Abstract: A method for evaluating robustness of one or more target neural network models using natural typos. The method includes receiving one or more natural typo generation rules associated with a first task associated with a first input document type, receiving a first target neural network model, and receiving a first document and corresponding its ground truth labels. The method further includes generating one or more natural typos for the first document based on the one or more natural typo generation rules, and providing, to the first target neural network model, a test document generated based on the first document and the one or more natural typos as an input document to generate a first output. A robustness evaluation result of the first target neural network model is generated based on a comparison between the output and the ground truth labels.

Abstract: Systems, methods, and devices facilitate generation of application programming interfaces. Systems include a computing platform implemented using a server system, the computing platform being configurable to cause monitoring application programming interface (API) traffic associated with a client service, storing a representation of the API traffic in a database system, the representation including at least an API call identifier and metadata, and generating one or more API models based, at least in part, on the representation of the API traffic, the one or more API models being configured to represent a plurality of API interactions associated with the client service. The computing platform is further configurable to cause generating one or more API objects based, at least in part, on the API models, the one or more API objects being one or more components of a client service API.

Abstract: Systems and methods are provided for rendering components of a user interface (UI) of an application such that primitive child elements that correspond to child components of the UI can be rendered without instantiating the child components. Each child data object corresponds to a particular child component of a parent component and includes data for that particular child component. Child data objects can be augmented to include a simple render method for rendering structured data of the augmented child data objects. The parent component be rendered from its template by iteratively evaluating each child data object and checking a property of that child data object to determine whether it includes a simple render method, and if so, the template of the parent component can instruct the framework to use that simple render method to get an atom object and render one or more primitive child elements.

Abstract: Disclosed are some implementations of systems, apparatus, methods and computer program products for implementing a scalable computing system. The scalable computing system includes an intermediate system that facilitates communications between a core server system and a third-party system. The core server system processes a client request for a third-party service in association with a web page having a corresponding web address. The intermediate system communicates with the core server system to obtain a session token, and transmits the session token and web address to the third-party system. The third-party system may then access the web page via the web address using the session token.

Abstract: Disclosed are some implementations of systems, apparatus, methods and computer program products for implementing a customizable intent driven action triggering system. The system retrieves, for at least a first intent, a corresponding list of actions. The system provides, to a first client device associated with a first user account, a first graphical user interface (GUI) identifying the first intent, the list of actions, and for at least the first action in the list of actions, a corresponding user-selectable option configurable to indicate whether the first action is activated for the first intent in association with the first user account. The system processes a first request received from the first client device in association with the user-selectable option, where the first request indicates that the first action is activated for the first intent in association with the first user account.

Abstract: Among other things, embodiments of the present disclosure relate to detecting and mitigating errors associated with software pull requests (PRs). Other embodiments may be described or claimed.

Abstract: Systems and methods are provided for securely sharing database records in the context of a conversation at a communication platform using a secure sharing service. The secure sharing service identifies an external system user identifier for a participant in the conversation at the communication platform using an identifier associated with the participant at the communication platform, identifies permissions associated with the data record at the external system, and identifies a subset of fields of the data record viewable by the participant using the external system user identifier. The secure sharing service obtains values for the identified subset of fields of the data record from the external computing system and automatically generates a shared data record message associated with the conversation at the communication platform that includes the obtained values for the identified subset of fields of the data record.

Abstract: Disclosed are some implementations of systems, apparatus, methods and computer program products for implementing a scalable computing system. The scalable computing system includes an intermediate system that facilitates communications between a core server system and a third-party system. The core server system processes a client request for a third-party service in association with a web page having a corresponding web address. The intermediate system communicates with the core server system to obtain a session token, and transmits the session token and web address to the third-party system. The third-party system may then access the web page via the web address using the session token.

Abstract: A method of early warning and risk assessment of incidents in a multi-tenant cloud environment is provided. The method includes: capturing a plurality of data metrics; automatically generating derived features from the plurality of captured data metrics; automatically selecting risk assessment features from the derived features and the captured data metrics; and predicting the risk of an incident in the multi-tenant cloud environment within a specified time window in the future and one or more possible root causes of the incident by applying the newly selected risk assessment features to a trained risk assessment model. The trained risk assessment model has been trained using machine learning techniques to predict the risk of an incident in the multi-tenant cloud environment within a specified time window in the future, provide an explanation of possible root causes of the incident, and assign a strength level to each possible root cause.

Abstract: Technologies are provided for supporting tenant customizations in a cloud computing platform. A standard application programming interface (API) can be defined to support a generic set of parameters that allow tenant-specified custom data to be passed to and from the standard API. The generic set of parameters allow the standard API to process requests and responses that include a payload that comprises tenant-specified custom data. The payload allows the tenant-specified custom data to be passed to and from the standard API. When a request is received at the cloud computing platform, a request payload of the request, that includes tenant-specified custom data, can be processed at the standard API, and a response can be sent from the cloud computing platform that includes a response payload that includes the tenant-specified custom data.

