Poster
Compositional generalization through meta sequence-to-sequence learning
Brenden Lake
East Exhibition Hall B + C #178
Keywords: [ Mem ] [ Algorithms -> Few-Shot Learning; Algorithms -> Meta-Learning; Applications -> Natural Language Processing; Deep Learning ] [ Neuroscience and Cognitive Science ] [ Cognitive Science ]
People can learn a new concept and use it compositionally, understanding how to "blicket twice" after learning how to "blicket." In contrast, powerful sequence-to-sequence (seq2seq) neural networks fail such tests of compositionality, especially when composing new concepts together with existing concepts. In this paper, I show how memory-augmented neural networks can be trained to generalize compositionally through meta seq2seq learning. In this approach, models train on a series of seq2seq problems to acquire the compositional skills needed to solve new seq2seq problems. Meta se2seq learning solves several of the SCAN tests for compositional learning and can learn to apply implicit rules to variables.
Live content is unavailable. Log in and register to view live content