Yahoo! Inc.

Oath Inc.

Yahoo Assets LLC

Verizon Media Inc.

Yahoo Holdings, Inc.

Jia Li

Xiangnan Kong

Bin Ni

Nadav Golbandi

Jenhao Hsiao

Jen-Hao Hsiao

XianXing Zhang

Yi Chang

Kuang-chih Lee

Suleyman Cetintas

Mohammad Saberian

Lyndon Kennedy

Nan Liu

Jian Yang

Haojian Jin

Sachin Sudhakar Farfade

Ioannis Kalantidis

Joseph Nathaniel Kaye

Briefly, embodiments of methods and/or systems of generating preference indices for contiguous portions of digital images are disclosed. For one embodiment, as an example, parameters of a neural network may be developed to generate object labels for digital images. The developed parameters may be transferred to a neural network utilized to generate signal sample value levels corresponding to preference indices for contiguous portions of digital images.

Generating preference indices for image content

At least one label prediction model is trained, or learned, using training data that may comprise training instances that may be missing one or more labels. The at least one label prediction model may be used in identifying a content item's ground-truth label set comprising an indicator for each label in the label set indicating whether or not the label is applicable to the content item.

System and method for large-scale multi-label learning using incomplete label assignments

Disclosed are systems and methods for improving interactions with and between computers in a search system supported by or configured with search servers or platforms. The systems interact to identify and retrieve data across platforms, which data can be used to improve the quality of results data used in processing interactions between or among processors in such systems. The disclosed systems and methods provide a Deep Fast Search (DFS) that improves content search accuracy that executes independent of the search database, while achieving an increased content retrieval speed. The disclosed systems and methods employ two complementary deep feature searches: 1) a coarse deep feature search and a 2) fine deep feature search. Thus, the disclosed systems and methods employ a coarse-to-fine strategy that embodies the efficiency and cost effectiveness of the coarse deep feature search and the accuracy of the fine deep feature search.

System and method for improved server performance for a deep feature based coarse-to-fine fast search

Example methods, apparatuses, and/or articles of manufacture are disclosed that may be implemented, in whole or in part, using one or more computing devices to facilitate or support one or more processes and/or operations for one or more on-line recommendations, such as product-related recommendations, for example.

On-line product related recommendations

Personalized criteria-based media organization is provided. Digital media items are organized into one or more albums. At least one album is associated with album membership criteria and includes only digital media items that satisfy the album membership criteria. An album is updated by receiving user input identifying at least one criteria creation element, generating user-defined album membership criteria based on the at least one criteria creation element, identifying any matching digital media items that satisfy the user-defined album membership criteria, and associating the matching digital media items with the album.

Personalized criteria-based media organization

An approach for performing mobile visual search uses deep variant coding of images to reduce the amount of data transmitted from mobile devices to a search server and to provide more efficient indexing and searching on the search server. The amount of data used to represent an image varies depending upon the content of the image and is less than conventional fixed bit length hashing approaches. Denser regions of a feature space are represented by more encoding bits and sparser regions of the feature space are represented by fewer encoding bits, so that the overall number of encoding bits for an image feature is reduced. The approach generally involves determining a set of hash functions that provide deep hashing with more evenly-distributed hash buckets. One or more additional hash functions may be selectively generated for particular hash buckets that contain more than a specified number of images.

Mobile visual search using deep variant coding

Generating notifications comprising text and image data for client devices with limited display screens is disclosed. An image to be included in the notification is resized and reshaped using image processing techniques. The resized image is further analyzed to identify optimal portions for placing the text data. The text data can also be analyzed and shortened for including at the identified portion of resized image to generate a notification. The resulting notification displays the text and image data optimally within the limited screen space of the client device so that a user observing the notification can obtain the information at a glance.

Automatically creating at-a-glance content

Disclosed herein is item recommender that uses a model trained using a combination of at least visual item similarity training data and social activity training data. The model may be used, for example, to identify a set of recommended products having similar visual features as a given product. The set of recommended products may be presented to the user along with the given product. The model may be continuously updated using feedback from users to identify the features considered to be important to the users relative to other features.

E-commerce recommendation system and method

System, method and architecture for providing improved visual recognition by modeling visual content, semantic content and an implicit social network representing individuals depicted in a collection of content, such as visual images, photographs, etc., which network may be determined based on co-occurrences of individuals represented by the content, and/or other data linking the individuals. In accordance with one or more embodiments, using images as an example, a relationship structure may comprise an implicit structure, or network, determined from co-occurrences of individuals in the images. A kernel jointly modeling content, semantic and social network information may be built and used in automatic image annotation and/or determination of relationships between individuals, for example.

Visual recognition using social links

Disclosed herein is a system and method that facilitate searching and/or browsing of images by clustering, or grouping, the images into a set of image clusters using facets, such as without limitation visual properties or visual characteristics, of the images, and representing each image cluster by a representative image selected for the image cluster. A map-reduce based probabilistic topic model may be used to identify one or more images belonging to each image cluster and update model parameters.

Image-based faceted system and method

Briefly, embodiments of methods and/or systems of detecting and image of a human face in a digital image are disclosed. For one embodiment, as an example, parameters of a neural network may be developed to generate object labels for digital images. The developed parameters may be refined by a neural network to generate signal sample value levels corresponding to probability that a human face may be depicted at a localized region of a digital image.

Face detection

Briefly, embodiments of methods and/or systems of training multiclass convolutional neural networks (CNNs) are disclosed. For one embodiment, as an example, an auxiliary CNN may be utilized to form an ensemble with the collection as a linear combination. The linear combination may be based, at least in part, on boost prediction error encountered during the training process.

Growing community of inventors

Santa Clara, CA, United States of America

Jia Li