Exhibitor presenting an AI music composer at the Science Museum

Machine learning – the form of artificial intelligence which allows machines to learn from data to make models, predictions or recommendations – is being put to use in an increasing range of everyday applications. During an event at the Science Museum on 31 August, the Royal Society invited several companies and research groups to showcase what they could create with this technology, and also introduced an interactive infographic explaining machine learning.

Here’s a quick overview of what we saw that machine learning can do during the event:

Making sense of words

Context Scout presented an advanced search engine, which uses natural language processing – a type of machine learning – to improve the accuracy or usefulness or your search results. Context Scout’s ‘Find Engine’ looks not only at the weblinks returned by a search, but also the content of webpages and the connections between them which other users have made most frequently, using a technique called reinforcement learning. This tool can be particularly helpful for recruiters, or others searching the web for contacts in particular fields.

Researchers from a cognitive computing team at Queen Mary University of London also showed how natural language processing can analyse our words. Their application composes poems based on keywords of your choice: using these key words and an online database, their machine learning system identifies relationships between words and uses this to compose original poems.

Creating new products

During the event, Museum-goers could make and listen to music from an AI composer – Jukedeck – that can create new soundtracks, based on the genre, mood and tempo of your choice, within 20 seconds. Each track can be edited to fit the user’s needs and is royalty-free. This is especially popular with YouTube users and other video creators, and could also potentially be used with fitness trackers to create music to match your pace as you work out. To do that, Jukedeck uses a different type of learning: it uses artificial neural networks, taking lots of examples of music and learning from these to build models of musical genres and moods.

The Museum also hosted an AI beer tasting session, which proved very popular with the Lates’ adult crowd. IntelligentX, showed their Automated Brewing Intelligence system learns from online feedback forms completed by consumers and, using statistical machine learning, comes up with a new beer recipe to fit better with their taste. In future, such learning could be used to personalise products ranging from coffee and chocolate to perfume.

Want to find out more?

The Royal Society will be supporting further public events as our machine learning project continues. For example, we’ll be at New Scientist Live on 23 September, discussing how and when machines will be able to explain themselves. Further details about this, and other forthcoming events, are on our events page.