Intelligent Selections | 2018

Museum-based Public Facing Installation Exploring Digital Collections by Leveraging Machine Learning

Intelligent Selections | 2018 - Description

Pete Haughie & Dominic Smith

Intelligent Selections was a joint project between me and artist Dominic Smith for the Tyne and Wear Archives and Museums Great North Museum: Hancock. It was to run from June to September 2018 as a part of the Great Exhibition of the North — a celebration of the heritage and history of arts, culture, and design in the North organised by TWAM and the National Lottery Fund.

We were given unprecedented access to a database containing descriptions of every individual item across the TWAM network and over a period of several weeks gradually trained a Machine Learning algorithm. This algorithm ran constantly in the background for the duration of its installation, eventually creating something in the region of thirty four million new records based on the original source materiél — the longer the installation was in sitú the larger and more varied the output could become. From this pool of possibilities the public could choose one to view via a tablet touchscreen interface where a random selection of nouns — also chosen from the TWAM data — was displayed. When the nouns were matched as closely as possible with textual data from the generated entries the result was displayed on a large back projected screen. These results were also sent to an accompianing website in realtime for later retrieval by the viewer. The piece would eventually be interacted with over thirteen thousand times.

Picture by Dominic Smith

Here I am inspecting the output on the back projection screen

In Memory of Those Yet to Come | 2018

A Museological Machine Learning installation for suggesting new names for the animals yet to become extinct or the creatures that may follow after we have gone

In Memory of Those Yet to Come | 2018 - Description

Using data from the IUCN Redlist containing all the taxonomic information of all known extinct creatures I extracted the English common names and combined it with the highest ranked results from the three most common Amazon product searches — drones, disposable plastic, and plastic toys.


I hesitate to call it music

For a few years now I've been keeping an active Bandcamp account filled with experimental recordings as a form of audio sketchbook.

Each recording I make is an examination of a set of physical properties of a subject or prototypical instrument, either in software or a tangible interface for expression, often in combination.

Mural | 2016 — 2019

A Tool for Creating Visual Longform Stories

Mural is a cross-platform application which attempts to if not solve then mitigate the difficulties of producing long-form rich media content for artists and journalists short on time, skills, or development budgets.

The origins of Mural start on a Serbain Mountaintop in the village of Drvengrad where Douglas Arellanes and I were drinking the local plum brandy and discussing the then recent triumphs of the Guardian and New York Times and their respective stories Fire Storm and Snow Fall - both wildly engaging yet incredibly expensive technical achievments. We decided to try and implement something as a proof of concept based on the now defunct content management system Newscoop, developed by Source Fabric. The theme for the CMS was first shown to the public as part of the Corners of Europe project for ISIS Arts (now D6). After the support for the CMS was dropped by Source Fabric we decided to separate the theme out and create something created specifically for longform content creation.

The first iteration was initially funded by the now suspended Google Digtial News Initiative in 2016 and was brought to a more realised useable state during 2017 whilst I was persuing my Masters at Newcastle University. The platform is still undergoing active development and we are always looking for partners and clients. Mural is open source and free, relies on the simplest server stack for delivery (HTML, JS, CSS), it has off-the-shelf support for YouTube and Vimeo or even local files, chaptered audio with automatic fade-in and out , automatic content advance, and works on all modern browsers - both desktop and mobile.

My own site is built with Mural.