Abstract: A method increases user interaction with deep learning agents. A generation request for a content slot is received. A subsequent tag vector, for the content slot, is generated from a previous tag, for a previous content slot, using a subsequent tag model. A context vector is generated from a set of subsequent tag vectors, which include the subsequent tag vector, using a context vector generator. A selection vector is generated from the context vector using a contextual bandit model. A content set is generated for a content slot using the selection vector. The content set for the content slot is presented.

Abstract: A method manages bot dialogue. A user input is converted to a phrase vector. A set of identified tokens are identified by a token identification engine from the phrase vector. An unsupervised token is selected from the set of identified tokens. A supervised token is selected from the set of identified tokens. A voted token selected from the unsupervised token and the supervised token. A next token is identified based on a set of recent tokens that includes the voted token. The next token is presented as one of a voice communication and an email communication.

Abstract: A method including receiving data including an unknown vector including a data structure populated with unknown features describing a first user and a score predicted by a MLM trained using a prediction data set. The score represents a prediction regarding the first user. The prediction data set includes the unknown vector stripped of a biased data set. The data also includes a prediction whether the first user belongs to the cohort. The method also includes hashing information types used by the primary MLM and the supervisory MLM to produce a first hashed data, the information types including at least the unknown vector, the score, and the prediction. The method also includes combining the first hash and a schema to produce a compliance document. The method also includes hashing the compliance document to produce a second hashed data. The method also includes storing the second hashed data in a blockchain.

Abstract: A method including receiving an unknown vector including a data structure populated with unknown features describing a user. The method also includes executing a primary machine learning model (MLM) trained using a prediction data set to predict a score representing a prediction regarding the user. The prediction data set includes the unknown vector stripped of a biased data set including markers set that directly indicate that the user belongs to a cohort against which bias is to be avoided. The method also includes executing a supervisory MLM trained using the prediction data set to predict whether the user belongs to the cohort. The method also includes performing, using an industry tool, a computer-implemented action using the score after executing the primary MLM and the supervisory MLM.

Abstract: A method uses personality-profiled language modeling for bots. An input phrase is received from a user. A language personality vector is determined using a language neural network and the input phrase. A response phrase to the input phrase is determined using the language personality vector from the language neural network. The response phrase is presented.

Abstract: A method includes receiving, by a loan program executing on a loan computing system, conditions from an originator program executing on an originator computing system. The conditions are sent, by the loan program, to a wallet program executing on a wallet computing system. Cryptographic proof that the conditions are met is received by the loan program from the wallet program. Ownership data of the loan program is updated by the loan program to identify a buyer program executing on a buyer computing system in response to invocation of a purchase function of the loan program. Payment receipt data is appended, by the loan program, to the loan program in response to invocation of a payment append function of the loan program by a loan service program executing on a loan service computing system.

Abstract: A method increases user interaction with deep learning agents. A generation request for a content slot is received. A subsequent tag vector, for the content slot, is generated from a previous tag, for a previous content slot, using a subsequent tag model. A context vector is generated from a set of subsequent tag vectors, which include the subsequent tag vector, using a context vector generator. A selection vector is generated from the context vector using a contextual bandit model. A content set is generated for a content slot using the selection vector. The content set for the content slot is presented.

Abstract: A method uses personality-profiled language modeling for bots. An input phrase is received from a user. A language personality vector is determined using a language neural network and the input phrase. A response phrase to the input phrase is determined using the language personality vector from the language neural network. The response phrase is presented.

Abstract: A method manages bot dialogue. A user input is converted to a phrase vector. A set of identified tokens are identified by a token identification engine from the phrase vector. An unsupervised token is selected from the set of identified tokens. A supervised token is selected from the set of identified tokens. A voted token selected from the unsupervised token and the supervised token. A next token is identified based on a set of recent tokens that includes the voted token. The next token is presented as one of a voice communication and an email communication.

Abstract: A method for image processing is disclosed. The method includes: obtaining an image associated with a check; obtaining target strings associated with a payor of the check and based on a user input; obtaining extracted strings by applying multiple optical character recognition (OCR) processes with different segmentation modes to the image; identifying, using an expandable and sliding window (ESW), matches between the plurality of target strings and the plurality of extracted strings; and selecting a winning match from the plurality of matches.

Abstract: A method for image processing is disclosed. The method includes: obtaining an image including a check with a magnetic ink character recognition (MICR) code; generating a mask including a plurality of shapes based on the image and an estimated rotation angle of the check; generating a stroke width map (SWM) by applying a stroke width transform (SWT) to a plurality of regions in the image corresponding to the plurality of shapes; generating a first word line associated with a first region based on a plurality of words in the SWM; rotating a portion of the SWM associated with the first word line; and detecting, after rotating, the MICR code by applying a plurality of OCR processes to the portion of the SWM.

Abstract: A method includes receiving, by a loan program executing on a loan computing system, conditions from an originator program executing on an originator computing system. The conditions are sent, by the loan program, to a wallet program executing on a wallet computing system. Cryptographic proof that the conditions are met is received by the loan program from the wallet program. Ownership data of the loan program is updated by the loan program to identify a buyer program executing on a buyer computing system in response to invocation of a purchase function of the loan program. Payment receipt data is appended, by the loan program, to the loan program in response to invocation of a payment append function of the loan program by a loan service program executing on a loan service computing system.

Abstract: A method including receiving data including an unknown vector including a data structure populated with unknown features describing a first user and a score predicted by a MLM trained using a prediction data set. The score represents a prediction regarding the first user. The prediction data set includes the unknown vector stripped of a biased data set. The data also includes a prediction whether the first user belongs to the cohort. The method also includes hashing information types used by the primary MLM and the supervisory MLM to produce a first hashed data, the information types including at least the unknown vector, the score, and the prediction. The method also includes combining the first hash and a schema to produce a compliance document. The method also includes hashing the compliance document to produce a second hashed data. The method also includes storing the second hashed data in a blockchain.

Abstract: A method for image processing is disclosed. The method includes: obtaining an image associated with a check; obtaining target strings associated with a payor of the check and based on a user input; obtaining extracted strings by applying multiple optical character recognition (OCR) processes with different segmentation modes to the image; identifying, using an expandable and sliding window (ESW), matches between the plurality of target strings and the plurality of extracted strings; and selecting a winning match from the plurality of matches.