Neural algorithmic reasoning
📰 The Gradient
Neural algorithmic reasoning combines classical computation with deep neural networks to improve AI's instructive and useful capabilities
Action Steps
- Understand the properties of classical algorithms, such as provable correctness, strong generalization, and interpretability
- Explore how to capture these properties in deep neural networks
- Investigate the application of neural algorithmic reasoning in areas like instructive AI and generally-intelligent agents
Who Needs to Know This
Machine learning researchers and practitioners can benefit from understanding neural algorithmic reasoning to develop more reliable and generalizable AI systems, while software engineers can apply these concepts to improve the performance of AI-powered applications
Key Insight
💡 Capturing classical computation in deep neural networks can improve AI's instructive and useful capabilities
Share This
💡 Neural algorithmic reasoning: combining classical computation with deep learning to create more reliable and generalizable AI #AI #MachineLearning
Key Takeaways
Neural algorithmic reasoning combines classical computation with deep neural networks to improve AI's instructive and useful capabilities
Full Article
Published Time: 2023-10-14T15:30:15.000Z
# Neural algorithmic reasoning
[](https://thegradient.pub/)
* [About](https://thegradient.pub/about/)
* [Articles](https://thegradient.pub/)
* [Newsletter](https://thegradientpub.substack.com/s/the-update)
* [Podcast](https://thegradientpub.substack.com/s/podcast)
* [Subscribe](https://thegradientpub.substack.com/)
* [Write](https://thegradient.pub/write/)
* [Support](https://thegradient.pub/support/)
# Neural algorithmic reasoning
14.Oct.2023 . 24 min read
In this article, we will talk about _classical computation_: the kind of computation typically found in an undergraduate Computer Science course on Algorithms and Data Structures [1]. Think shortest path-finding, sorting, clever ways to break problems down into simpler problems, incredible ways to organise data for efficient retrieval and updates. Of course, given _The Gradient_’s focus on Artificial Intelligence, we will not stop there; we will also investigate how to _capture_ such computation with deep neural networks.
## _Why_ capture classical computation?
Maybe it’s worth starting by clarifying where my vested interest in classical computation comes from. Competitive programming—the art of solving problems by rapidly writing programs that need to terminate in a given amount of time, and within certain memory constraints—was a highly popular activity in my secondary school. For me, it was truly the gateway into Computer Science, and I trust the story is similar for many machine learning practitioners and researchers today. I have been able to win several medals at international programming competitions, such as the Northwestern Europe Regionals of the ACM-ICPC, the top-tier Computer Science competition for university students. Hopefully, my successes in competitive programming also give me some credentials to write about this topic.
While this should make clear why _I_ care about classical computation, why should _we all_ care? To arrive at this answer, let us ponder some of the key properties that classical algorithms have:
* They are **provably correct**, and we can often have strong guarantees about the _resources_ (time or memory) required for the computation to terminate.
* They offer **strong generalisation**: while algorithms are often devised by observing several small-scale example inputs, once implemented, they will work without fault on inputs that are significantly larger, or distributionally different than such examples.
* By design, they are **interpretable**and **compositional**: their (pseudo)code representation makes it much easier to reason about what the computation is actually doing, and one can easily recompose various computations together through subroutines to achieve different capabilities.
Looking at all of these properties taken together, they seem to be exactly the issues that plague modern deep neural networks the most: you can rarely guarantee their accuracy, they often collapse on out-of-distribution inputs, and they are very notorious as black boxes, with compounding errors that can hinder compositionality.
However, it is _exactly_ those skills that are important for making AI _instructive_ and _useful_ to humans! For example, to have an AI system that reliably and instructively teaches a concept to a human, the quality of its output should not depend on minor details of the input, and it should be able to generalise that concept to novel situations. Arguably, these skills are also a missing key step on the road to building generally-intelligent agents. Therefore, if we are able to make any strides towards capturing traits of classical computation in deep neural networks, this is likely to be a very fruitful pursuit.
## First impressions: Algorithmic alignment
My
# Neural algorithmic reasoning
[](https://thegradient.pub/)
* [About](https://thegradient.pub/about/)
* [Articles](https://thegradient.pub/)
* [Newsletter](https://thegradientpub.substack.com/s/the-update)
* [Podcast](https://thegradientpub.substack.com/s/podcast)
* [Subscribe](https://thegradientpub.substack.com/)
* [Write](https://thegradient.pub/write/)
* [Support](https://thegradient.pub/support/)
# Neural algorithmic reasoning
14.Oct.2023 . 24 min read
In this article, we will talk about _classical computation_: the kind of computation typically found in an undergraduate Computer Science course on Algorithms and Data Structures [1]. Think shortest path-finding, sorting, clever ways to break problems down into simpler problems, incredible ways to organise data for efficient retrieval and updates. Of course, given _The Gradient_’s focus on Artificial Intelligence, we will not stop there; we will also investigate how to _capture_ such computation with deep neural networks.
## _Why_ capture classical computation?
Maybe it’s worth starting by clarifying where my vested interest in classical computation comes from. Competitive programming—the art of solving problems by rapidly writing programs that need to terminate in a given amount of time, and within certain memory constraints—was a highly popular activity in my secondary school. For me, it was truly the gateway into Computer Science, and I trust the story is similar for many machine learning practitioners and researchers today. I have been able to win several medals at international programming competitions, such as the Northwestern Europe Regionals of the ACM-ICPC, the top-tier Computer Science competition for university students. Hopefully, my successes in competitive programming also give me some credentials to write about this topic.
While this should make clear why _I_ care about classical computation, why should _we all_ care? To arrive at this answer, let us ponder some of the key properties that classical algorithms have:
* They are **provably correct**, and we can often have strong guarantees about the _resources_ (time or memory) required for the computation to terminate.
* They offer **strong generalisation**: while algorithms are often devised by observing several small-scale example inputs, once implemented, they will work without fault on inputs that are significantly larger, or distributionally different than such examples.
* By design, they are **interpretable**and **compositional**: their (pseudo)code representation makes it much easier to reason about what the computation is actually doing, and one can easily recompose various computations together through subroutines to achieve different capabilities.
Looking at all of these properties taken together, they seem to be exactly the issues that plague modern deep neural networks the most: you can rarely guarantee their accuracy, they often collapse on out-of-distribution inputs, and they are very notorious as black boxes, with compounding errors that can hinder compositionality.
However, it is _exactly_ those skills that are important for making AI _instructive_ and _useful_ to humans! For example, to have an AI system that reliably and instructively teaches a concept to a human, the quality of its output should not depend on minor details of the input, and it should be able to generalise that concept to novel situations. Arguably, these skills are also a missing key step on the road to building generally-intelligent agents. Therefore, if we are able to make any strides towards capturing traits of classical computation in deep neural networks, this is likely to be a very fruitful pursuit.
## First impressions: Algorithmic alignment
My
DeepCamp AI