Abstract: Systems and methods for user-agent anomaly detection are disclosed. In one embodiment, a user-agent string may be embedded into a numerical data vector representation using a sentence embedding algorithm (e.g., FastText). A predictive score may be calculated based on the numerical data vector representation and using a probability distribution function model that models a likelihood of occurrence of the observed user-agent based on patterns learned from historic payload data (e.g., a Gaussian Mixture Model). The predictive score may be compared to a threshold and, based on the comparison, it may be determined whether the user-agent is fraudulent.

Abstract: A plurality of user segments is defined. Each user segment has a respective profile corresponding to one or more characteristics shared by one or more users. A plurality of elements of a user interface for a mobile application is defined. Each user segment is associated with a different subset of the elements. A first request is received to display the user interface on a first mobile device of a first user. In response to the first request, user data of the first user is analyzed. Based on the analysis, a first user segment to which the first user belongs is determined. The mobile application is then instructed to display the user interface according to a first customized layout on the first mobile device. The first customized layout includes a first subset of elements associated with the first user segment.

Abstract: There are provided systems and methods for edge computing storage nodes based on location and activities for user data separate from cloud computing environments. A service provider, such as an online transaction processor, may provide additional services for to users via edge computing systems and edge computing storage nodes. The service may be for data that may be predictively loaded to the edge computing storage node for a particular location, where the edge computing storage node may reside more locally to the location on a network so that data may be served quicker and with less network resource consumption than providing data from a remote cloud computing storage. The data may be predicted to be needed or useful to the user at the location using a user profile for the user, monitored user activities, and/or one or more machine learning models that predict user behaviors at the location.

Abstract: A method includes receiving a set of training data and selecting a first machine learning platform based on a first optimization function that metrics past machine learning platforms used for training on the set of training data. The method also includes selecting a first algorithm supported by the first machine learning platform based on a second optimization function that metrics past algorithms used for training on the set of training data. Further, the method includes determining one or more hyperparameters supported by the first algorithm based on a third optimization function that metrics past combinations of hyperparameters from the set of hyperparameters used for training on the set of training data. The method also includes training a machine learning model on the set of training data using the first machine learning platform, the first algorithm, and the one or more hyperparameters.

Abstract: A method may include determining that input text data includes a first keyword from a first set of keywords. The method also includes determining a similarity between the input text data and a first stored text string that has previously been identified as a false positive match for the first keyword, and based on the similarity, generating a first false positive score corresponding to the input text data. Further, the method includes determining a number of keywords, from a second set of keywords, that are included in the input text data, and based on the number of keywords, generating a second false positive score corresponding to the input text data. The method also includes calculating a final false positive score corresponding to the input text data based on the first false positive score and the second false positive score.

Abstract: Methods and systems for using machine learning to automatically determine a data loading configuration for a computer-based rule engine are presented. The computer-based rule engine is configured to use rules to evaluate incoming transaction requests. Data of various data types may be required by the rule engine when evaluating the incoming transaction requests. The data loading configuration specifies pre-loading data associated with at least a first data type and lazy-loading data associated with at least a second data type. Statistical data such as use rates and loading times associated with the various data types may be supplied to a machine learning module to determine a particular loading configuration for the various data types. The computer-based rule engine then loads data according to the data loading configuration when evaluating a subsequent transaction request.

Abstract: Computer system security and efficiency of processing operations may be improved using techniques that are described relating to analyzing user actions based on time of day of occurrence, and using time of a day as a factor in determining whether a particular action should be allowed or disallowed. Past action times can be transformed to a two-dimensional representation using a radial time schema that avoids discontinuity. A probability distribution can indicate a likelihood of whether a new action fits a previous pattern. If a new user action is relatively unlikely due to time of day, the new user action might be denied/prevented from completing, thus enhancing computer system security and avoiding unnecessary computational processing costs.

Abstract: Methods and systems are presented for detecting groups within a graph using computer-based multi-threading techniques. These techniques provide technical improvements in computing power and efficiency for analysis of large graphs. A group detection system accesses a graph. Threads are instantiated to perform task related to a group detection process based on the nodes in the graph, where a thread is instantiated for a corresponding node. Each thread determines a neighbor count representing a number of neighbor nodes having one degree of separation from the corresponding node. Each thread also generates a list comprising an identity of the corresponding node and identities of the neighbor nodes. The thread transmits the list only to threads corresponding to a first subset of the neighbor nodes having more neighbors than the corresponding node, but not to threads corresponding to a second subset of the neighbor nodes having less neighbors than the corresponding nodes.

Abstract: Methods and systems for providing merchant in-context checkout are described. A user is authenticated based on credentials received from a first application running on a computing device. An authentication code is provided to the first application. A signed verifier and the authentication code is then received from a second application running on the computing device. The authentication code and the signed verifier received from the second application are then validated, and a device token is provided to the second application upon validation. The device token is exchangeable by the second application for an access token that is usable for making payment calls from the second application.

Abstract: Techniques for managing erasure of data are presented. In response to receiving a request for erasure of data from a set of data stores, an erasure component can analyze a set of rules and information relating to the user account, including an account status and erasure hold status associated with the user account. The set of rules can be based on legal or contractual obligations applicable to the set of data stores, and can indicate various conditions under which data associated with a user account of a user can be eligible to be erased from the set of data stores or an associated data vault repository. The erasure component can determine eligibility for erasure of all or a portion of the set of data from the set of data stores based on the analysis results. Erasure component can determine erasure eligibility scores to pre-qualify user accounts for erasure eligibility.

