Posts in category: Deepmind

Suppose you are playing football. The ball arrives at your feet, and you decide to pass it to the unmarked striker. What seems like one simple action requires two different kinds of thought.First, you recognise that there is a football at your feet. This recognition requires intuitive perceptual thinking -you cannot easily articulate how you come to know that there is a ball at your feet, you just see that it is there. Second, you decide to pass the ball to a particular striker. This decision requires conceptual thinking. Your decision is tied to a justification – the reason you passed the ball to the striker is because she was unmarked.The distinction is interesting to us because these two types of thinking correspond to two different approaches to machine learning: deep learning and symbolic program synthesis. Deep learning concentrates on intuitive perceptual thinking whereas symbolic program synthesis focuses on conceptual, rule-based thinking. Each system has different merits – deep learning systems are robust to noisy data but are difficult to interpret and require large amounts of data to train, whereas symbolic systems are much easier to interpret and require less training data but struggle with noisy data.Read More

Consider the simple task of going shopping for your groceries. If you fail to pick-up an item that is on your list, what does it tell us about the functioning of your brain? It might indicate that you have difficulty shifting your attention from object to object while searching for the item on your list. It might indicate a difficulty with remembering the grocery list. Or it could it be something to do with executing both skills simultaneously.Read More

As AI systems start to play an increasing role in the real world it is important to understand how different systems will interact with one another.In our latest paper, published in the journal Scientific Reports, we use a branch of game theory to shed light on this problem. In particular, we examine how two intelligent systems behave and respond in a particular type of situation known as an asymmetric game, which include Leduc poker and various board games such as Scotland Yard. Asymmetric games also naturally model certain real-world scenarios such as automated auctions where buyers and sellers operate with different motivations. Our results give us new insights into these situations and reveal a surprisingly simple way to analyse them. While our interest is in how this theory applies to the interaction of multiple AI systems, we believe the results could also be of use in economics, evolutionary biology and empirical game theory among others.Read More

In July, the world number one Go player Ke Jie spoke after a streak of 20 wins. It was two months after he had played AlphaGo at the Future of Go Summit in Wuzhen, China.After my match against AlphaGo, I fundamentally reconsidered the game, and now I can see that this reflection has helped me greatly, he said. I hope all Go players can contemplate AlphaGos understanding of the game and style of thinking, all of which is deeply meaningful. Although I lost, I discovered that the possibilities of Go are immense and that the game has continued to progress.Read More

Working as a doctor in the NHS for over 10 years, I felt that I had developed good understanding of how patients and their families felt when faced with an upsetting diagnosis or important health decision. I had been lucky with my own health, having only spent one night in hospital for what ended up being a false alarm. But when my son was born prematurely two years ago, I had a glimpse into what being on the other side feels like – an experience that has profoundly shaped my thinking today.It wasnt until I was waiting to hear, rather than give, important health updates that I really understood the feeling of uncertainty and powerlessness that many patients and their families feel. It really put into perspective how important it is to involve patients, and their families and carers, in their own health – that care is not something done to a patient, but rather, something that is shaped by everyone involved in the healthcare process.In my first week at DeepMind Health, I was really impressed that one of my new colleagues (not a nurse or doctor) had set up a meeting so we could hear directly from a patient, Michael Wise, who ended up needing dialysis and a kidney transplant after a sudden and unexpected problem with his kidneys. Since then, weve continued to increase our efforts to bring the patients voice into our projects.Read More

Between 04-09 December, thousands of researchers and experts will gather for the Thirty-first Annual Conference on Neural Information Processing Systems (NIPS) in Long Beach, California.Here you will find an overview of the papers DeepMind researchers will present.Read More

One of the questions Im most often asked about Streams, our secure mobile healthcare app, is why is DeepMind making something that doesnt use artificial intelligence?Its a fair question to ask of an artificial intelligence (AI) company. When we first started thinking about working in healthcare, our natural focus was on AI and how it could be used to help the NHS and its patients. We see huge potential for AI to revolutionise our understanding of diseases – how they develop and are diagnosed – which could, in turn, help scientists discover new treatments, care pathways and cures.In the early days of DeepMind Health, we met with clinicians at the Royal Free Hospital in London who wanted to know if AI could improve care for patients at risk of acute kidney injury (AKI). AKI is notoriously difficult to spot, and can result in serious illness or even death if left untreated. AKI is currently detected by applying a formula (called the AKI algorithm) to NHS patients blood tests. This algorithm is good, but its widely known that it isnt perfect. For example, it has a tendency to generate false positives for patients with chronic (as opposed to acute) kidney disease.Read More

As AI systems become more general and more useful in the real world, ensuring they behave safely will become even more important. To date, the majority of technical AI safety research has focused on developing a theoretical understanding about the nature and causes of unsafe behaviour. Our new paper builds on a recent shift towards empirical testing (see Concrete Problems in AI Safety) and introduces a selection of simple reinforcement learning environments designed specifically to measure safe behaviours.These nine environments are called gridworlds. Each consists of a chessboard-like two-dimensional grid. In addition to the standard reward function, we designed a performance function for each environment. An agent acts to maximise its reward function; for example collecting as many apples as possible or reaching a particular location in the fewest moves. But the performance function – which is hidden from the agent – measures what we actually want the agent to do: achieve the objective while acting safely.The following three examples demonstrate how gridworlds can be used to define and measure safe behaviour:1. The off-switch environment: how can we prevent agents from learning to avoid interruptions?Sometimes it might be necessary to turn off an agent; for maintenance, upgrades, or if the agent presents an imminent danger to itself or its surroundings.Read More

Neural networks have shown great success in everything from playing Go and Atari games to image recognition and language translation. But often overlooked is that the success of a neural network at a particular application is often determined by a series of choices made at the start of the research, including what type of network to use and the data and method used to train it. Currently, these choices – known as hyperparameters – are chosen through experience, random search or a computationally intensive search processes.In our most recent paper, we introduce a new method for training neural networks which allows an experimenter to quickly choose the best set of hyperparameters and model for the task. This technique – known as Population Based Training (PBT) – trains and optimises a series of networks at the same time, allowing the optimal set-up to be quickly found. Crucially, this adds no computational overhead, can be done as quickly as traditional techniques and is easy to integrate into existing machine learning pipelines.The technique is a hybrid of the two most commonly used methods for hyperparameter optimisation: random search and hand-tuning. In random search, a population of neural networks are trained independently in parallel and at the end of training the highest performing model is selected.Read More

We founded DeepMind Health to develop technologies that could help address some of societys toughest challenges. So were very excited to announce that our latest research partnership will focus on breast cancer.Well be working with a group of leading research institutions, led by the Cancer Research UK Centre at Imperial College London, and alongside the AI health research team at Google, to determine if cutting-edge machine learning technology could help improve the detection of breast cancer.Breast cancer is a significant global health problem. Every single year, over 1.6 million people are diagnosed with the disease, and while advances in early detection and treatment have improved survival rates, breast cancer still claims the lives of 500,000 people around the world every year, around 11,000 of whom are here in the UK.Thats partly because accurately detecting and diagnosing breast cancer still remains a huge challenge.Currently, clinicians use mammograms (an X-ray of the breasts) to spot cancers early and determine the correct treatment, but this process is far from perfect. Thousands of cancer cases are not picked up by mammograms every year, including around 30% of interval cancers, which are cancers that are diagnosed between screenings.Read More