Abstract: A testing environment in which offline simulations can be run to identify policies and/or prediction models that result in more valuable content being included in content pages is described herein. For example, the offline simulations can be run in an application executed by an experiment device using data gathered by a production content delivery system. The simulation application can test any number of different policies and/or prediction models without impacting users that use a production content delivery system to request content.

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: Technologies for optimized selection of content for delivery to a user that both optimizes the expected return from the delivery of the content to the user and that enables exploration of delivery of new content to users are disclosed. Content is selected for delivery to a user based on an exploitation score that defines an estimate of the feedback expected from the delivery of the content to the user and an exploration score that varies inversely with the number of times that the content has been transmitted to all users. The use of the exploration score enables the exploration of delivery of new content to users. The content might be delivered via e-mail messages, a web site, or using another mechanism.

Abstract: Described are techniques for storing and retrieving items using a robotic device for moving items. Any combinations of image data depicting a manipulator interacting with an item, sensor data from sensors instrumenting the manipulator or item, item data regarding characteristics of the item, and constraint data relating to characteristics of the robotic device may be used to generate one or more configurations for the robotic device. The configurations may include points of contact and force vectors for contacting the item using the robotic device.

Abstract: A system that verifies the attributes included in the description of an item using artificial intelligence is provided. For example, the system may use a feature extractor to identify color, shape, and/or texture features of a provided image. The system may then use a linear classifier to process the extracted features to identify attributes of the item depicted in the image. The system may compare the identified attributes with the attributes listed in the item's description. If there are any discrepancies, the system may revise the item description to include the identified attributes or provide suggested revisions to a user based on the identified attributes.

Abstract: Described are techniques for storing and retrieving items using a robotic manipulator. Images depicting a human interacting with an item, sensor data from sensors instrumenting the human or item, data regarding physical characteristics of the item, and constraint data relating to the robotic manipulator or the item may be used to generate one or more configurations for the robotic manipulator. The configurations may include points of contact and force vectors for contacting the item using the robotic manipulator.

Abstract: In multi-view learning, optimized prediction matrices are determined for V?2 views of n objects, and a prediction of a view of an object is generated based on the optimized prediction matrix for that view. An objective is optimized, wherein is a set of parameters including at least the V prediction matrices and a concatenated matrix comprising a concatenation of the prediction matrices, and comprises a sum including at least a loss function for each view, a trace norm of the prediction matrix for each view, and a trace norm of the concatenated matrix. may further include a sparse matrix for each view, with further including an element-wise norm of the sparse matrix for each view. may further include regularization parameters scaling the trace norms of the prediction matrices and the trace norm of the concatenated matrix.

Abstract: Described are techniques for storing and retrieving items using a robotic manipulator. Images depicting a human interacting with an item, sensor data from sensors instrumenting the human or item, data regarding physical characteristics of the item, and constraint data relating to the robotic manipulator may be used to generate one or more configurations for the robotic manipulator. Points of contact and force vectors of the configurations may correspond to the points of contact and force vectors determined from the images and sensor data.

Abstract: In multi-view learning, optimized prediction matrices are determined for V?2 views of n objects, and a prediction of a view of an object is generated based on the optimized prediction matrix for that view. An objective is optimized, wherein is a set of parameters including at least the V prediction matrices and a concatenated matrix comprising a concatenation of the prediction matrices, and comprises a sum including at least a loss function for each view, a trace norm of the prediction matrix for each view, and a trace norm of the concatenated matrix. may further include a sparse matrix for each view, with further including an element-wise norm of the sparse matrix for each view. may further include regularization parameters scaling the trace norms of the prediction matrices and the trace norm of the concatenated matrix.