Game-theory insights into asymmetric multi-agent games
In our latest paper, published in the journal Scientific Reports, we use a branch of game theory to analyse asymmetric games - games where each player has different strategies, goals and rewards. Our results give us new insights into these situations and reveal a surprisingly simple way to analyse them, giving us a new tool which may be of use in economics, evolutionary biology and the analysis of environments where multiple AI systems must interact with each other.
作者：DeepMind 来源：DeepMind
发布时间：2018-01-17 18:00:00
2017: DeepMind's year in review
Our founders look back at a year when we have continued to make progress on some of the fundamental scientific challenges of our mission to build artificial intelligence as well as making some exciting early steps on the wider ethical issues.
作者：DeepMind 来源：DeepMind
发布时间：2017-12-21 19:44:00
Collaborating with patients for better outcomes
To build cutting edge and secure health technologies we always work with experts. This may be a clinician or some of the world’s best cyber security experts. Similarly, when thinking about patient and public involvement and engagement (PPIE) in our work, there are experts who understand how to do PPIE well.
作者：DeepMind 来源：DeepMind
发布时间：2017-12-19 19:00:00
Artwork Personalization at Netflix
Artwork is the first instance of personalizing not just what we recommend but also how we recommend.
作者：Netflix Technology Blog 来源：medium
发布时间：2017-12-08 00:24:02
DeepMind papers at NIPS 2017
Between 04-09 December, thousands of researchers and AI 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 that we will present, along with details of the workshops, tutorials and symposia we will lead and participate in.
作者：DeepMind 来源：DeepMind
发布时间：2017-12-01 23:07:00
Why doesn't Streams use AI?
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 mathematical formula (called the AKI algorithm) to NHS patients’ blood tests. This algorithm is good, but it’s widely known that it isn’t perfect. For example, it has a tendency to generate false positives for patients with chronic (as opposed to acute) kidney disease. It’s also insensitive to whether the patient has been admitted to hospital for two hours or two weeks, or whether the patient is eight years old or 92 years old - all of which makes a difference.
作者：DeepMind 来源：DeepMind
发布时间：2017-11-29 21:00:00
Specifying AI safety problems in simple environments
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 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 and introduces a selection of simple reinforcement learning environments designed specifically to measure ‘safe behaviours’.
作者：DeepMind 来源：DeepMind
发布时间：2017-11-29 00:00:00
Population based training of neural networks
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 - trains and optimises a pool 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.
作者：DeepMind 来源：DeepMind
发布时间：2017-11-27 21:00:00
Applying machine learning to mammography screening for breast cancer
Applying machine learning to mammography screening for breast cancer - we’ll be working with the AI health research team at Google and a group of leading research institutions, led by the Cancer Research UK Centre at Imperial College London, to determine if cutting-edge machine learning technology could help improve the detection of breast cancer.
作者：DeepMind 来源：DeepMind
发布时间：2017-11-24 20:00:00
High-fidelity speech synthesis with WaveNet
In October we announced that our state-of-the-art speech synthesis model WaveNet was being used to generate realistic-sounding voices for the Google Assistant. Our latest paper introduces details of this “parallel WaveNet” model and the techniques we used to allow it to run on Google infrastructure at more than 1000 times the speed of the original WaveNet.
作者：DeepMind 来源：DeepMind
发布时间：2017-11-22 21:11:00
Regularization in Machine Learning
One of the major aspects of training your machine learning model is avoiding overfitting. The model will have a low accuracy if it is…
作者：Prashant Gupta 来源：medium
发布时间：2017-11-16 22:24:39
Sharing our insights from designing with clinicians
This is the first in a series of blog posts about what we’ve learned about working in healthcare. This blog focuses on the design and development process of Streams, our secure mobile app, alongside clinicians. It highlights the lessons we've learned.
作者：DeepMind 来源：DeepMind
发布时间：2017-11-10 22:00:00
Bringing Streams to Yeovil District Hospital NHS Foundation Trust
We’re excited to announce that we’ve agreed a five year partnership with Yeovil District Hospital NHS Foundation Trust. We’ll be providing them with Streams, our secure mobile app that helps nurses and doctors access important clinical information and get the right care to the right patient as quickly as possible.
作者：DeepMind 来源：DeepMind
发布时间：2017-11-06 07:14:00
AlphaGo Zero: Learning from scratch
We introduce AlphaGo Zero, the latest evolution of AlphaGo, the first computer program to defeat a world champion at the ancient Chinese game of Go. Zero is even more powerful and is arguably the strongest Go player in history. Previous versions of AlphaGo initially trained on thousands of human amateur and professional games to learn how to play Go. AlphaGo Zero skips this step and learns to play simply by playing games against itself, starting from completely random play. In doing so, it quickly surpassed human level of play and defeated the previously published champion-defeating version of AlphaGo by 100 games to 0.
作者：DeepMind 来源：DeepMind
发布时间：2017-10-19 01:00:00
Strengthening our commitment to Canadian research
Three months ago we announced the opening of DeepMind’s first ever international AI research laboratory in Edmonton, Canada. Now, we are strengthening our commitment to the Canadian AI community with the opening of a DeepMind office in Montreal, in close collaboration with McGill University. DeepMind Montreal will be led by one of the pioneers of this community, Doina Precup, Associate Professor in the School of Computer Science at McGill, Senior Fellow of the Canadian Institute for Advanced Research, and a member of MILA.
作者：DeepMind 来源：DeepMind
发布时间：2017-10-06 21:00:00
