Abstract: Techniques described herein include systems and methods for evaluating an unsupervised machine learning model. In some embodiments, the system identifies item-to-item similarity values based on historical transaction data. The system may also generate collection data for a number of users based on the historical transaction data. Similarity matrices may be created for each pair of users that include rows associated with a first collection and columns associated with a second collection. Each data field in the similarity matrix may indicate an item-to-item similarity value as identified by the system. In some embodiments, a similarity score may be calculated for the user pair based on the item-to-item similarity values included in the similarity matrix. In some embodiments, the system may generate a graphical summary representation of the similarity matrix.

Abstract: A machine learning system builds and uses computer models for identifying or predicting intensity of emotional reactions elicited by a particular video. Such computer models may also determine which particular emotional reaction corresponds to certain times during the video, and whether these reactions are positive or negative for a particular user. The computer models can also predict emotional reactions likely to be elicited by new videos based on learned correlations between video features and elicited emotional reactions.

Abstract: Systems and methods are provided for generating social networking recommendations. A color preference of a first user may be determined from a color palette of a first image associated with the user and/or a color palette of an item associated with the user. Other users may be identified that have a similar color preference as the first user based at least in part on the determined color preference of the first user. Interactions between the first user and one or more other users having similar color preferences with respect to the first user may be facilitated. A social networking recommendation may be generated with respect to the one or more other users having similar color preferences with respect to the first user.

Abstract: A machine learning system builds and uses computer models for identifying or predicting intensity of emotional reactions elicited by a particular video. Such computer models may also determine which particular emotional reaction corresponds to certain times during the video, and whether these reactions are positive or negative for a particular user. The computer models can also predict emotional reactions likely to be elicited by new videos based on learned correlations between video features and elicited emotional reactions.

Abstract: Techniques for providing instruction to devices in the network are provided. For example, based at least in part on the list of currently active devices and the command, a controller can determine a sequence of IR signals associated with at least one currently active device and transmit, utilizing an IR transmitter incorporated with the controller, the sequence of IR signals to one or more of the currently active devices. In another example, an audio command and location information can be received and used to determine where a controller should transmit an IR signal that corresponds with the audio command, based at least in part on the location information.

Abstract: A system uses a trained classifier to identify or predict the item attributes of an item depicted in an image, and compares these attributes to those specified in a corresponding item description. The system may, for example, be used to verify the accuracy of listings submitted by users to an electronic catalog. For example, if an item description submitted by a user does not specify all of the item attributes identified from the item image(s) submitted by the user, the system may generate a suggested edit to the item description.

Abstract: Features are provided for the distributed moderation of content annotations and the dynamic display of the moderated annotations. The annotations can be selectively distributed to subsets of user devices currently viewing a content item, thereby limiting the number of annotations that each user device receives. By selectively distributing only subsets of new annotations, users can give more attention to individual annotations and provide meaningful moderation information (e.g., ratings, “likes,” responses, etc.) regarding individual annotations. The annotations can be displayed in a manner that emphasizes the more highly-rated annotations and/or commonly-submitted annotations over other annotations. The annotations can also be displayed in a manner that provides visual cues regarding the sources of individual annotations.

Abstract: Techniques for curating audio and IR commands using machine learning may be provided. For example, the system can receive an audio stream that includes a plurality of audio segments provided by a particular user (e.g., as determined through acoustic fingerprinting) and the system can store the audio stream and/or segments in the user's profile. The system can also store a command associated with the audio segment(s). When a portion of the audio is provided in conjunction with a same command over a threshold number of repetitions, the next time that that audio segment is received, the system may provide the command that corresponds with that audio segment to an output device to cause an operation of the output device.

Abstract: Systems and methods are provided for associating colors or color names with a color image based on hue. A set of hue values can be extracted from color data, such as a color palette, corresponding to a color image. A representative hue can be derived from the set of hue values. Colors included in the color data that are attributable to the representative hue can be identified and one or more colors can be derived therefrom as representative of a subject depicted by the color image. Color names matching the representative colors can be identified and associated with the color image. In particular, a common color name can be determined therefrom and associated with the subject of the color image.

Abstract: An operating area for a virtual reality system may be defined based on the positions of sensors (e.g., infrared sensors) or fiducial markings within an environment where the virtual reality system is to be operated. The sensors or the fiducial markings may be provided on an operating surface in the form of a carpet, a mat or another like floor covering. When the virtual reality system is to be calibrated prior to use, positions of the sensors or the fiducial markings may be sensed by a base station, a headset or another virtual reality system unit, and an operating area may be defined based on virtual boundaries constructed using such positions.

