Abstract: A method includes providing a distributed instance of a machine learning model to a client computing device and access to a predetermined token associated with a predetermined label to a client-side application at the client computing device. The client-side application is configured to cause the client computing device to detect a user input that includes the first predetermined token, provide the user input to the distributed instance to obtain a first predicted label for the first predetermined token, and update a set of model parameters of the distributed instance based on the first predicted label and the first predetermined label. The method may also include updating the machine learning model by obtaining the one or more updated model parameters of the distributed instance from the client computing device.

Abstract: A secure method of real time money transferring between account holders utilizes Near Field Communication (NFC) technologies, dynamic transaction cards, and a mobile application to facilitate a seamless mobile transfer of funds. The dynamic transaction card of the customer requesting to make the transfer of funds may be placed in contact with the second recipient dynamic transaction card of the customer receiving the funds transfer to establish an NFC connection. The dynamic transaction card requesting the transfer may receive, via the NFC antenna on the card, the recipient customer account information stored on a passive NFC tag. The dynamic transaction card requesting the transfer may transmit the recipient account information to a mobile application, which may transmit the recipient account information to an account provider system for facilitating the funds transfer.

Abstract: Provided herein are systems and methods for extracting text from a document. Different optical character recognition (OCR) tools are used to extract different versions of the text in the document. Metrics evaluating the quality of the extracted text are compared to identify and select higher quality extracted text. A selected portion of text is compared to a threshold to ensure minimal quality. The selected portion of text is then saved. Error correction can be applied to the selected portion of text based on errors specific to the OCR tools or the document contents.

Abstract: In some embodiments, an authentication system may be configured to process one or more tokens that incorporates one or more tailored-wavelength-range coded patterns, codes represented within a steganographic image, or “overlapping” codes. As an example, such a token may include (i) one or more invisible-ink-printed patterns printed using infrared ink corresponding to different tailored wavelength ranges within the infrared light spectrum, (ii) one or more steganographic images in which invisible-ink-printed patterns are interweaved within the steganographic images, (iii) visible-ink-printed patterns and invisible-ink-printed patterns printed substantially within the visible ink portions of the visible-ink-printed patterns, or (iv) other features, where the invisible-ink-printed patterns correspond to one or more authentication codes or other data for token authentication.

Abstract: A device receives document information associated with a document, and receives a request to identify insights in the document information. The device performs, based on the request, natural language processing on the document information to identify words, phrases, and sentences in the document information, and utilizes a first machine learning model with the words, the phrases, and the sentences to identify information indicating abstract insights, concrete insights, and non-insights in the document. The device utilizes a second machine learning model to match the abstract insights with particular concrete insights that are different than the concrete insights, and utilizes a third machine learning model to determine particular insights based on the non-insights. The device generates an insight document that includes the concrete insights, the abstract insights matched with the particular concrete insights, and the particular insights determined based on the non-insights.

Abstract: In some implementations, a system may receive a first credential associated with a first account and a second credential associated with a second account and may instruct a cloud service, using the first credential, to generate a first snapshot of structured source data associated with the first account. The system may authorize the first account and the second account to use a master encryption key and instruct the cloud service to encrypt the first snapshot using the master encryption key. The system may instruct the cloud service, using the second credential, to copy the first snapshot to a second snapshot associated with the second account and to decrypt the second snapshot into structured target data using the master encryption key. The system may deauthorize the first account and the second account from using the master encryption key and output an indicator of completion to a user device.

Abstract: Methods, systems, and apparatuses are described herein for improving computer authentication processes by generating authentication questions based on the location of a user. Transaction data indicating a plurality of transactions associated with a user account may be received. Location data indicating a plurality of locations of a user device might be received. At least a subset of the plurality of transactions may be tagged, based on the location data, with an indication that a user was present for a respective transaction. For example, a location of a merchant might be compared to a user device location indicated by the location data. A plurality of authentication questions might be generated based on the subset of the plurality of transactions. Access to the user account might be provided based on responses to the plurality of authentication questions.

Abstract: Aspects described herein may relate to techniques for detecting login activity to a financial account during a knowledge-based authentication process. The login activity may be related to access to an online interface for the financial account. The detection of login activity during the authentication process my indicate that the integrity of the authentication process is compromised as login access may provide an individual with transaction data that may be used to answer transaction-based authentication questions. As a result of detecting login activity, an alternative authentication process may be initiated or an authentication request related to the financial account may be denied.

Abstract: Methods and systems are described herein for a transfer authorization system that may monitor for transfer requests for secondary cryptographic addresses maintained for one or more primary cryptographic addresses. When a transfer request is received, the transfer authorization system may determine that the transfer request is for a secondary cryptographic address. Based on that determination, the transfer authorization system may determine that the source cryptographic address of the request is authorized to receive control of cryptographic tokens from the secondary cryptographic address and that the request is associated with the source cryptographic address. The transfer authorization system may then transfer control of one or more cryptographic tokens from an associated primary cryptographic address to the secondary cryptographic address, and upon the successful transfer, may also transfer control of the one or more tokens to the requestor.

Abstract: A device attempts to access a resource that requires a multi-factor authentication (MFA), and receives, from an MFA server device, a challenge form. The device performs natural language processing on the challenge form to determine a first location of an input for a security code and a second location of a verify button. The device provides, to an email server device, a request to access emails associated with a user of the device, and receives access to multiple emails associated with the user. The device processes the multiple emails to identify an MFA email in the multiple emails, and identifies a security code in the MFA email. The device provides the security code at the first location, and selects the verify button at the second location. The device provides the security code to the MFA server device, and receives information indicating whether the device is authenticated.

