Abstract: Technology is provided for a memory drive storage tray. The memory drive storage tray includes a frame that defines one or more drive bays, each configured to contain a memory drive. Each drive bay includes an associated drive connector for interfacing the memory drive with the memory drive storage tray. Each drive bay also includes a first retainer tab positioned on the frame adjacent the drive connector and a second retainer tab, longer than the first retainer tab, positioned on the frame opposite the first retainer tab. A resilient member, such as a leaf spring, is centrally positioned in the drive bay opposite the drive connector to urge the memory drive toward the drive connector. The resilient member holds the memory drive in place against the drive connector and under the first and second retainer tabs.

Abstract: In one embodiment, a method includes accessing a first set of entities, with which a user has interacted, and a second set of entities in a social-networking system. A first set of vector representations of the first set of entities are determined using a deep-learning model. A target entity is selected from the first set of entities, and the vector representation of the target entity is removed from the first set. The remaining vector representations in the first set are combined to determine a vector representation of the user. A second set of vector representations of the second set of entities are determined using the deep-learning model. Similarity scores are computed between the user and each of the target entity and the entities in the second set of entities. Vector representations of entities in the second set of entities are updated based on the similarity scores using the deep-learning model.

Abstract: In particular embodiments, a social networking system queries a social graph of the social-networking system for social content associated with video content provided to one or more users of the social-networking system and accesses privacy settings associated with each of the one or more users. The privacy settings indicate whether a particular user has authorized the social-networking system to share social content associated with the particular user with one or more third parties. The social networking system provides the social content associated with video content provided to the one or more users to a particular third party if the privacy settings of the one or more users indicate that the one or more users have authorized the social-networking system to share the social content with one or more third parties.

Abstract: Techniques for social networking systems and methods for testing and applying user interfaces are disclosed herein. The method includes steps of presenting a user interface including a new user interface feature to a group of test users, collecting response data from the test users experiencing the user interface, performing analytics on the response data, and determining at least one interface rule of applying user interface features for a user depending on one or more user attributes of the user based on the analytics using a machine learning model.

Abstract: Embodiments relate to transferring dies or other electronic components from a carrier substrate to a target substrate of a device as part of chip assembly for the device. Bonding material is applied to selected dies on a carrier substrate by direct boning, or to corresponding die transfer locations on a target substrate. The carrier substrate is then brought into contact with the target substrate to transfer each of the selected dies to the carrier substrate. Dies can also be directly bonded to the target substrate even in the presence of other die in situ (e.g., from a previous bonding cycle), hence, enables more than one direct bond cycle to be carried out for a target substrate. As such, multi-color (RGB) display elements can be assembled in stages (e.g., separate bonding cycles) in a flexible manner to provide redundancy or to replace inoperative LED dies.

Abstract: The compatibility score of members of a social network is based on the compatibility of their expressed interests. The compatibility score between any two interests is calculated as the log of the estimated probability that a member of the social network will express both interests divided by the product of: (i) the estimated probability that a member of the social network will express the first of the two interests and (ii) the estimated probability that a member of the social network will express the second of the two interests. The compatibility score between two individuals is calculated as the sum of the compatibility scores between each interest appearing in a set of interests expressed by the first of the two individuals and each interest appearing in a set of interests expressed by the second of the two individuals.

Abstract: In one embodiment, a method includes a computer machine reviewing a picture image and an associated video with the associated video including several video frames. The computer machine then statistically analyzes the corresponding areas for each video frame. The computer machine then compares an area in the picture image with an associated area in a video frame image, wherein the video frame image is based in part on the statistical analysis of the corresponding area for each video frame. Next, the computer machine creates an area for a composite image based in part on the comparison of the area in the picture image with the associated area in the video frame image.

Abstract: A system for training a model to predict a sequence (e.g. a sequence of words) given a context is disclosed. A model can be trained to make these predictions using a combination of individual predictions compared to base truth and sequences of predictions based on previous predictions, where the resulting sequence is compared to the base truth sequence. In particular, the model can initially use the individual predictions to train the model. The model can then be further trained over the training data in multiple iterations, where each iteration includes two processes for each training element. In the first process, an initial part of the sequence is predicted, and the model and model parameters are updated after each prediction. In the second process, the entire remaining amount of the sequence is predicted and compared to the corresponding training sequence to adjust model parameters to encourage or discourage each prediction.

