Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating descriptions of input images. One of the methods includes obtaining an input image; processing the input image using a first neural network to generate an alternative representation for the input image; and processing the alternative representation for the input image using a second neural network to generate a sequence of a plurality of words in a target natural language that describes the input image.

Abstract: Generating a playlist may include designating a seed track in an audio library; identifying audio tracks in the audio library having constructs that are within a range of a corresponding construct of the seed track, where the constructs for the audio tracks are derived from frequency representations of the audio tracks, and the corresponding construct for the seed track is derived from a frequency representation of the seed track; and generating the playlist using at least some of the audio tracks that were identified.

Abstract: A system, computer readable storage medium, and computer-implemented method presents video search results responsive to a user keyword query. The video hosting system uses a machine learning process to learn a feature-keyword model associating features of media content from a labeled training dataset with keywords descriptive of their content. The system uses the learned model to provide video search results relevant to a keyword query based on features found in the videos. Furthermore, the system determines and presents one or more thumbnail images representative of the video using the learned model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for classifying images.

Abstract: Generating a playlist may include designating a seed track in an audio library; identifying audio tracks in the audio library having constructs that are within a range of a corresponding construct of the seed track, where the constructs for the audio tracks are derived from frequency representations of the audio tracks, and the corresponding construct for the seed track is derived from a frequency representation of the seed track; and generating the playlist using at least some of the audio tracks that were identified.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating descriptions of input images. One of the methods includes obtaining an input image; processing the input image using a first neural network to generate an alternative representation for the input image; and processing the alternative representation for the input image using a second neural network to generate a sequence of a plurality of words in a target natural language that describes the input image.

Abstract: This document generally describes systems, methods, devices, and other techniques related to speaker verification, including (i) training a neural network for a speaker verification model, (ii) enrolling users at a client device, and (iii) verifying identities of users based on characteristics of the users' voices. Some implementations include a computer-implemented method. The method can include receiving, at a computing device, data that characterizes an utterance of a user of the computing device. A speaker representation can be generated, at the computing device, for the utterance using a neural network on the computing device. The neural network can be trained based on a plurality of training samples that each: (i) include data that characterizes a first utterance and data that characterizes one or more second utterances, and (ii) are labeled as a matching speakers sample or a non-matching speakers sample.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating descriptions of input images. One of the methods includes obtaining an input image; processing the input image using a first neural network to generate an alternative representation for the input image; and processing the alternative representation for the input image using a second neural network to generate a sequence of a plurality of words in a target natural language that describes the input image.

Abstract: Methods, systems and apparatus for refining image annotations. In one aspect, a method includes receiving, for each image in a set of images, a corresponding set of labels determined to be indicative of subject matter of the image. For each label, one or more confidence values are determined. Each confidence value is a measure of confidence that the label accurately describes the subject matter of a threshold number of respective images to which it corresponds. Labels for which each of the one or more confidence values meets a respective confidence threshold are identified as high confidence labels. For each image in the set of images, labels in its corresponding set of labels that are high confidence labels are identified. Images having a corresponding set of labels that include at least a respective threshold number of high confidence labels are identified as high confidence images.

Abstract: Systems and techniques are disclosed for labeling objects within an image. The objects may be labeled by selecting an option from a plurality of options such that each option is a potential label for the object. An option may have an option score associated with. Additionally, a relation score may be calculated for a first option and a second option corresponding to a second object in an image. The relation score may be based on a frequency, probability, or observance corresponding to the co-occurrence of text associated with the first option and the second option in a text corpus such as the World Wide Web. An option may be selected as a label for an object based on a global score calculated based at least on an option score and relation score associated with the option.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a model using parameter server shards. One of the methods includes receiving, at a parameter server shard configured to maintain values of a disjoint partition of the parameters of the model, a succession of respective requests for parameter values from each of a plurality of replicas of the model; in response to each request, downloading a current value of each requested parameter to the replica from which the request was received; receiving a succession of uploads, each upload including respective delta values for each of the parameters in the partition maintained by the shard; and updating values of the parameters in the partition maintained by the parameter server shard repeatedly based on the uploads of delta values to generate current parameter values.

