Abstract: Methods, computer readable media, and devices for rapid prototyping of both user experience and technical features in a combined approach. One method may include creating a set of known user experience (UX) components including a user interface element representing at least one of a data element and a data interaction, creating a UX screen including a set of UX elements within a UX prototyping tool, translating the UX screen into a set of standardized UX components, publishing the set of standardized UX components by mapping the set of standardized UX components to synthetic data, and validating a technical aspect of the published set of standardized UX components based on user interaction with the published set of standardized UX components.

Abstract: Embodiments described herein provide a method for text summarization. The method includes receiving a training dataset having at least an uncompressed text, a compressed text, and one or more information entities accompanying the compressed text. The method also includes generating, using a perturber model, a perturbed text with the one or more information entities being inserted into the compressed text. The method further includes training the perturber model based on a first training objective, and generating, using the trained perturber model, a perturbed summary in response to an input of a reference summary. The method further includes generating, via an editor model, a predicted summary by removing information from the perturbed summary conditioned on a source document of the reference summary, and training the editor model based on a second training objective.

Abstract: Embodiments described herein provide training a prompt generator for text classification. A first training dataset associated with a first plurality of class labels is received for a first training process. For a first instance of the first training dataset, a set of labels of interest is generated by sampling from a set of possible class labels including the first plurality of class labels. The prompt generator generates a first prompt based on the set of labels of interest. A pretrained language model generates a task output in response to an input of the first instance prepended with the first prompt. A loss objective is generated based on the task output and the set of labels of interest. Parameters of the prompt generator are updated based on the computed loss function via backpropagation while the pretrained language model is frozen.

Abstract: A method by one or more electronic devices to provision an infrastructure for a machine learning application in a multi-tenant on-demand serving infrastructure. The method includes storing a plurality of templates, wherein each of the plurality of templates indicates a scoring interface, a web server, a definition of a continuous integration pipeline, and a definition of a continuous deployment pipeline, receiving a request to provision the infrastructure for the machine learning application using a specified template from the plurality of templates, and provisioning the infrastructure for the machine learning application using the specified template to create a version control system repository, a continuous integration pipeline, and a continuous deployment pipeline.

Abstract: An application server may identify a set of functions for the source code associated with an uninterruptable software service. In some examples, the subset of the set of functions may include one or more functions having a computational expense greater than a threshold. The application server may create a compilation template to compile a subset of the set of functions to a shared library, and may perform an instrumentation to identify the subset of the set of functions based on the shared library. The application server may create an optimized version of the shared library based on performing the instrumentation. The application server may then change one or more portions of the source code to repoint to the optimized version of the shared library based on creating the optimized version of the shared library.

Abstract: A database entry may be stored in a container in a database table corresponding with a partition key. The partition key may be determined by applying one or more partition rules to one or more data values associated with the database entry. The database entry may be an instance of one of a plurality of data object definitions associated with database entries in the database. Each of the data object definitions may identify a respective one or more data fields included within an instance of the data object definition.

Abstract: Disclosed herein are system, method, and computer program product embodiments for automatic non-code test suite generation of an application programming language (API) specification. An embodiment operates by receiving a specification of an API, wherein the API comprises a plurality of endpoints. The embodiment generates, using a parser, an abstraction model corresponding to the specification of the API, wherein the abstraction model comprises a plurality of entities corresponding to the plurality of endpoints. The embodiment identifies, based on the abstraction model, an operation that is applicable to an entity of the plurality of entities. The embodiment then generates a functional test based on a use case corresponding to the entity and the operation.

Abstract: Embodiments described a method of video-text pre-learning to effectively learn cross-modal representations from sparse video frames and text. Specifically, an align and prompt framework provides a video and language pre-training framework that encodes the frames and text independently using a transformer-based video encoder and a text encoder. A multi-modal encoder is then employed to capture cross-modal interaction between a plurality of video frames and a plurality of texts. The pre-training includes a prompting entity modeling that enables the model to capture fine-grained region-entity alignment.

Abstract: Embodiments described herein provide methods and systems for open vocabulary object detection of images. given a pre-trained vision-language model and an image-caption pair, an activation map may be computed in the image that corresponds to an object of interest mentioned in the caption. The activation map is then converted into a pseudo bounding-box label for the corresponding object category. The open vocabulary detector is then directly supervised by these pseudo box-labels, which enables training object detectors with no human-provided bounding-box annotations.

