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 search computing system receives an interaction history for one or more respective users describing interactions with one or more items and generates a co-interaction matrix, each value in the co-interaction matrix representing a number of common users determined to have had the interaction with an item of a corresponding row and an item of the corresponding column of the co-interaction matrix where the value is located. The search computing system generates an embeddings matrix comprising an item embedding value for each of the one or more items by applying matrix factorization to the co-interaction matrix and determines, in response to a search query of a particular user, a user embedding value for the searching user based on the interaction history for the searching user. The search computing system determines a similarity between each search result and user interaction history by comparing the user embedding value against each of the item embedding values.

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: 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: An image cropping technique includes detecting, at a computer system including one or more processors, faces within an image and evaluating, at the computer system, the faces to determine if any of the faces correspond to people within a user's online social network. The technique additionally includes defining, at the computer system, a first crop region including a subset of the faces when at least one face within the subset of the faces corresponds to a first person in the user's online social network and cropping, at the computer system, the image based on the first crop region.

Abstract: The present invention relates to methods and systems for automatic image cropping. An automatic image cropping system includes a downsampling system, a saliency analyzer, a spatial scatter, and a face detector configured to detecting one or more faces within the image. The saliency value of each pixel not associated with a detected face within the image is adjusted if one or more faces within the image are detected. A search system defines a saliency region based on a total value of saliency values within the saliency region and on a parameter, where the parameter value is based on a relative importance of a false positive pixel to a false negative pixel. The search system also defines a crop region using an aspect ratio that includes the saliency region and determines a position of the crop region based the saliency values within the crop region by performing a one dimensional search.

Abstract: A method for determining a salient region of an image is disclosed. For a plurality of different saliency cue functions, a single saliency value is calculated for each pixel in a plurality of adjacent pixels in an image using the saliency cue function, wherein one of the saliency cue functions is based on whether the pixel is in a region of the image whose colors contrast with the region's background and another of the saliency cue functions is based on a foreground and background color models of the image. A classifier is used to calculate a combined single saliency value for each pixel based on the single saliency values for the pixel. The salient region of the pixels is determined with a subwindow search based on the combined single saliency values.

Abstract: An image cropping technique includes detecting, at a computer system including one or more processors, faces within an image and evaluating, at the computer system, the faces to determine if any of the faces correspond to people within a user's online social network. The technique additionally includes defining, at the computer system, a first crop region including a subset of the faces when at least one face within the subset of the faces corresponds to a first person in the user's online social network and cropping, at the computer system, the image based on the first crop region.

Abstract: An image cropping technique includes detecting, at a computer system including one or more processors, faces within an image and evaluating, at the computer system, the faces to determine if any of the faces correspond to people within a user's online social network. The technique additionally includes defining, at the computer system, a first crop region including a subset of the faces when at least one face within the subset of the faces corresponds to a first person in the user's online social network and cropping, at the computer system, the image based on the first crop region.

Abstract: A method for determining a salient region of an image is disclosed. For a plurality of different saliency cue functions, a single saliency value is calculated for each pixel in a plurality of adjacent pixels in an image using the saliency cue function, wherein one of the saliency cue functions is based on whether the pixel is in a region of the image whose colors contrast with the region's background and another of the saliency cue functions is based on a foreground and background color models of the image. A classifier is used to calculate a combined single saliency value for each pixel based on the single saliency values for the pixel. The salient region of the pixels is determined with a subwindow search based on the combined single saliency values.

Abstract: Self-terminating prediction trees are a generalization of decision trees in which each node is associated with a real-valued prediction. Instead of having a separate pruning phase, a self-terminating tree may be constructed by applying various limits during tree growth that prevent nodes that add little or no additional decision power from being grown within the tree. The prediction tree is learned by performing a penalized empirical risk minimization task, based upon the use of prediction values and functional tree complexity. A separate pruning phase is not required, since the tree self-terminates further growth.

Abstract: The present invention relates to methods and systems for automatic image cropping. An automatic image cropping system includes a downsampling system, a saliency analyzer, a spatial scatter, and a face detector configured to detecting one or more faces within the image. The saliency value of each pixel not associated with a detected face within the image is adjusted if one or more faces within the image are detected. A search system defines a saliency region based on a total value of saliency values within the saliency region and on a parameter, where the parameter value is based on a relative importance of a false positive pixel to a false negative pixel. The search system also defines a crop region using an aspect ratio that includes the saliency region and determines a position of the crop region based the saliency values within the crop region by performing a one dimensional search.

Abstract: The present invention relates to methods and systems for automatic image cropping. An automatic image cropping system includes a downsampling system configured to downsample an image to a threshold number of pixels, a saliency analyzer configured to compute a saliency value for each pixel based on one or more detected color clusters and a spatial scatter of the one or more color clusters, and a face detector configured to detecting one or more faces within the image. However, the saliency value of each pixel not associated with the one or more detected faces within the image is adjusted if one or more faces within the image are detected. Further, a search system is configured to define a saliency region based on a total value of saliency values within the saliency region and on a parameter, where the parameter value is based on a relative importance of a false positive pixel to a false negative pixel.

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.