Abstract: Systems and methods are described herein to determine data associated with affiliated color palettes identified from keyword searches of color palettes. Color palettes may be searched by name or other data associated with the color palettes. Affiliated color palettes may be determined based at least in part on an input color. Furthermore, affiliated colors can be determined based at least in part on votes and/or rankings. The items and/or images associated with affiliated color palettes may be identified. Various user interfaces may be based at least in part on the keyword searches of color palettes and/or determination of affiliated color palettes.

Abstract: Systems and methods are provided for generating a blended reality view to a user, the blended reality view combining images reflected by a mirror with images transmitted from a screen behind the mirror. Systems for generating blended reality views can include a display device with a screen positioned behind a mirror. The display device can generate a pattern of illumination and non-illumination on the screen so that the illuminated portions of the screen substantially transmit through the mirror. Projectors can be used to illuminate objects in front of the mirror so that the illuminated objects are reflected by the mirror. In combination, the portions of the screen transmitted through the mirror and the illuminated objects reflected by the mirror can provide a blended reality view to a user viewing the mirror.

Abstract: Techniques described herein include systems and methods for enhancing content consumption experiences by synchronizing one or more devices to add immersive environmental effects to an area. In embodiments, a database of device capabilities and device identifiers for a plurality of devices associated with a location may be maintained. Information may be obtained that identifies that a particular portion of content has been consumed. Metadata associated with the particular portion of content may be identified based at least in part on the information and the database. The metadata may comprise definitions for providing environmental stimuli to the location where the environmental stimuli may correspond to the particular portion of content. Environmental stimuli may be presented within the location by invoking application programming interface calls for a portion of devices of the plurality of devices based on identifying the metadata and the device identifiers for the portion of devices.

Abstract: A dynamic content delivery service inserts users into groups based on when they access the service and/or known user interests, and the group provides reactions to videos or other content served to the group. Action rules determine what happens to the group as they interact with the service. For example, based on group reactions at specific time points, some or all users in the group can be switched to a new video or multiple different new videos. User reactions to content can be used to determine where to route the user when action rules indicate they should be redirected.

Abstract: Techniques for providing instruction to devices in the network are provided. For example, based at least in part on the list of currently active devices and the command, a controller can determine a sequence of IR signals associated with at least one currently active device and transmit, utilizing an IR transmitter incorporated with the controller, the sequence of IR signals to one or more of the currently active devices. In another example, an audio command and location information can be received and used to determine where a controller should transmit an IR signal that corresponds with the audio command, based at least in part on the location information.

Abstract: Systems and methods are described that facilitate searches based at least in part on a color name. Given a reference color name, one or more additional color names that are associated with the reference color name can be identified. Colors corresponding to these color names can be determined, which in turn serve as basis for determining one or more search colors for the reference color name in a search context. The system may further identify images that contain the determined search colors. The identified images and associated metadata can be analyzed, sorted and provided as a list of items.

Abstract: An operating area for a virtual reality system may be defined based on the positions of sensors (e.g., infrared sensors) or fiducial markings within an environment where the virtual reality system is to be operated. The sensors or the fiducial markings may be provided on an operating surface in the form of a carpet, a mat or another like floor covering. When the virtual reality system is to be calibrated prior to use, positions of the sensors or the fiducial markings may be sensed by a base station, a headset or another virtual reality system unit, and an operating area may be defined based on virtual boundaries constructed using such positions.

Abstract: Features are disclosed for dynamically generating enhanced content showing an item to simulate a physics property of the item via an interaction that will initiate such interactive features. A physics model may be generated to enhance existing content. To generate the physics model, the system analyzes the content such as to detect a feature of the item (e.g., zipper on a clothing item or button on an electronic device item) shown in the content. Some aspects also consider the access device rendering the enhanced content to ensure the enhanced content is appropriately selected and formatted for presentation via the access device.

Abstract: Systems and methods are provided for generating social networking recommendations. A color preference of a first user may be determined from a color palette of a first image associated with the user and/or a color palette of an item associated with the user. Other users may be identified that have a similar color preference as the first user based at least in part on the determined color preference of the first user. Interactions between the first user and one or more other users having similar color preferences with respect to the first user may be facilitated. A social networking recommendation may be generated with respect to the one or more other users having similar color preferences with respect to the first user.

Abstract: Features are disclosed for dynamically generating a navigation interface for rendering content related to an item. To generate the interface, the system analyzes the content available and organizes the content in an intuitive way that keeps the user oriented as they explore the product. Some aspects also consider the access device rendering the interface to ensure the interface includes content that is appropriately selected and formatted for presentation via the access device.