Abstract: Techniques for estimating user attention on a website or application are provided. First activity data for a first user of a website or an application may be identified. The first activity data may indicate activities of the first user on the website or the application. A first predetermined period of inactivity may be detected in the first activity data. A response triggering event may be initiated after the first predetermined period of inactivity. An indication of a user response to the response triggering event may be monitored for. Whether the first user is passively present on the website or the application may be determined based on the monitoring for the indication of the user response.

Abstract: A head-mounted display (HMD) includes a multifocal block having one or more possible focal distances and includes a multifocal structure. The multifocal structure has a first focal distance and a second focal distance of the one or more possible focal distances. The multifocal structure includes one or more optical components positioned in series such that light from an electronic display is received and passes through each of the one or more optical components at least once before being output from the multifocal structure. The one or more optical components includes a switchable half waveplate (SHWP). The SHWP has a first state that causes the multifocal structure to output image light at the first focal distance, and a second state that causes the multifocal structure to output the image light at the first focal distance.

Abstract: Embodiments are disclosed for transliterating text entries across different script systems. A method according to some embodiments includes steps of: receiving an input string in a first script system input using a keyboard; segmenting, using a probabilistic model, the input string into phonemes that correspond to characters or sets of characters in a second script system; converting the phonemes in the first script system into the characters or sets of characters in the second script system, the characters or sets of characters forming a word or a word prefix in the second script system; and outputting the word or the word prefix in the second script system.

Abstract: In one embodiment, a method includes a computing system accessing a first training data comprising a first image and a second image and an associated optical flow estimation. The system may input (1) the first image into a first machine-learning model configured to generate a first output and (2) the optical flow estimation into a second machine-learning model configured to generate a second output. The first output of the first machine-learning model is associated with first image segments of a predetermined number, and the second output of the second machine-learning model is associated with transformations of the predetermined number. The first output, the transformations, and the first image are configured to generate an estimated image. The system trains the first machine-learning model and the second machine-learning model based on at least a comparison of the estimated image and the second image.

Abstract: Technology is disclosed for inferring human attributes from images of people. The attributes can include, for example, gender, age, hair, and/or clothing. The technology uses part-based models, e.g., Poselets, to locate multiple normalized part patches from an image. The normalized part patches are provided into trained convolutional neural networks to generate feature data. Each convolution neural network applies multiple stages of convolution operations to one part patch to generate a set of fully connected feature data. The feature data for all part patches are concatenated and then provided into multiple trained classifiers (e.g., linear support vector machines) to predict attributes of the image.

Abstract: An online system maintains information describing interactions by its users with various applications. To allow evaluation of an application against other applications, the online system identifies additional applications having a threshold measure of similarity to the application and with which at least at threshold number of users interacted during a time interval. Based on a number of users who interacted with various additional applications and amounts of revenue obtained by additional applications, the online system selects a group of additional applications. The online system selects additional applications from the group based on scores for the additional applications determined from user interaction and revenue obtained by the additional applications and provides information about the additional applications selected from the group to an entity associated with the application.

Abstract: An online system receives information describing a target group of online system users from a third party system and stores the information describing the target group. The online system subsequently uses the target group to select content for presentation to one or more users. For example, users included in the target group are identified as eligible to be presented with content items. Based on revenue obtained by the online system from presenting content based on the target group, the online system determines a monetization value for the target group. The online system determines whether to continue storing the information describing the target group based on the monetization value.

Abstract: In one embodiment, a method includes receiving, from a client system of a first user, a text query comprising a character string of n characters, generating a plurality of suggested queries based on the text query, where each suggested query comprises the character string of the text query and one or more query tokens, and where at least one of the query tokens of each suggested query partially matches the character string of the text query, and sending, to the client system responsive to receiving the nth character of the text query, instructions for displaying one or more suggested queries to the first user, where at least one of the suggested queries is displayed with a preview of one or more search results matching the suggested query.

