Name
Abstract | ||
|---|---|---|
Recent work has shown that large text-based neural language models acquire a surprising propensity for one-shot learning. Here, we show that an agent situated in a simulated 3D world, and endowed with a novel dual-coding external memory, can exhibit similar one-shot word learning when trained with conventional RL algorithms. After a single introduction to a novel object via visual perception and language (\"This is a dax\"), the agent can manipulate the object as instructed (\"Put the dax on the bed\"), combining short-term, within-episode knowledge of the nonsense word with long-term lexical and motor knowledge. We find that, under certain training conditions and with a particular memory writing mechanism, the agent\u0027s one-shot word-object binding generalizes to novel exemplars within the same ShapeNet category, and is effective in settings with unfamiliar numbers of objects. We further show how dual-coding memory can be exploited as a signal for intrinsic motivation, stimulating the agent to seek names for objects that may be useful later. Together, the results demonstrate that deep neural networks can exploit meta-learning, episodic memory and an explicitly multi-modal environment to account for \u0027fast-mapping\u0027, a fundamental pillar of human cognitive development and a potentially transformative capacity for artificial agents. |
Year | Venue | DocType |
|---|---|---|
2021 | ICLR | Conference |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
6 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Felix Hill | 1 | 346 | 17.90 |
| Olivier Tieleman | 2 | 1 | 2.72 |
| Tamara von Glehn | 3 | 0 | 0.68 |
| Nathaniel Wong | 4 | 0 | 0.68 |
| Merzic Hamza | 5 | 0 | 0.68 |
| Stephen Clark | 6 | 2369 | 162.42 |