Microsoft Technology Licensing, LLC

Micorsoft Corpration

Zhigu Holdings Limited

Nebojsa Jojic

David Earl Heckerman

Brendan John Frey

Hagai Thomas Attias

Patrice Y Simard

Craig J Mundie

Darko Kirovski

Randy Hinrichs

Manuel Jesus Reyes Gomez

Vladimir Jojic

Matthew James Beal

Nemanja Petrovic

Sumit Gulwani

Carl Myers Kadie

Sumit Basu

Paul Cador Roberts

John Randall Hershey

Noah Aaron Zaitlen

Sing Bing Kang

Christopher A Meek

Kentaro Toyama

Charles Lawrence Zitnick, Iii

Ashish Kapoor

Trausti Thor Kristjansson

Jennifer Listgarten

Chris Pal

Manuel Jesus Reyes-Gomez

Trausti Thor Kristajanson

Delbert Dwayne Dueck

Vincent Cheung

Corey Benjamin Moore

Brendan J Frey

Matthew A Turk

Simon Mallal

Jaco Vermaak

Object tracking includes an audio model that receives at least two audio input signals and a video model that receives a video input. The audio model and the video model employ probabilistic generative models which are combined to facilitate object tracking. Expectation maximization can be employed to modify trainable parameters of the audio model and the video model.

Speaker detection and tracking using audiovisual data

Shift invariant predictors are described herein. By way of example, a system for predicting binding information relating to a binding of a protein and a ligand can include a trained binding model and a prediction component. The trained binding model can include a hidden variable representing an unknown alignment of the ligand at a binding site of the protein. The prediction component can be configured to predict the binding information by employing information about the protein's sequence, the ligand's sequence and the trained binding model.

Shift-invariant predictions

The subject invention provides systems and methods that facilitate AIDS vaccine cocktail assembly via machine learning algorithms such as a cost function, a greedy algorithm, an expectation-maximization (EM) algorithm, etc. Such assembly can be utilized to generate vaccine cocktails for species of pathogens that evolve quickly under immune pressure of the host. For example, the systems and methods of the subject invention can be utilized to facilitate design of T cell vaccines for pathogens such HIV. In addition, the systems and methods of the subject invention can be utilized in connection with other applications, such as, for example, sequence alignment, motif discovery, classification, and recombination hot spot detection. The novel techniques described herein can provide for improvements over traditional approaches to designing vaccines by constructing vaccine cocktails with higher epitope coverage, for example, in comparison with cocktails of consensi, tree nodes and random strains from data.

Systems and methods that utilize machine learning algorithms to facilitate assembly of aids vaccine cocktails

Systems and methodologies for efficient vaccine design are disclosed herein. A methodology for efficient vaccine design in accordance with one or more embodiments disclosed herein may be operable to receive a graph having vertices corresponding to epitope sequences present in the pathogen population, weights for respective vertices corresponding to respective frequencies with which corresponding epitope sequences appear in the pathogen population, and directed edges that connect vertices that correspond to overlapping epitope sequences. Such a methodology may also be operable to determine a candidate vaccine sequence of overlapping epitope sequences by identifying a path though the graph corresponding to a series of connected vertices and directed edges that maximizes the total weight of the vertices in the path for a desired vaccine sequence length.

Vaccine design methodology

Cluster models are described herein. By way of example, a system for predicting binding information relating to a binding of a protein and a ligand can include a trained binding model and a prediction component. The trained binding model can include a probability distribution and a hidden variable that represents a cluster of protein sequences, and/or a set of hidden variables representing learned supertypes. The prediction component can be configured to predict the binding information by employing information about the protein's sequence, the ligand's sequence and the trained binding model.

Cluster modeling, and learning cluster specific parameters of an adaptive double threading model

The present invention relates includes system and a method for automatically generating short segments of video (or video 'cliplets') from a larger source video. A cliplet has the properties that its length is determined prior to generation and that the cliplet ideally is semantically meaningful and contains a single short event or theme. Generally, the cliplet generation method processes a large source video and generates cliplet results for presentation (such as to a user). Specifically, the method processes the source video to determine editing points and then extracts cliplets from the source video based on the editing points. The extracted cliplets can overlap in time. Cliplet results then are presented, such as to a user. The cliplet generation system includes a video cliplet generator that processes a large source video and generates cliplets in accordance with the cliplet generation method.