Abstract: Systems, methods, and non-transitory computer readable media can obtain a first event stream including one or more events of a first type, where each event of the first type is associated with a timestamp. A second event stream including one or more events of a second type can be obtained, where each event of the second type is associated with a timestamp. The first event stream and the second event stream can be merged to generate information associated with a metric relating to a system, based on the timestamps associated with the one or more events of the first type and the timestamps associated with the one or more events of the second type.

Abstract: A virtual reality headset includes multiple illumination sources emitting light towards a user's eye and an image capture device capturing light reflected by the user's eye. The image capture device captures images of light from the illumination sources reflected by the user's corneas when the user looks at a specific location in the virtual reality headset. Based on locations of light having at least a threshold intensity in the captured images, the position of the center of user's eye's pupil is determined in three dimensions and used to determine a distance between the center of user's eye's pupil and a reference point relative to the illumination sources. Distances between centers of pupils of the user's eyes and reference points are used to determine a distance between the centers of the pupils of the user's eyes and a distance from the corneas to an optical system of the headset.

Abstract: An online system selects different geographic regions to evaluate content presented by the online system. The online system pairs a geographic region with another geographic region based on similarity of the geographic regions. To assist this pairing, the online system identifies users associated with locations within a threshold distance of different geographic regions and retrieves demographic and other characteristics of users associated with locations within the threshold distance of one or more geographic regions. Based on the retrieved characteristics and characteristics of the geographic regions, the online system generates vectors for different geographic regions. For each possible pairing of geographic regions, the online system computes a distance value and selects a pairing of geographic regions based on the distance values. For example, the online system selects a pair of geographic regions associated with a minimum distance value.

Abstract: A head-mounted display (HMD) includes an electronic display element and an optics block. The electronic display element includes a plurality of display panels that together output image light. The plurality of panels including a first display panel and a second display panel. The first display panel includes a first plurality of sub-pixels that are separated from each other by a non-emission area. The second panel includes a second plurality of sub-pixels. The second display panel is positioned offset from the first display panel such that the second plurality of sub-pixels emit light through the non-emission area of the first display panel. The optics block configured to direct the image light to an exit pupil of the HMD corresponding to a location of an eye of a user of the HMD.

Abstract: A ?LED device comprising: a substrate and an epitaxial layer grown on the substrate and comprising a semiconductor material, wherein at least a portion of the substrate and the epitaxial layer define a mesa; an active layer within the mesa and configured, on application of an electrical current, to generate light for emission through a light emitting surface of the substrate opposite the mesa, wherein the crystal lattice structure of the substrate and the epitaxial layer is arranged such that a c-plane of the crystal lattice structure is misaligned with respect to the light emitting surface.

Abstract: In one embodiment, a server computer may receive, from a client device, a request to download an application, wherein the client device is logged-in to a session associated with a user account in a communications system. In response to the request, a downloader module executable file may be appended with the login information and the authentication information. The server computer may transmit the downloader module executable file to the client device. In response to a second request (from the downloader module executable file), the server computer may transmit the installer file to the client device.

Abstract: In one embodiment, a method includes receiving, from a client system of a first user, a query comprising one or more n-grams, determining one or more search intents of the query based at least on whether one or more of the n-grams in the query match terms corresponding to a search intent indexed in a pattern-detection model, generating one or more search results based on the query, each search result corresponding to an object of a plurality of objects, and scoring the search results based on one or more of the search intents.

Abstract: In one embodiment, a sequence of input words is received. Each of the input words is encoded as an indicator vector, wherein a sequence of the indicator vectors captures features of the sequence of input words. The sequence of the indicator vectors is then mapped to a distribution of a contextual probability of a first output word in a sequence of output words. For each subsequent output word, the sequence of the indicator vectors is encoded with a context, wherein the context comprises a previously mapped contextual probability distribution of a fixed window of previous output words; and the encoded sequence of the indicator vectors and the context is mapped to the distribution of the contextual probability of the subsequent output word. Finally, a condensed summary is generated using a decoder by maximizing the contextual probability of each of the output words.

Abstract: In one embodiment, a method includes identifying a shared visual concept in visual-media items based on shared visual features in images of the visual-media items; extracting, for each of the visual-media items, n-grams from communications associated with the visual-media item; generating, in a d-dimensional space, an embedding for each of the visual-media items at a location based on the visual concepts included in the visual-media item; generating, in the d-dimensional space, an embedding for each of the extracted n-grams at a location based on a frequency of occurrence of the n-gram in the communications associated with the visual-media items; and associating, with the shared visual concept, the extracted n-grams that have embeddings within a threshold area of the embeddings for the identified visual-media items.

