Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a generative neural network. One of the methods includes training a generative neural network by performing a sequence of a plurality training stages each generating an expanded training data set. The method also involves performing a sequence of improve steps, each comprising training the generative neural network on the training examples in a corresponding subset of the expanded training data set.

Abstract: Embodiments are directed towards recommending and providing content to a user group for the social consumption of recommended content. Embodiments include generating an individual recommendation for the users based on user preferences of the plurality of users. A group recommendation is generated based on a combination of the individual recommendations. The group recommendation is provided to each of the plurality of users. The group may collaborate over the group recommendation. During collaboration, the group may generate a dialog that includes a discussion based on the group recommendation. Keyword and sentiment pairs, included in the dialog, are determined. The group recommendation is updated based on the keywords and sentiments. The updated group recommendation is provided to the users. In response to detecting a group consensus while monitoring the group collaboration, content associated with the group consensus is provided to the users.

Abstract: Embodiments are directed towards recommending and providing content to a user group for the social consumption of recommended content. Embodiments include generating an individual recommendation for the users based on user preferences of the plurality of users. A group recommendation is generated based on a combination of the individual recommendations. The group recommendation is provided to each of the plurality of users. The group may collaborate over the group recommendation. During collaboration, the group may generate a dialog that includes a discussion based on the group recommendation. Keyword and sentiment pairs, included in the dialog, are determined. The group recommendation is updated based on the keywords and sentiments. The updated group recommendation is provided to the users. In response to detecting a group consensus while monitoring the group collaboration, content associated with the group consensus is provided to the users.