System and method for automatically generating video cliplets from digital video

One embodiment is directed to synthesizing code fragments in a software routine using known inputs and corresponding expected outputs. A computer system provides a software routine with known inputs and corresponding expected outputs, infers software routine instructions based on the known inputs and corresponding expected outputs, and synthesizes a correctly functioning code fragment based on the inferred instructions. Another embodiment is directed to automatically resolving semantic errors in a software routine. A computer system provides the software routine with known inputs and corresponding expected outputs for portions of a program fragment where an error has been localized. The computer system learns a correctly functioning program fragment from pairs of input-output descriptions of the program fragment, determines the program statements that can transform given input states into given output states after execution of those program statements, and alters portions of the software routine with the learned program fragments.

Program synthesis and debugging using machine learning techniques

A system and method that facilitates and effectuates accurate source segmentation of multi-dimensional signals in a computationally efficient manner. By employing Queyranne's algorithm along with a model for combining adjacent multidimensional elements of signal into locally consistent regions, significant improvement in time to identify an optimal segmentation can be achieved. Additional, by saving values computed when executing the algorithm and recalling the values when needed during subsequent iterations of the algorithm provides an additional in algorithm execution speed.

Source segmentation using Q-clustering

Epitope prediction models are described herein. By way of example, a system for predicting epitope information relating to a epitope can include a classification model (e.g., logistic regression model). The trained classification model can illustratively operatively execute one ore logistic functions on received protein data, and incorporate one or more of hidden binary variables and shift variables that when processed represent the identification (e.g., prediction) of one or more desired epitopes. The classification model can be configured to predict the epitope information by processing data including various features of an epitope, MHC, MHC supertype, and Boolean combinations thereof.

T-cell epitope prediction

A system comprising a machine learning classifier trained on a plurality of associations between a host and a pathogen to predict a pathogen characteristic is described herein. The pathogen characteristic can relate to a disease state of the host. Computer-executable instructions for performing a method of forecasting a portion of a target molecule anticipated to influence an organism's condition also are described herein. The method comprises employing population data to automatically analyze one or more areas of the target molecule to determine the portion of the target molecule anticipated to influence the organism's condition. The population data can pertain to at least one relationship between at least one diverse organism trait and the target molecule. One or more epitopes forecast by employing the method also are contemplated.

Association-based predictions of pathogen characteristics

A 'Video Browser' provides interactive browsing of unique events occurring within an overall video recording. In particular, the Video Browser processes the video to generate a set of video sprites representing unique events occurring within the overall period of the video. These unique events include, for example, motion events, security events, or other predefined event types, occurring within all or part of the total period covered by the video. Once the video has been processed to identify the sprites, the sprites are then arranged over a background image extracted from the video to create an interactive static video montage. The interactive video montage illustrates all events occurring within the video in a single static frame. User selection of sprites within the montage causes either playback of a portion of the video in which the selected sprites were identified, or concurrent playback of the selected sprites within a dynamic video montage.

Interactive montages of sprites for indexing and summarizing video

Systems and methodologies for modeling data in accordance with one or more embodiments disclosed herein are operable to receive input data, create data patches from the input data, obtain long-range correlations between the data patches, and model the input data as a patch model based at least in part on the data patches and the long-range correlations. Various learning algorithms are additionally provided for refining the patch model created in accordance with one or more embodiments disclosed herein. Further, systems and methodologies for synthesizing a patch model created in accordance with various aspects of the present invention with a set of test data to perform a transformation represented by the patch model on the test data are provided.

Capturing long-range correlations in patch models

A simplified general model and an associated estimation algorithm is provided for modeling visual data such as a video sequence. Specifically, images or frames in a video sequence are represented as collections of flat moving objects that change their appearance and shape over time, and can occlude each other over time. A statistical generative model is defined for generating such visual data where parameters such as appearance bit maps and noise, shape bit-maps and variability in shape, etc., are known. Further, when unknown, these parameters are estimated from visual data without prior pre-processing by using a maximization algorithm. By parameter estimation and inference in the model, visual data is segmented into components which facilitates sophisticated applications in video or image editing, such as, for example, object removal or insertion, tracking and visual surveillance, video browsing, photo organization, video compositing, etc.

Generative models for constructing panoramas from an image sequence

Adaptive threading models for predicting an interaction between two or more molecules such as proteins are provided. The adaptive threading models have one or more learnable parameters that can be learned from all or some of the available data. The available data can include data relating to known interactions between the two or more molecules, the composition of the molecules and the geometry of the molecular complex.

Molecular interaction predictors

A hierarchical statistical framework for separating mixed data is provided. The approach is a shift from chaining hard-decision modules to an integrated soft-decision approach. The framework facilitates separating, for instance, sequencing data obtained from a mixture of two or more different sequences. The sequencing data can be separated using machine learning techniques to determine the correspondence between the sequencing data and the two or more different sequences.

Growing community of inventors

Redmond, WA, United States of America

Nebojsa Jojic