-digital converter (ADC) configured to convert an analog voltage to generate a first subset of digital output bits from a most significant bit (MSB) to a bit k and a second subset of digital output bits from a bit k?1 to a least significant bit (LSB) of the first ADC. The bit k is between the MSB and the LSB. The first ADC comprises a residual output configured to output a residual voltage of the analog voltage after converting the bit k. The conversion circuit further comprises an amplifier stage connected to the residual output and configured to generate an amplified residual voltage. The conversion circuit further comprises a second ADC connected to the amplifier stage and configured to convert the amplified residual voltage to generate extra digital output bits.

Abstract: Examples are disclosed that relate to a keyboard for a touch-sensitive display device. One example provides a touch-sensitive display device comprising a touch sensor, a display, a wireless receiver, a logic subsystem, and a storage subsystem. The storage subsystem comprises instructions executable by the logic subsystem to, in a first mode in which a keyboard satisfies a detection condition of the touch sensor, detect, via the touch sensor, input applied to the keyboard, and, in a second mode in which the keyboard does not satisfy the detection condition of the touch sensor, receive input from the keyboard via the wireless receiver, the input being detected at least in part by the keyboard operating in the second mode.

Abstract: Systems and methods are directed to generating a synoptic video from one or more documents. The system receives an indication to generate the synoptic video and in response, accesses the one or more documents and user preferences of a user that are derived by the network system. The one or more documents are analyzed to determine relevant text content, whereby the analyzing includes tagging the relevant text content. Based on the user preferences, the system generates the synoptic video that summarizes the tagged relevant text content. The content and style of the synoptic video is based on a user style indicated by the user preferences. For example, the user preference can determine a depth of the summarized content, a type of audio or text used, an amount of images to provide, and an amount of summarized text to provide. The synoptic video can then be displayed to the user.

Abstract: Examples are disclosed that relate to providing image data to a user in a defined space of a surrounding environment from a perspective of the user. One example provides a computing system, comprising a logic subsystem, and a storage subsystem comprising instructions executable by the logic subsystem to obtain information regarding a pose of a user within a defined space, based upon the pose of the user, determine a portion of an environment surrounding the defined space toward which the user is looking, obtain image data representing the portion of the environment from a perspective of the user, and provide the image data for display via a display device within the defined space.

Abstract: A method, system, and storage device storing a computer program, for generating questions based on provided content, such as, for example, a document having words. The method comprises automatically estimating the probability of interesting phrases in the provided content, and generating a question in natural language based on the estimating. In one example embodiment herein, the estimating includes predicting the interesting phrases as answers, and the estimating is performed by a neural model. The method further comprises conditioning a question generation model based on the interesting phrases predicted in the predicting, the question generation model generating the question. The method also can include training the neural model. In one example, the method further comprises identifying start and end locations of the phrases in the provided content, and the identifying includes performing a dot product attention mechanism parameterizing a probability distribution.

Abstract: Some embodiments process log data into traffic patterns, using call graph normalization and aggregation. Traffic patterns are characterized, by comparing traffic patterns, finding redundancies, locating coverage gaps, delimiting impact regions, spotting optimization candidates, performing timing analyses, predicting traffic flow, or other operations. Characterizations are prioritized, and priorities help guide operational changes in a computing system. Gaps between the traffic patterns of different environments or different scenarios may be located, leading to changes in signoff coverage or testing coverage or both. Deployment traffic anomalies may trigger rollback. Traffic patterns indicative of cyberattacks may trigger security countermeasures.

Abstract: Machine learning multiple features of an item depicted in images. Upon accessing multiple images that depict the item, a neural network is used to machine train on the plurality of images to generate embedding vectors for each of multiple features of the item. For each of multiple features of the item depicted in the images, in each iteration of the machine learning, the embedding vector is converted into a probability vector that represents probabilities that the feature has respective values. That probability vector is then compared with a value vector representing the actual value of that feature in the depicted item, and an error between the two vectors is determined. That error is used to adjust parameters of the neural network used to generate the embedding vector, allowing for the next iteration in the generation of the embedding vectors. These iterative changes continue thereby training the neural network.

Abstract: Innovations in the use of base color index map (“BCIM”) mode dining encoding and/or decoding simplify implementation by reducing the number of modifications made to support BCIM mode and/or improve coding efficiency of BCIM mode. For example, some of the innovations involve reuse of a syntax structure that is adapted for transform coefficients to instead signal data for elements of an index map in BCIM mode. Other innovations relate to mapping of index values in BCIM mode or prediction of elements of an index map in BCIM mode. Still other innovations relate to handling of exception values in BCIM mode.

Abstract: A system includes reception of a database query, determination of result set output columns associated with the database query, and determination, for each of the determined result set output columns, of one or more data sources associated with the result set output column. Sensitivity information is determined for each of the one or more data sources based on metadata, and result set sensitivity information is determined based on the determined sensitivity information. A result set is determined based on the database query, and the result set and the result set sensitivity information are transmitted.