Visa International Service Association

Verint Americas Inc.

Abhinav Aggarwal

Cynthia Freeman

Ian Roy Beaver

Mihai Christodorescu

Mahdi Zamani

Jonathan Patrick Merriman

Mastooreh Salajegheh

Jingjing Ren

Rohit Sinha

Systems and methods are described to combine two or more linear models into a combined linear model. Two or more linear models and an observation of interest are selected. The linear models are concurrent with respect to the observation of interest. The observation of interest includes a class value and a feature vector. A distance is selected, and a plurality of feature vectors are selected that are within the distance of the feature vector associated with the observation of interest. These feature vectors are input to the selected linear models and a plurality of class values are generated. These class values and the selected feature vectors are used to generate a combined linear model. The combined model is either a mean local linear surrogate model or a linearly weighted local surrogate model.

Systems and methods for combining model interpreters

A method for maintaining user privacy in advertisement networks may include receiving first persona data associated with a first user from at least one publisher system. The first persona data may be generalized to form first generalized persona data. A session key may be generated. The first generalized persona data and the session key may be encrypted with a first public key of an advertisement network system to form a first ciphertext. The first ciphertext and first user identity data may be communicated to a mediator system. The first user identity data may be associated with first user's identity. A first encrypted targeted advertisement may be received based on the first generalized persona data from the advertisement network system via the mediator system. The first encrypted targeted advertisement may be decrypted with the session key to form a first targeted advertisement. A system and computer program product are also disclosed.

System, method, and computer program product for maintaining user privacy in advertisement networks

Methods and systems for generating a security policy at a gateway are disclosed. A server computer and a gateway can perform a protocol in order to train a security model at a gateway, such that it can detect attack packets and prevent those attack packets from reaching the server computer via the gateway. In a learning phase, the server computer can provide training packets and test packets to the gateway. The gateway can use the training packets to train a security model, and the gateway can classify the test packets using the security model in order to test its accuracy. When the server computer is satisfied with the accuracy of the security policy, the server computer can transmit an acceptance of the security policy to the gateway, which can subsequently deploy the model in order to detect and filter attack packets.

Remote configuration of security gateways

Attention weights in a hierarchical attention network indicate the relative importance of portions of a conversation between an individual at one terminal and a computer or a human agent at another terminal. Weighting the portions of the conversation after converting the conversation to a standard text format allows for a computer to graphically highlight, by color, font, or other indicator visible on a graphical user interface, which portions of a conversation led to an escalation of the interaction from an intelligent virtual assistant to a human customer service agent.

System and method of highlighting influential samples in sequential analysis

A method is disclosed. A node in a plurality of nodes can perform an identity set generation process. The node can then determine a leader node. The node may diffuse an identity set from each node of the plurality of nodes to the plurality of nodes. The node can then determine a majority set including identities occurring in at least one half of the identity sets, wherein the leader node diffuses the majority set of the leader node to the plurality of nodes. The node can verify the majority set of the leader node. The node may then update the identity set based on the majority set of the leader node.

Sybil-resistant identity generation

A method for maintaining user privacy in advertisement networks may include receiving first persona data associated with a first user from at least one publisher system. The first persona data may be generalized to form first generalized persona data. A session key may be generated. The first generalized persona data and the session key may be encrypted with a first public key of an advertisement network system to form a first ciphertext. The first ciphertext and first user identity data may be communicated to a mediator system. The first user identity data may be associated with the first user's identity. A first encrypted targeted advertisement may be received based on the first generalized persona data from the advertisement network system via the mediator system. The first encrypted targeted advertisement may be decrypted with the session key to form a first targeted advertisement. A system and computer program product are also disclosed.

A technique for oblivious filtering may include receiving an input data stream having a plurality of input elements. For each of the input elements received, a determination is made as to whether the input element satisfies a filtering condition. For each of the input elements received that satisfies the filtering condition, a write operation is performed to store the input element in a memory subsystem. For those of the input elements received that do not satisfy the filtering condition, at least a dummy write operation is performed on the memory subsystem. The contents of the memory subsystem can be evicted to an output data stream when the memory subsystem is full. The memory subsystem may include a trusted memory and an unprotected memory.

Oblivious filtering of data streams

A method of determining influence of language elements in script to an overall classification of the script includes performing a sentiment analysis of the language elements. In some instances, for example, in a conversation, the language elements and turns within the conversation (e.g., in a chat bot) are analyzed for their influence in escalation or non-escalation of the conversation to a higher level of resolution, e.g., to a human representative or manager.

Growing community of inventors

Albuquerque, NM, United States of America

Abhinav Aggarwal