Hybrid computing using a neural network with dynamic external memory
Artificial neural networks are remarkably adept at sensory processing, sequence learning and reinforcement learning, but are limited in their ability to represent variables and data structures and to store data over long timescales, owing to the lack of an external memory. Here we introduce a machin...
Saved in:
| Published in: | Nature (London) Vol. 538; no. 7626; pp. 471 - 476 |
|---|---|
| Main Authors: | , , , , , , , , , , , , , , , , , , , |
| Format: | Journal Article |
| Language: | English |
| Published: |
London
Nature Publishing Group UK
27.10.2016
Nature Publishing Group |
| Subjects: | |
| ISSN: | 0028-0836, 1476-4687, 1476-4687 |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Artificial neural networks are remarkably adept at sensory processing, sequence learning and reinforcement learning, but are limited in their ability to represent variables and data structures and to store data over long timescales, owing to the lack of an external memory. Here we introduce a machine learning model called a differentiable neural computer (DNC), which consists of a neural network that can read from and write to an external memory matrix, analogous to the random-access memory in a conventional computer. Like a conventional computer, it can use its memory to represent and manipulate complex data structures, but, like a neural network, it can learn to do so from data. When trained with supervised learning, we demonstrate that a DNC can successfully answer synthetic questions designed to emulate reasoning and inference problems in natural language. We show that it can learn tasks such as finding the shortest path between specified points and inferring the missing links in randomly generated graphs, and then generalize these tasks to specific graphs such as transport networks and family trees. When trained with reinforcement learning, a DNC can complete a moving blocks puzzle in which changing goals are specified by sequences of symbols. Taken together, our results demonstrate that DNCs have the capacity to solve complex, structured tasks that are inaccessible to neural networks without external read–write memory.
A ‘differentiable neural computer’ is introduced that combines the learning capabilities of a neural network with an external memory analogous to the random-access memory in a conventional computer.
A neural network/computer program hybrid
Conventional computer algorithms can process extremely large and complex data structures such as the worldwide web or social networks, but they must be programmed manually by humans. Neural networks can learn from examples to recognize complex patterns, but they cannot easily parse and organize complex data structures. Now Alex Graves, Greg Wayne and colleagues have developed a hybrid learning machine, called a differentiable neural computer (DNC), that is composed of a neural network that can read from and write to an external memory structure analogous to the random-access memory in a conventional computer. The DNC can thus learn to plan routes on the London Underground, and to achieve goals in a block puzzle, merely by trial and error—without prior knowledge or ad hoc programming for such tasks. |
|---|---|
| Bibliography: | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 content type line 23 |
| ISSN: | 0028-0836 1476-4687 1476-4687 |
| DOI: | 10.1038/nature20101 |