Abstract: Certain aspects involve facilitating the integration of sensitive data from a data provider into an instance of a web-based, third-party application. For example, a data provider service can receive an authentication API call from a third-party system. The authentication API call can include a user identifier and a request for an access token usable by a web-based interface of the third-party system. The data provider service can generate an access token for the third-party system from which the authentication API call is received. The data provider service can subsequently receive, from the user device, a feature API call including the access token and a feature request for sensitive data. The data provider service can generate output data specific to the user identified by the access token included in the feature API call. The data provider service can provide the output to the user device via the web-based interface.

Abstract: This disclosure involves development and deployment platforms for decision algorithms. For example, a computing system provides software development interface to a client device. The system sets, based on an input from the client device via the interface, a decision engine to a test mode that causes the decision engine to operate on test data stored in a first database and that prevents the decision engine from applying operations from the client device to production data stored in a second database. The system also configures the decision engine in the test mode to execute a different decision algorithms on the test data. The system also sets, based on another input via the interface, the decision engine to a deployment mode that causes the decision engine to operate on the production data. The system configures the decision engine in the deployment mode to execute one or more of the tested decision algorithms.

Abstract: According to certain implementations, a data-entry system synchronizes a region of an image with a data-entry field. For example, a data-entry interface may include data-entry fields. The data-entry interface may be updated to display an image of a document, such as a scanned handwritten form. The document image may include regions that are associated with respective data-entry fields. The data-entry system may detect a selection of a target data-entry field. Based on the detected selection, the data-entry system may determine an associated region of the document image. The associated region may correspond to the target data-entry field. The data-entry system may modify the displayed document image, and the modification may include a visual enhancement of the associated region. In some cases, the data-entry system determines the associated region corresponding to the target data-entry field based on metadata.

Abstract: The present disclosure involves systems and methods for identity authentication across multiple institutions using a trusted mobile device as a proxy for a user login. In one example, the operations include identifying a request to trust a particular user associated with a first entity in a digital ID network. A set of personally identifiable information (PII) associated with the user is obtained via the first entity and an identity verification (IDV)/fraud risk analysis is performed. In response to satisfying the analysis, instructions are transmitted to the user to verify the identity via a mobile trust application on an associated mobile device. Upon verification, the mobile device is bound to the user within the digital ID network along with a digital ID associated with the particular user. The digital ID can be used by other entities registered within the digital ID network to authenticate the user.

Abstract: Certain aspects involve building segment-specific executable program code. In one example, a code-building service can execute segmentation logic that assigns different target entity records to different segments based on differences between sets of attribute values among the target entity records. The code-building service can select, for each segment, a set of data assets that is specific to the assigned segment and a set of source code portions that is specific to the selected data assets. The code-building service can order each set of the source code portions based on an identified modeling output type for the target entity records. The code-building service can generate, from the ordered source code portions, a set of program code referencing the selected subset of the data assets. For instance, the generated program code, if executed, can generate and transmit different modeling outputs for the first target entity record and the second target entity record.

Abstract: Certain embodiments involve generating or optimizing a neural network for generating analytical or predictive outputs. The neural network can be generated using a relationship between various predictor variables and an outcome (e.g., a condition's presence or absence). The neural network can be used to determine a relationship between each of the predictor variables and a response variable. The neural network can be optimized by iteratively adjusting the neural network such that a monotonic relationship exists between each of the predictor variables and the response variable. The optimized neural network can be used both for accurately determining response variables using predictor variables and determining adverse action codes for the predictor variables, which indicate an effect or an amount of impact that a given predictor variable has on the response variable.

Abstract: In some aspects, a computing system can generate and optimize a neural network for risk assessment. The neural network can be trained to enforce a monotonic relationship between each of the input predictor variables and an output risk indicator. The training of the neural network can involve solving an optimization problem under a monotonic constraint. This constrained optimization problem can be converted to an unconstrained problem by introducing a Lagrangian expression and by introducing a term approximating the monotonic constraint. Additional regularization terms can also be introduced into the optimization problem. The optimized neural network can be used both for accurately determining risk indicators for target entities using predictor variables and determining explanation codes for the predictor variables. Further, the risk indicators can be utilized to control the access by a target entity to an interactive computing environment for accessing services provided by one or more institutions.