Abstract: Systems and methods for recommending merchant actions include a physical merchant location having a plurality of beacon devices. A system provider device receives a plurality of first location information from the plurality of beacon devices that is collected from a customer beacon communication device that is associated with a customer. The system provider device also receives a plurality of second location information over the network from the plurality of beacon devices that is collected from a merchant beacon communication device that is associated with an asset of the merchant. The system provider device analyzes the plurality of first location information and the plurality of second location information to determine at least one merchant action recommendation related to the customer and the asset of the merchant and provides the at least one merchant action recommendation over the network for display on a display device.

Abstract: Techniques are disclosed relating to methods that include receiving an indication of an access by a user to a web page that includes a beacon, and calculating a readiness score for triggering the beacon. The methods may also include determining, based on the readiness score, whether to perform a client-side or server-side triggering of the beacon. The triggering causes data associated with the access to be transmitted to a third-party computer system.

Abstract: Techniques for data lifecycle discovery and management are presented. Data lifecycle discovery platform (DLDP) can identify data of users, data type, and language of data stored in data stores (DSs) of entities based on scanning of data from databases. DLDP determines compliance of DLDP and DSs with obligations relating to data protection arising out of jurisdictional laws or agreements. DLDP generates rules to facilitate complying with and enforcing laws and agreements. DLDP can determine, and present to authorized users, risk scores relating to levels of compliance of the DLDP, associated platforms, or entities, risk indicator metrics, or a privacy health index of the organization associated with DLDP. DLDP can manage user rights regarding data, and access to data in DSs and information relating thereto stored in secure data store of DLDP. DLDP can remediate issues involving anomalies indicating non-compliance. DLDP can utilize machine learning to enhance various functions of DLDP.

Abstract: Methods and systems for managing loyalty programs using a decentralized blockchain are provided. Transactions between users and a merchant are monitored via an application programming interface (API) of a service provider. Based on the monitoring, it may be determined that at least one of the users has satisfied reward criteria for a loyalty program associated with the merchant. The loyalty program includes a plurality of tokens to be distributed to users who satisfy the reward criteria configured by the merchant via a merchant interface. At least one token from among the plurality of tokens may be selected for the user, based on at least one smart contract associated with the merchant. The smart contract is stored on the blockchain in association with a unique identifier for the merchant. A transaction is broadcasted to the decentralized blockchain for transferring the at least one token to a digital wallet of the user.

Abstract: Computing systems and methods for facilitating mobile transactions are disclosed. An example method includes: detecting, via a hardware communication interface, a user check-in request based at least on a communication exchange between a user device and a signaling device at a location. The communication exchange corresponds to a proximity of the user device to the signaling device. The method further includes: determining a user identifier associated with the user device based at least on the user check-in request; and transmitting one or more promotional items to the user device based at least on the user identifier.

Abstract: Methods and systems are presented for detecting a boundary of a document within a digital image. Upon receiving an image, the image is converted into a binary image. One or more kernel-based transformations are performed on the binary image using a horizontal kernel and a vertical kernel. A plurality of edges are identified based on the one or more kernel-based transformations. The plurality of edges includes a plurality of horizontal edges and a plurality of vertical edges. Multiple quadrilaterals are constructed using different combinations of horizontal edges and vertical edges from the plurality of edges. A particular quadrilateral is selected from the multiple quadrilaterals based on how well the edges fit the perimeters of the quadrilaterals. The selected quadrilateral is used to define a boundary of the document within the digital image.

Abstract: Computer systems and methods for interfacing with a third party website, comprising determining that a webpage of the third party website is being output in an interface of a web browser, and in response, generating an interface element that at least partially overlaps the webpage in the interface. The interface element includes a user-selectable graphical trigger that, when selected, causes digital codes to be automatically input into a data entry interface identified in the webpage, thereby triggering the third party website to receive each of the digital codes and to return a respective response. Based on monitoring the respective response to each of the digital codes, a particular digital code is identified that causes a greatest change in a numerical value displayed in the webpage. The particular digital code is applied to obtain a resulting numerical value that is displayed in the webpage.

Abstract: According to an aspect of an embodiment of the present disclosure, a method may include receiving a quick response (QR) code from a computing device, the QR code including a first entity identifier, and analyzing the QR code to determine a request to initiate a data transfer between a first entity associated with the first entity identifier and a second entity, and to determine one or more characteristics associated with the data transfer. The method may also include, based on the one or more characteristics, identifying a first QR code processing flow from multiple QR code processing flows to follow to process the data transfer, and verifying the first entity in the transaction based on the first entity identifier of the QR code and based on the first QR code processing flow. The method may also include processing the data transfer between the second entity and the first entity based on the first QR code processing flow.

Abstract: Methods and systems are presented for generating a machine learning model using enhanced gradient boosting techniques. The machine learning model is configured to receive inputs corresponding to a set of features and to produce an output based on the inputs. The machine learning model includes multiple layers, wherein each layer includes multiple models. To generate the machine learning model, multiple models are built and trained in parallel for each layer of the machine learning model. The multiple models use different subsets of features to produce corresponding output values. After a layer in built and trained, a collective error may be determined for the layer based on the output values from the different models in the layer. An additional layer of models may be added to the machine learning model to reduce the collective error of a previous layer.

Abstract: Systems and methods for determining activated neural pathways in knowledge graphs are disclosed. In an embodiment, a computer system may retrieve data associated with a user account and generate a knowledge graph centered around an account under a current adjudication. The computer system may determine paths based on the knowledge graph using similarity and sentiment scores. By reducing the amount of data in the knowledge graph to specific data associated with paths of interest, unnecessary computer operations may be avoided, which increases processing times related to the knowledge graph and allows near real-time outputs based on the knowledge graph to be utilized. The computer system may rank the paths based on their similarity scores and sentiments scores to determine which paths are considered activated neural pathways in the knowledge graph. The computer system may base a determination of the current adjudication for the account on the activated neural pathways.