Abstract: A system, having a memory that stores computer executable components, and a processor that executes the computer executable components, reduces data size in connection with training a neural network by exploiting spatial locality to weight matrices and effecting frequency transformation and compression. A receiving component receives neural network data in the form of a compressed frequency-domain weight matrix. A segmentation component segments the initial weight matrix into original sub-components, wherein respective original sub-components have spatial weights. A sampling component applies a generalized weight distribution to the respective original sub-components to generate respective normalized sub-components. A transform component applies a transform to the respective normalized sub-components.

Abstract: A method of providing physiological self-regulation challenges prior to participating in a series of activities or exercises at a series of predefined locations includes determining a physiological goal associated with each location. A sensing device measures a physiological state of a user, and a mobile communication device communicates to a user whether or not the user has achieved the physiological goal for the challenge. The level of difficulty of the physiological goal may be reduced if the user does not meet the goal. The physiological goal may comprise a brain state that is conducive to learning, and the sensing device may be configured to measure brain state values representing cognitive engagement. Upon achievement of each physiological goal, the participant is provided with a reward such as information concerning the current predefined location and/or information concerning the next predefined location.

Abstract: A neural network method and apparatus is provided. A processor-implemented neural network method includes determining, based on a determined number of classes of input data, a precision for a neural network layer outputting an operation result, and processing parameters of the layer according to the determined precision.

Abstract: Various implementations of the invention for predicting customer behavior are described.

Abstract: A system and method of determining a policy to prevent fading drivers is described. The system and method creates virtual trajectories of incentives such as coupons offered to drivers in a transportation hailing system and corresponding states of drivers in response to the incentives. A joint policy simulator is created from an incentive policy, a confounding incentive policy, and an incentive object policy to generate the simulated actions of drivers in response to different incentives. The rewards of the simulated actions of the drivers is determined by a discriminator. The incentive policy for preventing fading drivers is optimized by reinforcement learning based on the virtual trajectories generated by the joint policy simulator and discriminator.

Abstract: A multi-party, multi-point type decentralized loyalty system and associated processes for promotion and point issuance using a permission-based distributed ledger are disclosed for promotion and point issuance across multiple point/promotion issuers and redeemers via a permission-based distributed ledger. This system allows an issuer and a set of partners to gain transparent access to both (a) the rules of the system in the form of promotions and exchange rates defined in a distributed ledger and (b) consumer accrual and redemption transactions posted to the distributed ledger, and thereby to allow the complete coalition to operate in a decentralized fashion, with quick on-boarding of new partners and transparent reconciliation of point balances between partner and issuer.

Abstract: We describe an artificial neural network comprising: an input layer of input neurons, one or more hidden layers of neurons in successive layers of neurons above the input layer, and at least one further, concept-identifying layer of neurons above the hidden layers. The neural network includes an activation memory coupled to an intermediate, hidden layer of neurons between the input concept-identifying layers to store a pattern of activation of the intermediate layer. The neural network further includes a system to determine an overlap between a plurality of the stored patterns of activation and to activate in the intermediate hidden layer an overlap pattern such that the concept-identifying layer of neurons is configured to identify features of the overlap patterns. We also describe related methods, processor control code, and computing systems for the neural network. Optionally further, higher level concept-identifying layers of neurons may be included.

Abstract: An automatic tag identification system identifies tags for color themes, a color theme referring to a set of multiple colors that work well together to create (e.g., are visually appropriate for creating) a desired effect, and a tag referring to one or more words that describe a color theme. The automatic tag identification system receives an indication of the multiple colors (e.g., five colors) for a color theme. A first machine learning system uses the indicated multiple colors to generate a color theme embedding for the color theme, which is a vector encoding or embedding of the color theme. The second machine learning system uses the color theme embedding generated by the first machine learning system to generate one or more tags that label the color theme. These one or more tags can then be saved as associated with or corresponding to the multiple colors for the color theme.

Abstract: Methods and apparatus to estimate census level audience sizes, impression counts, and duration data are disclosed. Example instructions cause one or more processor to at least set up a system of equations based on first census data; select a census parameter value based on a constraint for census parameter selection, the constraint based on media access represented in panel data; iteratively solve for second census data that satisfies the system of equations using the panel data and the census parameter value, the second census data including an audience size value, an impression count value, and a duration value for a media item accessed by anonymous audience members in a demographic group; and generate a report including the second census data.

Abstract: This application relates to apparatus and methods for automatically determining and providing digital advertisements to targeted users. In some examples, a computing device receives campaign data identifying items to advertise on a website, and generates campaign user data identifying a user that has engaged all of the items on the website. The computing device may then determine a portion of the users based on a relationship between each user and the campaign user data, and may determine user-item values for each of the items for each user of the portion of users, where each user-item value identifies a relational value between the corresponding user and item. The computing device may then identify one or more of the items to advertise to each user of the portion of users based on the user-item values, and may transmit to a web server an indication of the items to advertise for each user.

