Biology And Machine Learning: A Shared Vision

January 24, 2020
February 10, 2020

As machine learning and artificial intelligence become more widely adopted, with the technology being used for everything from genetics, to smart-homes and social media, it is worth remembering where the inspiration for these powerful algorithms came from. The field was created in an attempt to approximate ‘how human beings learn’, and as the fundamental technologies improve, they may aid in our understanding of their underlying biological processes.

Machine learning is a branch of data science that analyses and learns from data in order to improve the algorithm itself. We call this ‘training’: a period in which the algorithm learns from different datasets before it is ultimately ready to be used for analytical applications. This is much like the human brain. Indeed, the human brain’s ability to learn and iterate is much of the inspiration behind this field. This is true practically as well as conceptually, as the building blocks of machine learning are themselves heavily inspired by the biology of our brains.

A brain is a complex multi-layered network of neural connections, or a biological neural network (BNN). In trying to model and understand the brain, and its associative learning capabilities, scientists have tried to replicate this with a type of algorithm called an artificial neural network (ANN). One of the current best approximations to the jumbled network of neurons that make up the human brain is a convolutional neural network (CNN), itself a complex web of these algorithms.

However, these algorithms cannot match the true complexity of a neuron, or of the human connectome (the brain’s ‘wiring’); it is perhaps a loose inspiration at best. We don’t yet know exactly how the brain works so we can’t come close to building something that works exactly like it. Instead, ANNs (and CNNs) are finely engineered systems, built upon the principles of computer science but containing many elements derived from biology.

These algorithms can then be used to help us discover more about biology itself, for example helping us to learn more about how the brain works. An interesting example of the symbiotic relationship between biology and machine learning can be seen by examining vision….

  • Biological vision inspiring machine learning: the visual stream is comprised of hierarchical areas. An example of this would be that “simple shapes” have a higher hierarchy than “complex shapes”: that is, the brain decides there is a simple shape present before then decoding the complexity of that shape. This is something that CNNs mimic through use of a layered architecture: algorithms are worked through in a specific order, adding complexity at each stage.
  • Machine learning in the study of vision: exploratory machine learning models, such as CNNs, help scientists to understand ways in which the human eye processes information. This is used in research such as predicting where a human’s eye is likely to fixate on an object, and how that information is communicated to the brain.

The biological inspiration behind machine learning is improving research in both fields: as the technology becomes “cleverer” and drives the study of biology (and the wider sciences), the insights gained could themselves influence the computational architectures of the future, allowing us to solve increasingly complex problems.

  • Written and edited by Belle Taylor, Strategic Communications and Partnerships Manager at
This is some text inside of a div block.