Abstract: Certain aspects involve optimizing neural networks or other models for assessing risks and generating explanatory data regarding predictor variables used in the model. In one example, a system identifies predictor variables. The system generates a neural network for determining a relationship between each predictor variable and a risk indicator. The system performs a factor analysis on the predictor variables to determine common factors. The system iteratively adjusts the neural network so that (i) a monotonic relationship exists between each common factor and the risk indicator and (ii) a respective variance inflation factor for each common factor is sufficiently low. Each variance inflation factor indicates multicollinearity among the common factors. The adjusted neural network can be used to generate explanatory indicating relationships between (i) changes in the risk indicator and (ii) changes in at least some common factors.

Abstract: Examples are disclosed for detecting synthetic online entities that may be used for fraudulent purposes or other purposes. In some aspects, a computing system can generate a data structure that includes nodes and links between the nodes. The nodes can represent online entities and the links can represent geographic associations or transactional associations between pairs of online entities. These associations can be identified from electronic transactions involving the online entities. The computing system can determine, from the links between the nodes, that a degree of connectivity among a subset of the nodes exceeds a threshold connectivity. The degree of connectivity indicates electronic communications involving online entities represented by the subset of the nodes. The computing system can transmit, based on the degree of connectivity exceeding the threshold connectivity, an alert indicating a potential synthetic entity (e.g., potentially fraudulent activity) within the subset of the nodes.

Abstract: Certain embodiments involve generating or optimizing a neural network for risk assessment. The neural network can be generated using a relationship between various predictor variables and an outcome (e.g., a condition's presence or absence). The neural network can be used both for accurately determining risk indicators or other outputs using predictor variables and for determining adverse action codes explaining the predictor variables' effect or an amount of impact on the risk indicator.

Abstract: An automated model development tool can be used for automatically developing a model (e.g., an analytical model). The automated model development tool can perform various automated operations for automatically developing the model including, for example, performing automated operations on variables in a data set that can be used to develop the model. The automated operations can include automatically analyzing the predictor variables. The automated operations can also include automatically binning (e.g., combining) data associated with the predictor variables to provide monotonicity between the predictor variables and one or more output variables. The automated operations can further include automatically reducing the number of predictor variables in the data set and using the reduced number of predictor variables to develop the analytical model.

Abstract: Certain aspects involve training tree-based machine-learning models for computing predicted responses and generating explanatory data for the models. For example, independent variables having relationships with a response variable are identified. Each independent variable corresponds to an action or observation for an entity. The response variable has outcome values associated with the entity. Splitting rules are used to generate the tree-based model, which includes decision trees for determining relationships between independent variables and a predicted response associated with the response variable. The tree-based model is iteratively adjusted to enforce monotonicity with respect to representative response values of the terminal nodes. For instance, one or more decision trees are adjusted such that one or more representative response values are modified and a monotonic relationship exists between each independent variable and the response variable.

Abstract: Certain embodiments involve generating or optimizing a neural network for risk assessment. The neural network can be generated using a relationship between various predictor variables and an outcome (e.g., a condition's presence or absence). The neural network can be used to determine a relationship between each of the predictor variables and a risk indicator. The neural network can be optimized by iteratively adjusting the neural network such that a monotonic relationship exists between each of the predictor variables and the risk indicator. The optimized neural network can be used both for accurately determining risk indicators using predictor variables and determining adverse action codes for the predictor variables, which indicate an effect or an amount of impact that a given predictor variable has on the risk indicator. The neural network can be used to generate adverse action codes upon which consumer behavior can be modified to improve the risk indicator score.

