What if machine learning meant that you didn't have to have a definitive starting point and the reams of records in the archives could be explored and enjoyed visually?
That is the vision of Martin Disley who has been creating datasets from across the National Library of Scotland's (NLS) map collection.
His project, which is part of the Creative Informatics Resident Entrepreneur project at the University of Edinburgh, curated datasets of images previously scanned by the NLS to feed a machine learning model.The newly-created machine learning model then creates 'fake' versions of the images that it is trained upon.
The generated output from this process can be animated to produce visions of machines dreaming, in this case the fake maps animated and brought to life. This has the effect of synthesising these large collections down in short videos.
Fake maps and towns can be created from the model and then animated.
When the animation starts with a small town and ends with a large developed town, the viewer can watch the town grow in an organic manner as the model has been trained on how towns of every size grow over time.
He said: "People can view thousands of images online but this can quickly become overwhelming and it is a struggle to get people to get people to engage with the content.
“We are working on a tool that will allow users to interact with the model, to be able to control what it produces.
The technology which drives Martin's machine learning model is based on the GAN machine learning architecture which gained national attention when it was used to create the website thispersondoesnotexist.com.
Over 70,000 facial images from Flickr were used to train the model meaning it was able to learn patterns in human face composition and then create new faces.
Martin said: "If you consider maps, you are already starting with a fake. It is a pictorial representation of reality.
The models I have made have learnt the grammar of these maps; you can read these fake maps like you can read any of the originals. You are able to build an internal representation of the map in your head; you can imagine what these places might look like.”
Martin said the process of creating his model was one of fine tuning having started with a large dataset and then whittling out the images of maps that were creating bad results.
"When you are training the model you get to see the dataset in motion.
"As I go through the dataset I take out what I don't like and pick points that are producing interesting results.
"The National Library are excited about the potential for the increased public engagement that synthesising these overwhelmingly large collections into visually exciting media might bring”
Martin Disley is a participant in Creative Informatics Resident Entrepreneur project which delivered by the University of Edinburgh in partnership with Edinburgh Napier University, CodeBase and Creative Edinburgh and is one of nine programmes across the UK that make up the Creative Industries Clusters Programme, funded by the Arts and Humanities Research Council as part of the UK Government’s Industrial strategy. Creative Informatics is part of the Edinburgh and South East Scotland City Region Deal initiative (DDI Programme) and is also supported by the Scottish Funding Council.