Abstract: Systems, methods, and non-transitory computer-readable media can determine a first faceplate installed in the computing device, the first faceplate being an attachment that is capable of being physically installed in the computing device. A faceplate profile associated with the first faceplate can be determined. At least one setting of the computing device or operation performed by the computing device can be modified based at least in part on information specified by the faceplate profile associated with the first faceplate.

Abstract: Methods for quantifying pupil swim are disclosed in order to compensate for the same. A target image, in one embodiment, is displayed on a display of a head mounted display (HMD). Images of the target image are captured from a plurality of positions relative to an optical axis of an optics block of the HMD at an exit pupil of the HMD. The target image includes features and differences between observed locations of the features and their expected locations absent the optics block are determined. From these differences, a wavefront of the optics block is reconstructed and distortion corrections for the optics block are generated using the wavefront. The distortion corrections, when applied to a virtual scene, add pre-distortion that is canceled by the optical imperfections of the optics block as light of the virtual scene with the pre-distortion passes through the optics block.

Abstract: In one embodiment, a method includes receiving data that includes an association between a nickname and a primary name, determining whether the association meets a nickname association threshold, and generating at least one expanded index entry that associates the nickname with the primary name when the association meets the nickname association threshold. The data may include a log entry indicating that a search for the nickname that returned the primary name as a result was followed by a user selection of the primary name, and the method may further include incrementing a counter value associated with the nickname and the primary name. The data may include a first log entry indicating that a search for the nickname returned a result the user did not select, followed by a second log entry indicating that a search for the primary name returned the primary name, and a user selected the primary name.

Abstract: Online systems, for example, social networking systems evaluate expressions based on features describing relations between entities represented in the online system. These expressions are represented using an expression language. The expression language allows features to be specified as functions of attributes from user accounts. The expressions support use of variables to represent computations, for example, sub-expressions. The expressions are dynamic, since expressions can be specified and executed at call time. The same set of expressions is used many times, e.g., to compute the same function for multiple feature sets, for example, user accounts. Expressions are preferably represented using postfix representation. However some expressions, for example, expressions using variables are represented as trees. To optimize the expressions at runtime, the expressions are cached using a representation determined to be efficient for executing the expression.

Abstract: A head mounted display (HMD) includes a display and a pancake lens block. The display with a circular polarizer, comprised of an initial linear polarizer and a first quarter-waveplate with polarizer transmission axis 45° from the waveplate fast axis, emits polarized light. The pancake lens includes a partial reflector, a second quarter-waveplate, and a beam splitting polarizer. The pancake lens receives polarized light from the display. Light propagating through the pancake lens undergoes multiple reflections and transmissions achieved by coordinating changes in polarization of light through these optical elements. To mitigate parasitic light from degrading image quality of the HMD, the fast axis orientation of the first quarter-waveplate is oriented 90° relative to the fast axis orientation of the second quarter-waveplate, and thus the transmission axis of the first polarizer is oriented 90° relative to the transmission axis of the beam splitting polarizer.

Abstract: Embodiments are disclosed for transliteration decoding using a tree structure. A method according to some embodiments includes steps of: generating a tree structure for an input string in a first script system, the tree structure including nodes representing segments of the input string; identifying segmentation candidates for the input string based on paths of the tree structure, the segmentation candidates segmenting the input string into character groups; selecting a segmentation candidate based on probabilities of the segmentation candidates predicted by a probabilistic model; segmenting the input string into character groups that correspond to characters in a second script system; decoding the character groups in the first script system into the characters in the second script system, the characters forming a word or a word prefix in the second script system; and outputting the word or the word prefix in the second script system.

Abstract: Some embodiments include a method of machine learner workflow processing. For example, a workflow execution engine can receive an interdependency graph of operator instances for a workflow run. The operator instances can be associated with one or more operator types. The workflow execution engine can assign one or more computing environments from a candidate pool to execute the operator instances based on the interdependency graph. The workflow execution engine can generate a schedule plan of one or more execution requests associated with the operator instances. The workflow execution engine can distribute code packages associated the operator instances to the assigned computing environments. The workflow execution engine can maintain a memoization repository to cache one or more outputs of the operator instances upon completion of the execution requests.