Abstract: Method, apparatus and computer program product for multi-modal user authorization are described herein. For example, the apparatus includes at least one processor and at least one non-transitory memory including program code. The at least one non-transitory memory and the program code are configured to, with the at least one processor, receive a request to enable a first client device associated with a first user profile to access a group-based communication system; determine a first access privilege status for the first user profile; in response to determining that the first access privilege status is a general access privilege status, transmit, to the first client device, first user interface data associated with a general access user interface; and in response to determining that the first access privilege status is a limited access privilege status, transmit, to the first client device, second user interface data associated with a limited access user interface.

Abstract: Disclosed herein are system, method, and device embodiments for a data bridge adapter in a data integration platform that models any application programming interface as an entity-relationship model. This technique allows an individual using an integration platform to map and transform the entity-relationship model without having to create any custom code. The user may specify target parameters as part of an ETL process, and the entity relationship model may allow appropriate data and API calls to be generated to pass the data to the specified target. By creating an entity-relationship model from any API, the technique further supports versioning, schema change resolution, compatibility detection and other features. Furthermore, data visualization software may use the entity-relationship model to allow users to explore and analyze the data represented in the entity-relationship model.

Abstract: Embodiments described herein combine both masked reconstruction and predictive coding. Specifically, unlike contrastive learning, the mutual information between past states and future states are directly estimated. The context information can also be directly captured via shifted masked reconstruction—unlike standard masked reconstruction, the target reconstructed observations are shifted slightly towards the future to incorporate more predictability. The estimated mutual information and shifted masked reconstruction loss can then be combined as the loss function to update the neural model.

Abstract: Embodiments described herein provide an automated testing pipeline for providing a testing dataset for testing a trained neural network model trained using a first training dataset. A first testing dataset for the trained neural network including a first plurality of user queries is received. A dependency parser is used to filter the first plurality of user queries based on one or more action verbs. A pretrained language model is used to rank the remaining user queries based on respective relationships with queries in the first training dataset. Further, user queries that are classified as keyword matches with the queries in the first training dataset using a bag of words classifier are removed. A second testing dataset is generated using the ranked remaining user queries. Testing outputs are generated, by the trained neural network model, using the second testing dataset.

Abstract: The technology disclosed provides a quasi-recurrent neural network (QRNN) encoder-decoder model that alternates convolutional layers, which apply in parallel across timesteps, and minimalist recurrent pooling layers that apply in parallel across feature dimensions.

Abstract: In accordance with disclosed embodiments, there are provided methods, systems, and apparatuses for aggregating social network feed information including, for example, means for receiving user activity data from one or more social networks; parsing the user activity data to render a plurality of targeting parameters culled from the user activity data; matching one or more of the plurality of targeting parameters with advertising conditions for a social media campaign, wherein the advertising conditions of the social media campaign are contextually relevant to the one or more targeting parameters matched; recommending the social media campaign via a user interface; receiving authorization to launch the social media campaign via input received at the user interface or automatically launching the social media campaign based on pre-defined parameters; and deploying the social media campaign to one or many social media networks. Other related embodiments are disclosed.

Abstract: Disclosed herein are system, method, and computer program product embodiments for providing asynchronous backend processing of complex, time consuming, and/or expensive jobs. A command line interface (CLI) provides a command interface between a user and a backend processing server. The CLI sends a job request to the backend server, which immediately replies to the request with a confirmation message that includes a job identifier. This response is sufficient to allow the CLI to end its wait for the job response. Meanwhile, the backend server carries out the job by parsing the job into component tasks according to a declaration file and assigns those different tasks to different work servers. The backend server functions as a master server, tracking the statuses of the different tasks and assigning out new tasks until the job is complete. An overall job status is provided to the CLI upon request by referencing the job identifier.

Abstract: Techniques are disclosed relating to a database system includes worker nodes operable to perform transactions and director nodes operable to ensure transactional consistency for the transactions. A worker node may receive a request to perform a transaction involving writing a record. The worker node may then issue, to director nodes of the database system, a request for information that facilitates performance of an operation for the transaction. A director node may determine whether to approve the request based on whether the operation could cause transactional inconsistency in the database system. The worker node may proceed to perform the operation for the transaction in response to receiving approval responses from a majority of the director nodes, with none of the received responses indicating a disapproval of the transaction.

Abstract: A computing platform may be configurable to receive a request from a video conferencing application while a video conference is in session, the request including an indication of a selection of one or more participants in a video conference session. The computing platform may also be configurable to implement a query of a database system, the query being configured to retrieve user object data, and generate a result object based, at least in part, on the retrieved one or more user data objects, the result object being configured to display information about the at least one participant. The computing platform may be configurable to provide the result object to a client device to display a graphical user interface (GUI) window including the information about the at least one participant, the GUI window comprising a user interface element to provide access to at least one function associated with a computing platform.