# Improving Differentiable Neural Computers Through Memory Masking, De-allocation, and Link Distribution Sharpness Control

arXiv: https://arxiv.org/abs/1904.10278 · HTML: https://mlsanity.com/paper/1904.10278/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 23 Apr 2019.

## Built on by (most cited first, 9 of 9)

- [Combinatorial optimization and reasoning with graph neural networks](https://mlsanity.com/paper/2102.09544/) (2021)
- [The Devil is in the Detail: Simple Tricks Improve Systematic Generalization of Transformers](https://mlsanity.com/paper/2108.12284/) (2021)
- [The Neural Data Router: Adaptive Control Flow in Transformers Improves Systematic Generalization](https://mlsanity.com/paper/2110.07732/) (2021)
- [Learning to Rehearse in Long Sequence Memorization](https://mlsanity.com/paper/2106.01096/) (2021)
- [Think Before You Act: Decision Transformers with Working Memory](https://mlsanity.com/paper/2305.16338/) (2023)
- [State-Space Constraints Can Improve the Generalisation of the Differentiable Neural Computer to Input Sequences With Unseen Length](https://mlsanity.com/paper/2110.09138/) (2021)
- [Differentiable Neural Computers with Memory Demon](https://mlsanity.com/paper/2211.02987/) (2022)
- [DNCs Require More Planning Steps](https://mlsanity.com/paper/2406.02187/) (2024)
- [SSAKG 2.0: An Open-Source Package for Structural Associative Sequence Memory and Context-Based Retrieval](https://mlsanity.com/paper/2609.01849/) (2026)

## Abstract (arXiv)

The Differentiable Neural Computer (DNC) can learn algorithmic and question answering tasks. An analysis of its internal activation patterns reveals three problems: Most importantly, the lack of key-value separation makes the address distribution resulting from content-based look-up noisy and flat, since the value influences the score calculation, although only the key should. Second, DNC's de-allocation of memory results in aliasing, which is a problem for content-based look-up. Thirdly, chaining memory reads with the temporal linkage matrix exponentially degrades the quality of the address distribution. Our proposed fixes of these problems yield improved performance on arithmetic tasks, and also improve the mean error rate on the bAbI question answering dataset by 43%.