Abstract: Systems and methods are described for migrating a workflow from a source system to a target system. The systems and methods receive metadata defining a workflow configured for processing by the source system. A set of rules associated with transforming the metadata for processing by the target system are identified. A first rule is applied to the metadata to execute a first transformation of an element of the metadata. A second rule is applied to the metadata to execute a second transformation of a position of an element of the metadata. The systems and methods generate a transformed workflow including the first transformation and the second transformation. The transformed workflow is displayed via the target system.

Abstract: Verification of discriminative models includes receiving an input; receiving a prediction from a discriminative model for the input; encoding, using an encoder, a latent variable based on the input; decoding, using a decoder, a reconstructed input based on the prediction and the latent variable; and determining, using an anomaly detection module, whether the prediction is reliable based on the input, the reconstructed input, and the latent variable. The encoder and the decoder are jointly trained to maximize an evidence lower bound of the encoder and the decoder. In some embodiments, the encoder and the decoder are further trained using a disentanglement constraint between the prediction and the latent variable. In some embodiments, the encoder and the decoder are further trained without using inputs that are out of a distribution of inputs used to train the discriminative model or that are adversarial to the discriminative model.

Abstract: Methods, and apparatuses are provided for access limitations to a network in a session using a formatted web token. The method includes: formatting a web token by a schema to create a formatted web token for user access to the network; receiving a log-in request for the user access to the network server via an app hosted by a computing device remotely located to the network server; in response to receipt of a user access request, creating the session by the network server with network limitations for user access to data and resources of the network; passing the formatted web token to a client for enabling user access to the data and resources of the network; decoding payload data of the formatted web token at the client to authenticate the user access; and enabling the client with access limitations based on decoded payload data.

Abstract: Systems and methods for unifying question answering and text classification via span extraction include a preprocessor for preparing a source text and an auxiliary text based on a task type of a natural language processing task, an encoder for receiving the source text and the auxiliary text from the preprocessor and generating an encoded representation of a combination of the source text and the auxiliary text, and a span-extractive decoder for receiving the encoded representation and identifying a span of text within the source text that is a result of the NLP task. The task type is one of entailment, classification, or regression. In some embodiments, the source text includes one or more of text received as input when the task type is entailment, a list of classifications when the task type is entailment or classification, or a list of similarity options when the task type is regression.

Abstract: Disclosed herein are system, method, and computer program product embodiments for providing a data model application programming interface (“API”) to customizable live applications in a cloud collaboration platform. The cloud collaboration platform provides data storage, a data model, and an API to embedded third-party live applications to allow the live applications to securely create, store, and access data within the cloud collaboration platform. The data model allows records stored in the cloud collaboration platform to sync independently, merge intelligently, and function offline.

Abstract: A method for managing a user's email subscriptions has been developed. The method provides a user interface (UI) to the user that allows the user to view and select multiple stand-alone components across multiple dashboard displays for an email subscription to an email address. The user's selections are received via the UI. A display of each of the subscribed stand-alone components is rendered periodically based on the user's preference. The rendered displays are combined into a single email message that is sent to the email address upon completion of the rendering of the displays.

Abstract: Methods and systems are provided for automatically determining a need to invoke functionality of a feature of an application while using a primary application and for automatically invoking that functionality without user action. When input information is input into the primary application via a user interface of a user system, it can be received by an application server from a sender process at the user system, and a pre-fix that occurs before a text string can be identified. The text string that follows the pre-fix can be extracted by performing a matching analysis against a library of keywords and extracting a matching keyword. The text string comprises extracted text from the input information. The matching keyword can be matched to an entry that is mapped to a command that can be executed to generate an action that automatically invokes the functionality of the feature of the application without user action.

Abstract: A method and system uses a test cloud system for monitoring cloud to cloud performance. The method includes initiating a first trust relationship by the test cloud system with a first target cloud system, receiving a success indicator for the trust relationship, in response to the first target cloud verifying the test cloud system is whitelisted, generating a security token using a private key of the test cloud system, and invoking a function of an application programming interface of the first target cloud using the security token to validate functionality of the function of the application programming interface of the first target cloud.

Abstract: Database systems and methods are provided for data migration between source and target computing systems using a configurable transformation service. The configurable transformation service obtains transformation configuration metadata associated with the particular combination of source and target computing systems, obtains a subset of source data in accordance with the transformation configuration metadata and transforms the subset of source data from a format associated with the source computing system to an intermediate, platform-independent format. The service transforms the subset of source data from the intermediate format to a destination format associated with the target computing system in accordance with the transformation configuration metadata and maintains the transformed subset of source data in the destination format in association with the target computing system.

Abstract: Described herein are systems, apparatus, methods and computer program products configured for image detection/recognition of products. The disclosed systems and techniques utilize video data to provide the necessary number of images and view angles needed to train a machine learning product detection/recognition system to recognize a specific product within later provided images. In various embodiments, a user may provide video data and the video data may be transformed in a manner that may aid in training of the machine learning system.