Abstract: In one embodiment, one or more computing devices access a social graph including nodes and edges between nodes; the nodes include user nodes corresponding to users of a social-networking system and concept nodes corresponding to concepts of the social-networking system; the computing devices identify first user nodes in the social graph corresponding to users having a particular user characteristic; the computing devices assign a characteristic score to each first user node; for a second user node without an assigned characteristic score, the computing devices also select neighboring user nodes connecting to the second user node based on affinity, and estimate a characteristic score for the second user node by aggregating iteratively characteristic scores while traversing the social graph from at least one first user node to the second user node through paths that each includes at least one selected neighboring node.

Abstract: A waveguide display includes a source assembly, an output waveguide, and a controller. The source assembly includes a light source and an optics system. The light source includes source elements arranged in a 1D or 2D array that emit image light. The optics system includes a scanning mirror assembly that scans the image light to particular locations based on scanning instructions. The output waveguide receives the scanned image light from the scanning mirror assembly and outputs an expanded image light. In some embodiments, the waveguide display includes a source waveguide and the 1D array of source elements. The source waveguide receives a conditioned image light from the source assembly. The controller generates the scanning instructions and provides the scanning instructions to the scanning mirror assembly. In some embodiments, the controller provides the scanning instructions to an actuator assembly of the source waveguide.

Abstract: An electronic device has one or more processors, a display, and memory. The memory stores one or more programs configured for execution by the one or more processors. The device receives, from a content delivery network, a program manifest including one or more video segments of a video. The electronic device parses the program manifest to identify a timeline for the video that spans the video segments. The electronic device receives, from a social-networking server, a playback offset for the video. In accordance with the playback offset and the timeline for the video, the electronic device determines a designated video segment and a playback position within the designated video segment. The electronic device then plays the video segments sequentially on the electronic device, commencing at the playback position within the designated video segment.

Abstract: In one embodiment, a computer server running a social networking application aggregates raw local area network (LAN) traffic data received from one or more listening nodes in one or more LANs. The aggregated LAN traffic data is comprised of multiple entries, each of which includes a MAC address for a networked device, as well as an association between each MAC address and a user of a social networking system. The computer server may then detect, identify, and qualify recurring patterns when a particular user is on the same LAN as other users of the social networking system. Based upon the qualified patterns, the social networking system may suggest friend connections or other interactions on the social networking system to the particular user.

Abstract: In one embodiment, a computing system may access a social graph comprising a plurality of nodes and a plurality of edges connecting the nodes, each of the edges between two of the nodes representing a single degree of separation between them, where the nodes comprise a first node corresponding to a first user of an online social network, and a plurality of second nodes that each correspond to a concept or a second user associated with the online social network. The computing system may receive a search query from the first user. The computing system may generate one or more search results corresponding to the search query, where each search result corresponds to a node of the plurality of second nodes. The computing system may score each search result based on a proximity coefficient between the first node and the node corresponding to the search result.

Abstract: An interface adapter may include a printed circuit board that includes an edge connector dimensioned to be inserted into an expansion socket of a computing device. The interface adapter may also include a plurality of modular computing components removably mounted to a top surface of the printed circuit board, where each modular computing component includes a connector dimensioned to be inserted into a socket of the printed circuit board and the connector of each modular computing component includes a pinout that is more compact than a pinout of the edge connector of the printed circuit board. In addition, the interface adapter may include one or more modular heatsinks coupled to a top surface of each modular computing component to facilitate heat dissipation from the top surface of the modular computing component. Various other apparatuses, systems, and methods are also disclosed.

Abstract: A social networking system notifies its users of different events and actions taking place inside the social networking system. To present notifications that are most likely to be of interest to the user, the social networking system observes the interactions of the user with notifications presented to the user and suggests an option to opt-out of notifications associated with a source based on the observed interactions. The option to opt-out of notifications associated with the source may be presented if the users interactions with the notifications presented are below a threshold. Notifications associated with a social group are presented to the user based on factors describing the group and the relation between the user and other users of the social group, for example, seniority of the user, size of the group, and the rate at which user actions associated with the group are received.