Abstract: In some implementations, a cloud system may obtain data via a copy of an application executing in a first cloud deployment zone of the multiple cloud deployment zones, wherein the application has copies executing in respective cloud deployment zones of the multiple cloud deployment zones. The cloud system may store, in a database executing in the first cloud deployment zone, the data with a deployment zone identifier that indicates the first cloud deployment zone, where the multiple cloud deployment zones include respective databases of multiple databases including the database. The cloud system may perform, via a cloud resource in the first cloud deployment zone, an operation using the data based on the data including the deployment zone identifier that indicates the first cloud deployment zone and based on the cloud resource being in the first cloud deployment zone.

Abstract: A computing system identifies a plurality of virtualization elements executing on a cloud computing environment. The computing system retrieves a first set of data related to software resources executing across the plurality of virtualization elements. The computing system retrieves a second set of data related to hardware resources associated with the plurality of virtualization elements. The computing system retrieves delta updates to the first set of data, wherein the delta updates are retrieved in real-time. The computing system generates a first database. The first database includes the first set of data related to the software resources, the second set of data related to the hardware resources associated with the plurality of virtualization elements, and the delta updates. The computing system generates a configuration management database hosted by a configuration management database system.

Abstract: A computer-implemented method for demonstrating a personalized automated teller machine (ATM) presentation to a user may include: obtaining transaction data of the user via a device associated with the user; obtaining, via the one or more processors, geographic data of the user based on the transaction data; generating, via the one or more processors, ATM data based on the geographic data of the user; obtaining, via the one or more processors, user feedback data based on the ATM data, wherein the user feedback data comprises a selection of an ATM of the list of the ATMs; transmitting, to the selected ATM of the list of the ATMs, presentation data based on the transaction data and the user feedback data; and demonstrating, via the selected ATM of the list of the ATMs, the personalized ATM presentation to the user based on the presentation data.

Abstract: In some implementations, an autonomous vehicle may receive information associated with an order including multiple items. The autonomous vehicle may, after traveling to a first pickup location and obtaining a subset of the items in the order, identify one or more alternative pickup locations having respective inventories including one or more items that are unavailable at the first pickup location. The autonomous vehicle may generate one or more virtual credentials linked to a primary credential associated with a user that submitted the order and use the one or more virtual credentials to obtain, at each alternative pickup location, one or more items that were unavailable at the first pickup location. The autonomous vehicle may travel to a final destination to deliver the subset of the multiple items obtained from the first pickup location and the one or more items obtained from each of the one or more alternative pickup locations.

Abstract: Systems and methods for implementing an automated process for detection of a shared recurring bills and identification of a set of corresponding person-to-person transactions associated with shared portions of the recurring bill. The identification process is prediction-based and contingent upon an analysis of a user's transaction records. A key aspect of the analysis is directed towards transaction amount data (fraction test) and occurrence patterns, derived from the transaction records, in order to identify a shared bill and a set of person-to-person (P2P) transfer transactions associated therewith. Upon identification of a set of P2P transfer transactions associated with a recurrent shared bill, an automation option may be provided to the user.

Abstract: Disclosed embodiments may include a system for card authentication. The system may receive, via a user device, a voice request to activate a card associated with a customer account. The system may generate, using a first machine learning model (MLM), a voice fingerprint confidence score based on comparing a first voice in the voice request to a second voice in a stored audio file associated with a customer of the customer account. The system may generate, using a second MLM, a device authentication score based on comparing an address from the user device to a stored address of a known user device associated with the customer account. The system may determine whether the voice request is valid based at least in part on the voice fingerprint confidence score and the device authentication score. The system may activate the card based on the determination that the voice request is valid.

Abstract: Systems and methods are described for generating code summaries using bifurcated model architectures. The bifurcated model architecture may comprise a first model that generates code summaries based on native code script, and the second model may compare, de-duplicate, and/or categorize the code summaries into clusters that perform the same or similar functions. That is, the systems and methods may comprise a first model (e.g., a large language model) to predict new content (e.g., a code summary for a given native script code). The outputs of the first model are then inputted into a natural language processing (NLP) model to compare the code summaries. For example, the first model of the bifurcated architecture performs a crucial pre-processing step that prevents the NLP model from generating cluster descriptions that are specific to semantic structures, programming languages, and/or code formatting.

Abstract: Systems and methods for novel uses and/or improvements to data labeling applications, particularly data labeling applications involving sensitive data. As one example, systems and methods are described herein for preventing sensitive data leakage, using weak learner libraries, during label propagation.

Abstract: In certain embodiments, image-feature malicious-activity detection and access prevention may be facilitated, where an activity notification associated with a transaction triggers feature detection for at least one image related to a different transaction. In some embodiments, a first image (received in connection with a first transaction) may be associated with the first transaction, and a second image (received in connection with a second transaction) may be associated with a second transaction. In response to a malicious activity notification associated with the first transaction, an identifying characteristic (e.g., a facial feature) detected in the first image may be used to determine whether to prevent access to an account related to the second transaction. As an example, in response to determining that the identifying characteristic is present in the second image, access to the account related to the second transaction may be prevented.

Abstract: Systems and methods are described for performing secure communications with rotating key communications. For example, the system may receive a first applet request to perform a first rotating key communication across one or more computer networks to access a first server host. The system may then determine a first processing characteristic for the first applet request and a first plurality of server locations of the first server host (e.g., across the one or more computer networks). The system may then determine a first plurality of connection paths to a first subset of the first plurality of server locations. The system may then filter the first plurality of connection paths to generate a first subset of the plurality of connection paths. Based on the first subset of the plurality of connection paths, the system may generate a first rotating key for performing the rotating key communication.