Microsoft Technology Licensing, LLC

Yahoo! Inc.

Excalibur IP, LLC

John C Langford

Alekh Agarwal

Siddhartha Sen

Miroslav Dudik

Aleksandrs Slivkins

Eric J Horvitz

Dong Yu

Gang Li

James G Droppo

Frank Torsten Bernd Seide

Chi Wang

Kilian Quirin Weinberger

Lihong Li

Leon Bottou

Sharad Goel

Kishore A Papineni

Olivier Chapelle

Randolph Preston McAfee

Richard A Caruana

Debadeepta Dey

Marco Rossi

Markus Cozowicz

Cyril Zhang

Akanksha Saran

Jacob A Alber

Danielle Karen Bragg

Akshay Hall-krishnamurthy

Rajan N Chari

Olalekan Patrick Ogunmolu

Dylan J Foster

Dipendra Kumar Misra

Jonatan Efroni

Riashat Islam

Alexander Matthew Lamb

Paul S Mineiro

Qingyun Wu

Hanzhang Hu

Sham Kakade

Alexander L Strehl

Sarah L Bird

Generally discussed herein are devices, systems, and methods for calibration of a gaze indicator. A method can include receiving gaze tracker output data that indicates a gaze location on a display at which a user is gazing, determining an expected gaze location to which the user is expected to gaze after the gaze location, receiving further gaze tracker output data that indicates a subsequent gaze location on the display at which the user is gazing, and adjusting, based on the subsequent gaze location and the expected gaze location, a location of a visual gaze indicator on the display.

Autocalibration of gaze indicator

Systems and methods for tuning hyperparameters for a machine learning model using a challenger champion model are described. A set of challenger configurations are generated based on a hyperparameter for tuning and a subset of the set of challenger configurations are scheduled for evaluation based on a loss function. A loss value derived from the loss function for the challenger configurations is compared to a loss value derived from the loss function for a champion configuration, and the champion configuration is replaced with the challenger configuration based on the comparison of the loss value derived from the loss function for the challenger configuration and the loss value derived from the loss function for the champion configuration. When the champion is replaced, a new set of challenger configurations is generated based on the new champion configuration.

System and method for automatic hyperparameter selection for online learning

Generally discussed herein are devices, systems, and methods for determining a minimal controllable latent state and operating a model trained to implement the minimal controllable latent state. A method can include receiving first and second observations of a temporal series of observations produced by a sensor, the second observation separated from the first observation by more than one other observation, the first observation associated with a first action performed by an agent responsive to the first observation, encoding, by an encoder, the first and second observations into respective first and second hidden state representations, combining the first and second hidden state representations resulting in a combined representation, operating, by an action predictor, on the combined representation resulting in a second action, and altering weights of the encoder and the action predictor based on a difference between the first and second actions.

Controllable latent space discovery using multi-step inverse model

This document relates to automated generation of machine learning models, such as neural networks. One example system includes a hardware processing unit and a storage resource. The storage resource can store computer-readable instructions cause the hardware processing unit to perform an iterative model-growing process that involves modifying parent models to obtain child models. The iterative model-growing process can also include selecting candidate layers to include in the child models based at least on weights learned in an initialization process of the candidate layers. The system can also output a final model selected from the child models.

Automated generation of machine learning models

The claimed subject matter includes techniques for providing an application testing service with a user interface that enables a user to evaluate performance data for computer implemented decision policies. An example method includes displaying a first reward statistic comprising an actual performance result for a policy implemented by an application. The method also includes obtaining experimental data corresponding to previously implemented policies, computing a second reward statistic for a hypothetical policy using a reward function applied to the experimental data. The method also includes displaying the second reward statistic together with the first reward statistic to enable a user to compare the first reward statistic and the second first reward statistic.

Application testing

Techniques and constructs can reduce the time required to determine solutions to optimization problems such as training of neural networks. Modifications to a computational model can be determined by a plurality of nodes operating in parallel. Quantized modification values can be transmitted between the nodes to reduce the volume of data to be transferred. The quantized values can be as small as one bit each. Quantization-error values can be stored and used in quantizing subsequent modifications. The nodes can operate in parallel and overlap computation and data transfer to further reduce the time required to determine solutions. The quantized values can be partitioned and each node can aggregate values for a corresponding partition.

Computing system for training neural networks

The claimed subject matter includes techniques for providing an application testing service. An example method includes receiving context information from a client system, the context information comprising parameters that describe details of a user interaction with an application under test (AUT). The method also includes receiving a set of potential actions from the client system. The method also includes identifying a selected action from the set of potential actions and sending the selected action to the client system, wherein the AUT is customized based on the selected action. The method also includes receiving reward data from the client system based on a user's interaction with the AUT. The method also includes storing the context information, the selected action, and the reward data to a log of application test data.

Method, system, and programs for distributed machine learning on a cluster including a plurality of nodes are disclosed. A machine learning process is performed in each of the plurality of nodes based on a respective subset of training data to calculate a local parameter. The training data is partitioned over the plurality of nodes. A plurality of operation nodes are determined from the plurality of nodes based on a status of the machine learning process performed in each of the plurality of nodes. The plurality of operation nodes are connected to form a network topology. An aggregated parameter is generated by merging local parameters calculated in each of the plurality of operation nodes in accordance with the network topology.

Method and system for distributed machine learning

Embodiments are directed towards employing an admission controller (AC) network device to coordinate voluntary requests by traffic source devices (TSDs) to transmit traffic over a network. The TSDs submit voluntary requests to transmit network traffic during an allocated time frame to the AC. The AC monitors historical network traffic data and, based on various allocation policies, provides permission to at least some of the TSDs in the form of a nonexclusive lease of bandwidth with a rate cap for an allocated time frame. The TSDs receiving the lease voluntarily agree to transmit traffic not exceeding the rate cap for the time frame of the lease. TSDs that receive a zero rate cap voluntarily agree not to transmit. However, urgent network traffic bypasses the AC. The allocation policies used to determine the rate cap and number of permitted senders include a reactive approach, a predictive approach, and a predictive-reactive approach.

Voluntary admission control for traffic yield management

Embodiments are directed towards using a community of weighted results from local and global message classifiers to determine whether a message is spam. Each local classifier may receive a message that is to be evaluated to determine whether it is spam. A local classifier receives the message and performs a classification of the message. The local classifier may receive predictions of whether the message is spam from at least one global classifier. The local and global predictions are combined using, in one embodiment, a regression analysis to generate a single local message classification. Combining the local and global predictions is directed towards enabling a community of predictions to be used to classify messages. The user may then re-classify this output, which in turn is used as feedback to modify weights to the local and received global predictions for a next message.

Distributed spam filtering utilizing a plurality of global classifiers and a local classifier

A system and method for personalized search ranking may use a user's feedback to immediately reorder search results for this particular user so as to improve click-through rate. Upon receiving a query including one or more words, a search engine may identify a list of search results and display the search results on a search result page. A machine-learning module may collect information about a user's browsing activities on the result page, update estimates of relevance of the search results, and reorder the search result list to personalize it for the user.

Personalized web search ranking

Growing community of inventors

Scarsdale, NY, United States of America

John C Langford