Abstract: In one embodiment, a user of a social networking system requests to look up a contact's communication channel information in the user's address book. The social networking system provides to the user the contact's communication channels in an ordered sequence based on the user's social graph and communication information.

Abstract: A self-contained circuit board actuator including a pivot bearing connectable to a circuit board and an actuator lever pivotably coupled to the pivot bearing. The lever is moveable between a closed position and an open position. A pair of opposed pawls extend from a first end of the lever and a latch notch is formed in the first end adjacent the pair of opposed pawls. The latch notch engages a catch tab on the pivot bearing when the lever is in the closed position. The lever is coupled to the pivot bearing by a pivot member and the lever is slidable along the pivot member between a latched position and an unlatched position. The lever includes a pivot bore to receive the pivot member and a resilient member is positioned in the pivot bore around the pivot member to urge the lever toward the latched position.

Abstract: In one embodiment, a method includes accessing a post associated with a first user of an online social network; extracting n-grams from the post; determining whether the post is associated with a trending topic based on whether one or more of the extracted n-grams are associated with the trending topic; identifying one or more second users, wherein each second user is a first-degree connection of the first user within the online social network and wherein each second user is a subscribing user; generating, for each second user, a live social module comprising the post; and sending, to a system of each second user, the live social module.

Abstract: In one embodiment, a method includes receiving an input corresponding to a search query from an online page of a social-networking system. The online page corresponds to a target user of the social-networking system. The method also includes identifying one or more of a number of content objects associated with the target user. The association is based at least in part on analyzing text, audio, or visual content of each of the plurality of content objects as well as interactions by the target user with each of the plurality of content objects. The method also includes scoring each of the identified content objects based at least in part on an affinity score of the content object relative to a querying user; and ranking each of the identified content objects based at least in part on the affinity score.

Abstract: In one embodiment, a method includes automatically determining, in response to user input that includes an indication that the user is generating social-network content, a location of the user. The social network has a number of nodes and a number of edges connecting the nodes, and at least one node corresponds to the user. The method includes sending, automatically and without requiring further user input, the location from a client computing device of the user to an external server computing device and receiving from the server an identification of one or more places corresponding to the location.

Abstract: A method clusters audio sources in virtual environments. The method is performed at a virtual-reality device displaying a virtual environment. The device identifies two audio sources in the virtual environment. For each of the two audio sources, the device determines a bounding box in the virtual environment. Each bounding box includes termination points for a respective plurality of rays emanating from a point in the virtual environment corresponding to the audio source. The device applies an overlap test to the bounding boxes to determine whether the two audio sources are in a same room. The device forms an angle according to rays from the location of the listener to the audio source points. When the two audio sources are in the same room and the angle is less than a predetermined threshold angle, the device clusters the two audio sources together, including rendering combined audio for the two audio sources.

Abstract: A computer-implemented method for charging partial-use batteries may include (1) determining a discharge duration for a partial-use battery, (2) estimating an age of the partial-use battery, (3) calculating, based on the age, a partial charge voltage for the partial-use battery that will enable a discharge cycle of the partial-use battery to last for the discharge duration, and (4) floating the voltage of the partial-use battery at the partial charge voltage to prolong the partial-use battery's life. The partial-use battery may be a backup battery of a power supply that will experience a transition from a first power source to a second power source, the transition may be expected to last for a transitional duration, the backup battery may supply power to the power supply during the transition, and the discharge duration may be greater than or equal to the transitional duration. Various other methods, systems, and apparatus are also disclosed.

Abstract: A key assembly for a keyboard may include a key cap that includes a touch sensor that is communicatively coupled to a control device. When the touch sensor senses a touch event, the touch sensor may send a touch signal to the control device, and when the touch sensor senses an absence of touch, the touch sensor may send a control signal to the control device. The key assembly may further include a switch that is communicatively coupled to the control device and that is movable between a pressed position and an un-pressed position. When the switch is in the pressed position, the switch may send a closed signal to the control device, and when the switch is in the un-pressed position, the switch may send an open signal to the control device. Various other apparatuses, systems, and methods are also disclosed.