Abstract: Systems and techniques for labeling objects within an image. The objects may be labeled by selecting an option from a plurality of options such that each option is a potential label for the object. An option may have an option score associated with. Additionally, a relation score may be calculated for a first option and a second option corresponding to a second object in an image. The relation score may be based on a frequency, probability, or observance corresponding to the co-occurrence of text associated with the first option and the second option in a text corpus such as the World Wide Web. An option may be selected as a label for an object based on a global score calculated based at least on an option score and relation score associated with the option.

Abstract: This document generally describes systems, methods, devices, and other techniques related to speaker verification, including (i) training a neural network for a speaker verification model, (ii) enrolling users at a client device, and (iii) verifying identities of users based on characteristics of the users' voices. Some implementations include a computer-implemented method. The method can include receiving, at a computing device, data that characterizes an utterance of a user of the computing device. A speaker representation can be generated, at the computing device, for the utterance using a neural network on the computing device. The neural network can be trained based on a plurality of training samples that each: (i) include data that characterizes a first utterance and data that characterizes one or more second utterances, and (ii) are labeled as a matching speakers sample or a non-matching speakers sample.

Abstract: Methods, systems and apparatus for refining image relevance models. In general, one aspect includes receiving a trained image relevance model that generates relevance measures of content feature values of images to a query, identifying a first threshold number of common content feature values for the set of training images, the common content feature values being identified as a set of content feature values that are each shared by at least a portion of the training images, identifying a subset of the set of training images having a quantity of the common content feature values greater than a second threshold number of content features values, and generating a re-trained image relevance model based on content feature values of the set of training images, wherein content feature values of the subset of training images are weighted higher than content feature values of the training images not in the subset.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a model using parameter server shards. One of the methods includes receiving, at a parameter server shard configured to maintain values of a disjoint partition of the parameters of the model, a succession of respective requests for parameter values from each of a plurality of replicas of the model; in response to each request, downloading a current value of each requested parameter to the replica from which the request was received; receiving a succession of uploads, each upload including respective delta values for each of the parameters in the partition maintained by the shard; and updating values of the parameters in the partition maintained by the parameter server shard repeatedly based on the uploads of delta values to generate current parameter values.

Abstract: A method, system, and computer readable storage medium is provided for identifying textual terms in response to a visual query is provided. A server system receives a visual query from a client system. The visual query is responded to as follows. A set of image feature values for the visual query is generated. The set of image feature values is mapped to a plurality of textual terms, including a weight for each of the textual terms in the plurality of textual terms. The textual terms are ranked in accordance with the weights of the textual terms. Then, in accordance with the ranking the textual terms, one or more of the ranked textual terms are sent to the client system.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating descriptions of input images. One of the methods includes obtaining an input image; processing the input image using a first neural network to generate an alternative representation for the input image; and processing the alternative representation for the input image using a second neural network to generate a sequence of a plurality of words in a target natural language that describes the input image.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a model using parameter server shards. One of the methods includes receiving, at a parameter server shard configured to maintain values of a disjoint partition of the parameters of the model, a succession of respective requests for parameter values from each of a plurality of replicas of the model; in response to each request, downloading a current value of each requested parameter to the replica from which the request was received; receiving a succession of uploads, each upload including respective delta values for each of the parameters in the partition maintained by the shard; and updating values of the parameters in the partition maintained by the parameter server shard repeatedly based on the uploads of delta values to generate current parameter values.

Abstract: An image classification system trains an image classification model to classify images relative to text appearing with the images. Training images are iteratively selected and classified by the image classification model according to feature vectors of the training images. An independent model is trained for unique n-grams of text. The image classification system obtains text appearing with an image and parses the text into candidate labels for the image. The image classification system determines whether an image classification model has been trained for the candidate labels. When an image classification model corresponding to a candidate label has been trained, the image classification subsystem classifies the image relative to the candidate label. The image is labeled based on candidate labels for which the image is classified as a positive image.

Abstract: Methods, systems and apparatus for refining image relevance models. In general, one aspect of the subject matter described in this specification can be implemented in methods that include re-training an image relevance model by generating a first re-trained model based on content feature values of first images of a first portion of training images in a set of training images, receiving, from the first re-trained model, image relevance scores for second images of a second portion of the set of training images, removing, from the set of training images, some of the second images identified as outlier images for which the image relevance score received from the first re-trained model is below a threshold score, and generating a second re-trained model based on content feature values of the first images of the first portion and the second images of the second portion that remain following removal of the outlier images.