Six real-life use cases for Google DeepMind’s machine learning systems
DeepMind has attracted mixed headlines since Google paid £400 million for the artificial intelligence (AI) startup in January 2014. When an AI system built by DeepMind won a series of Go matches against the world's best player of the Chinese board game, the possibilities of the technology inspired wonder. But that was tempered by criticisms of its controversial access to personal health records, with concerns raised over data privacy.
From their lab in King's Cross, London, 140 DeepMind researchers develop software systems capable of independent learning through AI algorithms. The company has worked with a small number of organisations to implement its technology so far, but there are already some intriguing real-life examples emerging.
1. DeepMind use cases: The National Grid
Google is in talks with the National Grid to apply its AI frameworks to the UK's energy supply, according to City AM.
The potential partnership could be used to apply DeepMind's technology to make the supply of energy across the UK more efficient. And while talks are said to be in the early stages, both firms are exploring how artificial intelligence and 'smart' systems can be deployed.
"One really interesting possibility is whether we could help the National Grid maximise the use of renewables through using machine learning to predict peaks in demand and supply," a spokesperson for the National Grid told City AM.
2. DeepMind use cases: Royal Free NHS Trust
In February Google launched DeepMind Health with the goal of using its machine learning systems to improve healthcare treatment and digitise medical processes.
The first project announced was a partnership with the Royal Free NHS Trust in north London involving the creation of a mobile app that provides clinicians with cutting-edge analytics. Known as Streams, the app was designed to improve the detection of acute kidney injury by instantly reviewing blood test results before sending an alert to an appropriate clinician via a handheld device.
The data-sharing agreement gives access to encrypted healthcare data from an estimated 1.6 million patients who use the three hospitals run by the trust annually: the Royal Free, Barnet, and Chase Farm. The project attracted controversy for providing Google with confidential patient information, despite assurances from the tech giant that the information would only be used to inform diagnosis and treatment.
3. DeepMind use cases: Moorfields Eye Hospital
The second NHS partnership announced by Google was a collaboration with Moorfields Eye Hospital. The project was set up to develop a machine learning–based system that can recognise sight-threatening eye diseases from a digital scan of the eye. While its initial collaboration with the NHS at the Royal Free focused on patient care, this is the first that is entirely dedicated to medical research.
The programme involves the analysis of more than one million anonymous eye scans to produce an algorithm that detects early signs of emerging eye conditions and increases the speed of diagnosis.
A Moorfields ophthalmologist called Pearse Keane is credited with coming up with the idea. He contacted the company after seeing their technology help computers learn how to play video games, and believed it could be applied to images of the eye.
4. DeepMind use cases: University College London Hospital
DeepMind's latest research partnership with the NHS aims to improve the treatment of head and neck cancers. Before radiotherapy can begin, clinicians currently spend around four hours preparing a detailed map of each patient's body to avoid targeting the delicate surrounding tissue that can be damaged in treatment. The information is then fed into a radiotherapy machine to target the cancer without harming the healthy tissue.
Researchers at DeepMind believe machine learning can cut this time down to an hour. The team will analyse anonymised scans from UCLH patients to develop a radiotherapy segmentation algorithm that can automate parts of the process. They hope to eventually apply the algorithm to other parts of the body.
5. DeepMind use cases: WaveNet
While healthcare technology dominates the current DeepMind developments, its machine learning systems have also been extended to audio analysis. Talking machines have a long history in science fiction and have already entered mainstream usage through commercial products such as Apple's personal voice assistant Siri. But the gap between computer and human speech remains substantial.
DeepMind has developed a text-to-speech system that can close that gap by more than 50 per cent. Known as WaveNet, it uses a neural network to replicate the sound waves produced by human speakers rather than copying the language that they use. The computing power required to power Wavenet limits its practical applicability for now, but Google has already presented samples of its automatically generated piano pieces.
6. DeepMind use cases: Google
Google currently uses machine learning algorithms in a range of its own products, including Maps, Gmail, YouTube and Android. It believes that DeepMind’s technology could in future also be extended to search, robots and the Internet of Things. A DeepMind agent has already matched human performance at 49 Atari games including Pac-Man and Space Invaders, and developed the aforementioned AlphaGo, the first computer programme to ever win a game of Go against a professional player.
Google has even used DeepMind to cut the electricity bills at its huge data centres. DeepMind algorithms predicted the air conditioning required to cool the vast number of servers powering its services, which vary depending on user demand. The results were efficiency savings of 40 percent in the cooling systems, and a 15 percent reduction to the overall energy used in the data centres.
Neural Turing Machines attempt to emulate the brain's short-term memory
Google, the search company that's investing heavily in robotics, is teaming with Oxford University researchers to work on artificial intelligence.
DeepMind is a 'vacuum for very, very clever people', says Ocado Technology’s head of data