Abstract: A method for training a weight-sharing neural network with stochastic architectures is disclosed. The method includes (i) selecting a mini-batch from a plurality of mini-batches, a training data set for a task being grouped into the plurality of mini-batches and each of the plurality of mini-batches comprising a plurality of instances: (ii) stochastically selecting a plurality of network architectures of the neural network for the selected mini-batch; (iii) obtaining a loss for each instance of the selected mini-batch by applying the instance to one of the plurality of network architectures; and (iv) updating shared weights of the neural network based on the loss for each instance of the selected mini-batch.

Abstract: An example apparatus for visual question answering includes a receiver to receive an input image and a question. The apparatus also includes an encoder to encode the input image and the question into a query representation including visual attention features. The apparatus includes a knowledge spotter to retrieve a knowledge entry from a visual knowledge base pre-built on a set of question-answer pairs. The apparatus further includes a joint embedder to jointly embed the visual attention features and the knowledge entry to generate visual-knowledge features. The apparatus also further includes an answer generator to generate an answer based on the query representation and the visual-knowledge features.

Abstract: An example apparatus for visual question answering includes a receiver to receive an input image and a question. The apparatus also includes an encoder to encode the input image and the question into a query representation including visual attention features. The apparatus includes a knowledge spotter to retrieve a knowledge entry from a visual knowledge base pre-built on a set of question-answer pairs. The apparatus further includes a joint embedder to jointly embed the visual attention features and the knowledge entry to generate visual-knowledge features. The apparatus also further includes an answer generator to generate an answer based on the query representation and the visual-knowledge features.