Abstract: Certain aspects involve building and debugging models for generating source code executed on data-processing platforms. One example involves receiving an electronic data-processing model, which generates an analytical output from input attributes weighted with respective modeling coefficients. A target data-processing platform is identified that requires bin ranges for the modeling coefficients and reason codes for the input attributes. Bin ranges and reason codes are identified. Modeling code is generated that implements the electronic data-processing model with the bin ranges and the reason codes. The processor outputs source code, which is generated from the modeling code, in a programming language used by the target data-processing platform.

Abstract: In some aspects, an interface is provided that has options for defining entities and attributes. A model of a remote data source is defined based on inputs received via the interface. The model includes an entity representing a data segment from the remote data source. The entity has properties, which represent fields of a data structure in the remote data source and have data type descriptors indicating data types used by the fields. The model also includes an attribute specifying a function, which uses the properties as inputs. The entity is matched to exported data from the remote data source. The match is performed based on the exported data elements being arranged in a hierarchy corresponding to the properties and the data elements having the data types corresponding to the data type descriptors. The attribute's function is executed using, as input data, the data elements mapped to the properties.

Abstract: In some aspects, a computing system can obtain, from a host server, a data network identifier that identifies a mobile device involved in an online transaction within an interactive computing environment provided by the host server. The computing system can also obtain, from a telecommunication provider server, a credential corresponding to the data network identifier. The computing system can match the credential to a set of authorization elements. Each authorization element can allow the mobile device to access a function for advancing the online transaction within the interactive computing environment. The computing system can generate a dynamic authorization data structure that includes the authorization elements. The computing system can cause the host server to advance, responsive to receiving a selection from the mobile device identifying one of the authorization elements, the online transaction within the interactive computing environment.

Abstract: Attribute data structures can be updated to indicate joint relationships among attributes and predictive outputs in training data that can be used for training automated modeling system. A data structure that stores training data for training an automated modeling algorithm can be accessed. The training data can include first data for a first attribute and second data for a second attribute. The data structure can be modified to include a derived attribute that indicates a joint relationship among the first attribute, the second attribute, and a predictive output variable. The automated modeling algorithm can be trained with the first attribute, the second attribute, and the derived attribute.

Abstract: According to certain implementations, a data-entry system synchronizes a region of an image with a data-entry field. For example, a data-entry interface may include data-entry fields. The data-entry interface may be updated to display an image of a document, such as a scanned handwritten form. The document image may include regions that are associated with respective data-entry fields. The data-entry system may detect a selection of a target data-entry field. Based on the detected selection, the data-entry system may determine an associated region of the document image. The associated region may correspond to the target data-entry field. The data-entry system may modify the displayed document image, and the modification may include a visual enhancement of the associated region. In some cases, the data-entry system determines the associated region corresponding to the target data-entry field based on metadata.

Abstract: In some aspects, an entity-resolution computing system for entity resolution is provided. The entity-resolution computing system includes an entity resolution computing device configured as an interface between a client computing device and an encrypted identity data repository that contain resolved entity dataset. The entity resolution computing device is configured for servicing a resolution request from the client computing device by matching encrypted indexes generated from identity data objects stored in a client identity database to encrypted data objects stored in the encrypted identity data repository. The resolution computing device retrieves and transmits a common entity identifier associated with the encrypted data objects so that the client computing device can link the identity data objects stored in a client identity database via the common entity identifier.

Abstract: Certain aspects produce a scoring model that can automatically classify future text samples. In some examples, a processing device perform operations for producing a scoring model using active learning. The operations includes receiving existing text samples and searching a stored, pre-trained corpus defining embedding vectors for selected words, phrases, or documents to produce nearest neighbor vectors for each embedding vector. Nearest neighbor selections are identified based on distance between each nearest neighbor vector and the embedding vector for each selection to produce a text cloud. Text samples are selected from the text cloud to produce seed data that is used to train a text classifier. A scoring model can be produced based on the text classifier. The scoring model can receive a plurality of new text samples and provide a score indicative of a likelihood of being a member of a selected class.