Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for controlling agents. In particular, an agent can be controlled using a discretization neural network that generates a state-dependent discretization of an original action space and a policy neural network that is used to select an action from the state-dependent quantization rather than from the original action space.

Abstract: Implementations disclose identifying content appropriate for children algorithmically without human intervention. A method includes identifying, by a processing device, entities corresponding to topics relevant to children, determining, by the processing device, a children's affinity score for each of the identified entities, and selecting, by the processing device, content based on the children's affinity score for the identified entities corresponding to the content.

Abstract: Methods, systems, and media for presenting content organized by category are provided. In some embodiments, the method comprises: receiving a request to present a page indicating content related to a first entity; identifying a first group of entities related to the first entity based at least in part on user interaction information; identifying a group of properties corresponding to the first entity; identifying, for each property, a second group of entities corresponding to the property; calculating, for each property, a score for the property; determining a subset of the properties in the group of properties based on the score of each property; identifying, for each property in the subset of properties, a group of content items corresponding to the property; determining a subset of content items from the group of content items; and causing a user interface indicating the subset of content items to be presented.

Abstract: Methods, systems, and media for presenting content organized by category are provided. In some embodiments, the method comprises: receiving a request to present a page indicating content related to a first entity; identifying a first group of entities related to the first entity based at least in part on user interaction information; identifying a group of properties corresponding to the first entity; identifying, for each property, a second group of entities corresponding to the property; calculating, for each property, a score for the property; determining a subset of the properties in the group of properties based on the score of each property; identifying, for each property in the subset of properties, a group of content items corresponding to the property; determining a subset of content items from the group of content items; and causing a user interface indicating the subset of content items to be presented.

Abstract: Implementations disclose identifying content appropriate for children. A method includes selecting a list of candidate channels from a plurality of channels based on an automatically generated rating of each of the plurality of channels, filtering the list of candidate channels based on feedback received from a first group of reviewers with respect to appropriateness of content of each candidate channel for children, selecting a list of candidate videos based on an automatically generated rating of each video from every channel in the filtered list of candidate channels, and upon receiving, via a graphical user interface, a confirmation that a candidate video from the list of candidate videos is appropriate for children, setting a confirmation indicator of the candidate video to enable presentation of the candidate video to children.

Abstract: Implementations disclose identifying content appropriate for children algorithmically without human intervention. A method includes identifying, by a processing device, entities corresponding to topics relevant to children, determining, by the processing device, a children's affinity score for each of the identified entities, and selecting, by the processing device, content based on the children's affinity score for the identified entities corresponding to the content.