The essence of the research, conducted at the University of Toronto, is centered around the question – ‘how well can the fruit fly see?’ And also a related question – ‘do all fruit flies visualize the world in the same way?’ This latter question arises because some researchers think there is evidence of individual recognition and the ability for visual learning in flies. The fruit fly Drosophila melanogaster is a versatile model organism that has been used in biomedical research for over a century.
To explore this, a research study was concocted applying machine learning. The outcome of the study was that individual fruit flies, of the species D. melanogaster, are distinct in terms of their ability to visualize their environment. This is despite the fruit fly having relatively limited optical resolution.
The research also showed that the fruit fly has better vision than many other researchers had previously thought, in that the flies are not solely reliant upon taste and odor in order to navigate a defined space.
The reason for the differences between flies and the surprising ability of D. melanogaster to ‘see’ is based on revelations about the fruit fly’s neuronal architecture, a fact drawn out from the machine learning software. This big data analysis, drawn from many hours of recording fly behavior, revealed that the fly has the ability to extract and encode visual data which enables the fly to re-identify spatial patterns.
Commenting on the research, Graham Taylor, who is a machine learning specialist notes: “A lot of Deep Neural Network applications try to replicate and automate human abilities like facial recognition, natural language processing, or song identification.”
“But rarely do they go beyond human capacity. So it’s exciting to find a problem where algorithms can outperform humans.”
Based on the research, the scientists have concluded that fruit flies exist in a world that is more complex than previously considered. The research also further develops the application of machine learning for biological applications. For example, neuroscientists are using deep learning to examine supervised, unsupervised and reinforcement learning. Furthermore, machine learning has a number of emerging applications in the field of bioinformatics, an area of science that deals with computational and mathematical approaches for understanding and processing biological data.
The research is published in the journal PLoS One, with the research paper titled “Can Drosophila melanogaster tell who’s who?”