Other

Quotient Technology Inc.

Sun Microsystems, Inc.

Coupons.com Incorporated

Sunil Keshav Bopardikar

Matthias Eichstaedt

Deepak Goel

Thomas E Saulpaugh

Rohan Bopardikar

Gregory K Slaughter

Xiaoyan Zheng

Nikhil Sunil Bopardikar

Zi-Hui Li

Systems and methods for improving a machine learning model are described. In an embodiment, a computing system stores a plurality of training examples comprising training inputs and training outputs. The computing system generates a machine learning model and training the machine learning model using the plurality of training examples. The computing system receives a particular input for the machine learning system and, using the particular input and the machine learning system, computes a particular output. For each training example of the plurality of training examples, the computing system adjusts a weight of the training example on the machine learning system and computes a relative numerical impact on the particular output for the training example, the relative numerical impact reflecting an importance of each training example on the particular output relative to an importance of the other training examples of the plurality of training examples on the particular output. The server computer generates training example relevance data comprising identifiers of the plurality of training examples and the relative numerical impact values for the plurality of training examples. The server computer stores the training example relevance data in the computing system.

Identifying contributing training datasets for outputs of machine learning models

Systems and methods for reducing the size of neural networks are disclosed. In an embodiment, a server computer stores a plurality of training datasets, each of which comprise a plurality of training input matrices and a plurality of corresponding outputs. The server computer initiates training of a neural network using the plurality of training input matrices, a weight matrix, and the plurality of corresponding outputs. While the training of the neural network is being performed, the server computer identifies one or more weight values of the weight matrix for removal. The server computer removes the one or more weight values from the weight matrix to generate a reduced weight matrix. The server computer then stores the reduced weight matrix with the neural network.

Reducing the size of a neural network through reduction of the weight matrices

In an embodiment, a method comprises causing, to be displayed to a user through a web page of a social network provider that provides a social network, contest data that indicates an online contest; receiving, at a particular entity, input that indicates an intention by the user to participate in the online contest; sending, from the particular entity, request data that indicates a request, by the particular entity, for one or more permissions that allows the particular entity to access information from the social network provider about the user; after sending the request data, receiving, at the particular entity, acceptance data that indicates that the user accepts the request; in response to receiving the acceptance data, causing the user to be entered in the online contest; wherein the method is performed by one or more computing devices.

Online contests with social networks

A server receives an indication that a user of a client device has selected social network data associated with a social network provider that maintains a social graph. The selected social network data indicates approval of a coupon offer by the user. In response to receiving the indication, the server computer provides the coupon to the user and sends a link message that causes a user node that is associated with the user to be linked, within the social graph, with a coupon node that is associated with the coupon offer. A message including data associated with the coupon offer is sent to a subset of a set of other users identified, in the social graph that the social network provider maintains, as connected to the user.

Unlocking coupon offers

In an embodiment, a method comprises receiving, at a social network coupon platform, a request to display coupon data regarding a particular coupon that is provided by a coupon provider, wherein a user that is registered with a social network provider initiated the request; in response to receiving the request, sending, from the social network coupon platform, the coupon data to be displayed to the user; receiving, at the social network coupon platform, from the social network provider, profile data that includes profile information of the user; in response to receiving the profile data, allowing the user access to the particular coupon; wherein the method is performed by one or more computing devices.

Coupon social networking platform

Systems and methods for reducing the size of deep neural networks are disclosed. In an embodiment, a server computer stores a plurality of training datasets, each of which comprise a plurality of training input matrices and a plurality of corresponding outputs. The server computer initiates training of a deep neural network using the plurality of training input matrices, a weight matrix, and the plurality of corresponding outputs. While the training of the deep neural network is being performed, the server computer identifies one or more weight values of the weight matrix for removal. The server computer removes the one or more weight values from the weight matrix to generate a reduced weight matrix. The server computer then stores the reduced weight matrix with the deep neural network.

In an embodiment, a method comprises receiving an indication that a user selected social network data that is displayed, on a client device, concurrently with offer data that indicates a coupon offer; wherein prior to receiving the indication, the user of the client device is incapable of accessing the coupon offer; in response to receiving the indication that the user selected the social network data, allowing the user to access the coupon offer, and causing a message to be sent to each user of a plurality of users that are identified, in a social network, as connected to the user, wherein the message includes data about the coupon offer or a coupon provider that initiated the coupon offer; wherein the method is performed by one or more computing devices.

Techniques are provided for allowing a user to access a coupon. In one approach, a social network coupon platform receives a request to display coupon data regarding a particular coupon that is provided by a coupon provider, wherein a user that is registered with a social network provider initiated the request. In response to receiving the request, sending the coupon data is sent form the social network coupon platform to be displayed to the user. Profile data that includes profile information of the user is received at the social network coupon platform from the social network provider. In response to receiving the profile data, the user is allowed access to the particular coupon.

An object-oriented interrupt processing system in a computer system creates a system database including a device namespace containing an entry for each device in the computer system and an interrupt namespace containing an entry for each interrupt source, arranged as an Interrupt Source Tree. Each entry in the Interrupt Source Tree is cross-referenced to a corresponding entry in the namespace and contains a reference to an interrupt handler for the corresponding interrupt source. When an interrupt occurs, a single interrupt dispatcher is invoked, to access the Interrupt Source Tree and cause execution of the corresponding interrupt handler.

Method and apparatus for object-oriented interrupt system

A method and apparatus for providing memory functionality to object-oriented client software components of a computer system having a CPU using a first set of memory classes, each class of the first set being platform-independent; a second set of memory classes, each class of the second set being a subclass of a class of the first set and being platform-dependent; and performing client component memory functions by accessing only objects of classes of the first set. Descriptions of the classes and methods are provided, enabling platform-independent device drivers to be implemented.

Apparatus and method for object-oriented memory system

An object-oriented interrupt processing system in a computer system creates a system database including a device namespace containing an entry for each device in the computer system and an interrupt namespace containing each entry in the interrupt source, arranged as an Interrupt Source Tree. Each entry in the Interrupt Source Tree is cross-referenced to a corresponding entry in the device namespace and contains a reference to an interrupt handler for the corresponding interrupt source. When an interrupt occurs, a single interrupt dispatcher is invoked, to access the Interrupt Source Tree and cause execution of the corresponding interrupt handler.

Growing community of inventors

Sunnyvale, CA, United States of America

Sunil Keshav Bopardikar