# Differentiable Adaptive Computation Time for Visual Reasoning

arXiv: https://arxiv.org/abs/2004.12770 · HTML: https://mlsanity.com/paper/2004.12770/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 27 Apr 2020.

## Built on by (most cited first, 12 of 12)

- [Can You Learn an Algorithm? Generalizing from Easy to Hard Problems with Recurrent Networks](https://mlsanity.com/paper/2106.04537/) (2021)
- [Attention, please! A survey of Neural Attention Models in Deep Learning](https://mlsanity.com/paper/2103.16775/) (2021)
- [Path Independent Equilibrium Models Can Better Exploit Test-Time Computation](https://mlsanity.com/paper/2211.09961/) (2022)
- [End-to-end Algorithm Synthesis with Recurrent Networks: Logical Extrapolation Without Overthinking](https://mlsanity.com/paper/2202.05826/) (2022)
- [A survey on VQA_Datasets and Approaches](https://mlsanity.com/paper/2105.00421/) (2021)
- [DACT-BERT: Differentiable Adaptive Computation Time for an Efficient BERT Inference](https://mlsanity.com/paper/2109.11745/) (2021)
- [Adaptive recurrent vision performs zero-shot computation scaling to unseen difficulty levels](https://mlsanity.com/paper/2311.06964/) (2023)
- [Efficient Encoders for Streaming Sequence Tagging](https://mlsanity.com/paper/2301.09244/) (2023)
- [Learning to Stop Overthinking at Test Time](https://mlsanity.com/paper/2502.10954/) (2025)
- [Understanding Complexity in VideoQA via Visual Program Generation](https://mlsanity.com/paper/2505.13429/) (2025)
- [When in Doubt, Think Slow: Iterative Reasoning with Latent Imagination](https://mlsanity.com/paper/2402.15283/) (2024)
- [DART: Distributional Adversarial Recurrent Training for Algorithm Learning](https://mlsanity.com/paper/2609.05988/) (2026)

## Abstract (arXiv)

This paper presents a novel attention-based algorithm for achieving adaptive computation called DACT, which, unlike existing ones, is end-to-end differentiable. Our method can be used in conjunction with many networks; in particular, we study its application to the widely known MAC architecture, obtaining a significant reduction in the number of recurrent steps needed to achieve similar accuracies, therefore improving its performance to computation ratio. Furthermore, we show that by increasing the maximum number of steps used, we surpass the accuracy of even our best non-adaptive MAC in the CLEVR dataset, demonstrating that our approach is able to control the number of steps without significant loss of performance. Additional advantages provided by our approach include considerably improving interpretability by discarding useless steps and providing more insights into the underlying reasoning process. Finally, we present adaptive computation as an equivalent to an ensemble of models, similar to a mixture of expert formulation. Both the code and the configuration files for our experiments are made available to support further research in this area.