Abstract: A system including one or more processors and one or more non-transitory computer-readable media storing computing instructions configured to run on the one or more processors and perform extracting meta features for an item to generate sparse feature embeddings for the item; reducing, using a multilayer perceptron, a dimension of the sparse feature embeddings to generate a representation vector for the meta features; extracting, using a recurrent neural network, sequential data from dense traffic features for the item over a period of time; and inputting the representation vector for the meta features and the sequential data from the dense traffic features into a multilayer neural network with a rectified linear unit (ReLU) activation function and a scoring layer to generate one or more performance metrics for the item. Other embodiments are disclosed.

Abstract: Methods and systems for deep neural networks using dynamically selected feature-relevant points from a point cloud are described. A plurality of multidimensional feature vectors arranged in a point-feature matrix are received. Each row of the point-feature matrix corresponds to a respective one of the multidimensional feature vectors, and each column of the point-feature matrix corresponds to a respective feature. Each multidimensional feature vector represents a respective unordered point from a point cloud and each multidimensional feature vector includes a respective plurality of feature-correlated values, each feature-correlated value represents a correlation extent of the respective feature. A reduced-max matrix having a selected plurality of feature-relevant vectors is generated.

Abstract: Content presented to a user of an online system is presented along with an option allowing the user to view one or more reasons for presentation of the advertisement content to the user and one or more characteristics of the user satisfying targeting criteria associated with the content. For example, a description identifying one or more selected characteristics of the user satisfying the targeting criteria is presented along with the content. The online system selects the one or more characteristics presented along with the content by applying one or more rules to characteristics of the user satisfying targeting criteria included in the content. The rules may account for accuracy of models used to determine whether the user has characteristics, revenue received by the online system from presenting content associated with targeting criteria identifying various characteristics, or prevalence of characteristics among users or targeting criteria of content.

Abstract: Techniques are described for automatically extracting items from a user's browse history on one or more e-commerce websites, and automatically searching other e-commerce websites and sources for offers (including coupons, deals, and/or promotions) for those items and/or related items and/or stores. Such automated techniques allow users to gain the benefit of online offers from various sources, without having to go through the effort of manually searching for such offers. Instead, the system and method automatically locate such offers based on the user's browse history, without requiring any action to be taken on the part of the user.

Abstract: A system and computer-implemented method are provided for enabling control of a physical system based on a state of the physical system which is inferred from sensor data. The system and method may iteratively infer the state by, in an iteration, obtaining an initial inference of the state using a mathematical model representing a prior knowledge-based modelling of the state, and by applying a learned model to the initial inference of the state and the sensor measurement, wherein the learned model has been learned to minimize an error between initial inferences provided by the mathematical model and a ground truth and to provide a correction value as output for correcting the initial inference of the state of the mathematical model. Output data may be provided to an output device to enable control of the physical system based on the inferred state.

Abstract: Systems, methods, and apparatuses for implementing commerce rewards across tenants for commerce cloud customers utilizing blockchain technologies in conjunction with a cloud based computing environment are described herein.

Abstract: This application relates to apparatus and methods for automatically determining and providing digital advertisements to targeted users. In some examples, a computing device receives campaign data identifying items to advertise on a website, and generates campaign user data identifying a user that has engaged all of the items on the website. The computing device may then determine a portion of the users based on a relationship between each user and the campaign user data, and may determine user-item values for each of the items for each user of the portion of users, where each user-item value identifies a relational value between the corresponding user and item. The computing device may then identify one or more of the items to advertise to each user of the portion of users based on the user-item values, and may transmit to a web server an indication of the items to advertise for each user.

Abstract: Aspects of the subject disclosure may include, for example, a method for processing an advertising request message, including, receiving an advertising request message from a mobile device, determining if an advertising exchange server associated with the advertising request is permitted to receive enhanced information associated with the mobile device if a service provider is permitted to provide location information associated with the mobile device responsive to determining that the advertising exchange server associated with the advertising request is permitted to receive the enhanced information, responsive to determining that the service provider is permitted to provide location information, adding location information associated with the mobile device to a header of the advertising request message, and transmitting the advertising request message to the advertising exchange server. Other embodiments are disclosed.

Abstract: A computer-implemented method is disclosed for training neural networks of a group recommender to provide item recommendations for ephemeral groups having group interaction sparsity. A preference encoder and aggregator generate user and group preference embeddings from user-item interactions, wherein the preference embeddings form a latent user-group latent embedding space.

Abstract: A method and computer-readable medium for carrying out a promotional campaign, comprising: identifying at least one influencer of at least one target of the promotional campaign; identifying at least one indicator of association between the influencer and a brand associated with the promotional campaign; and releasing towards the target a data element derived from the indicator of association.