WaveNet launches in the Google Assistant
Twelve months ago we published details of WaveNet, a deep neural network for generating raw audio waveforms that was capable of producing more realistic-sounding speech than existing text-to-speech techniques. We have now updated the model so that it is faster, higher quality and able to run at Google scale. As of today, we are proud to announce that this new model is being used to generate the Google Assistant voices for US English and Japanese across all platforms.
作者：DeepMind 来源：DeepMind
发布时间：2017-10-05 02:00:00
Why we launched DeepMind Ethics & Society
Today we’re launching a new research unit, DeepMind Ethics & Society, to complement our work in AI science and application. This new unit will help us explore and understand the real-world impacts of AI. It has a dual aim: to help technologists put ethics into practice, and to help society anticipate and direct the impact of AI so that it works for the benefit of all.
作者：DeepMind 来源：DeepMind
发布时间：2017-10-04 03:54:00
The hippocampus as a 'predictive map'
In our new paper, in Nature Neuroscience, we apply a neuroscience lens to a longstanding mathematical theory from machine learning to provide new insights into the nature of learning and memory. Specifically, we propose that the area of the brain known as the hippocampus offers a unique solution to this problem by compactly summarising future events using what we call a “predictive map.”
作者：DeepMind 来源：DeepMind
发布时间：2017-10-02 23:30:00
DeepMind and Blizzard open StarCraft II as an AI research environment
Along with our partner Blizzard Entertainment, we are excited to announce the release of the Starcraft II Learning Environment (SC2LE), a suite of tools that we hope will accelerate AI research in the real-time strategy game and make it easier for researchers to focus on the frontiers of our field.
作者：DeepMind 来源：DeepMind
发布时间：2017-08-10 01:00:00
DeepMind papers at ICML 2017 (part one)
The first of three blogs that give an overview of the papers we are presenting at the ICML 2017 Conference in Sydney, Australia.
作者：DeepMind 来源：DeepMind
发布时间：2017-08-05 01:00:00
DeepMind papers at ICML 2017 (part two)
The second of three blogs that give an overview of the papers we are presenting at the ICML 2017 Conference in Sydney, Australia.
作者：DeepMind 来源：DeepMind
发布时间：2017-08-05 01:00:00
DeepMind papers at ICML 2017 (part three)
The third of three blogs that give an overview of the papers we are presenting at the ICML 2017 Conference in Sydney, Australia.
作者：DeepMind 来源：DeepMind
发布时间：2017-08-05 01:00:00
AI and Neuroscience: A virtuous circle
Recent progress in AI has been remarkable. While this is attributed to several factors, one often overlooked contribution is the use of ideas from experimental and theoretical neuroscience. We argue that if we are to continue to make rapid advances, researchers should not lose sight of this valuable interplay. We urge researchers in neuroscience and AI to find a common language, allowing a free flow of knowledge that will allow continued progress in both fields.
作者：DeepMind 来源：DeepMind
发布时间：2017-08-02 15:00:00
Going beyond average for reinforcement learning
In reinforcement learning (RL) applications, random perturbations influence the exact amount of reward received. A typical algorithm predicts the average reward across multiple trials, and uses this prediction to decide how to act. In our latest work, we show it is equally possible to model not only the average but also the full variation of this reward, what we call the value distribution. This not only results in RL systems that are more accurate and faster to train than previous models, but more importantly opens up the possibility of rethinking the whole of reinforcement learning.
作者：DeepMind 来源：DeepMind
发布时间：2017-07-24 21:43:00
Agents that imagine and plan
Imagining the consequences of your actions before you take them is a powerful tool of human cognition and is a crucial tool in our everyday lives. If we are to develop sophisticated algorithms they too must have the capability to ‘imagine’ and reason about the future. Beyond that they must be able to construct a plan using this knowledge. But even for the most intelligent agents, imagining in complex environments is a long and costly process. In two new papers, we describe a new family of approaches for imagination-based planning.
作者：DeepMind 来源：DeepMind
发布时间：2017-07-20 19:30:00
Imagine this: Creating new visual concepts by recombining familiar ones
The ability to learn new concepts by recombining existing ones through symbolic instructions allows humans to reason about abstract concepts like the universe or mathematics. But this idea of “compositionality” remains a challenge in AI research. In our new paper, we propose a novel theoretical approach to address this problem. We also demonstrate a new neural network component called the Symbol-Concept Association Network (SCAN), that can, for the first time, learn a grounded visual concept hierarchy in a way that mimics human vision and word acquisition, enabling it to imagine novel concepts guided by language instructions.
作者：DeepMind 来源：DeepMind
发布时间：2017-07-12 16:00:00
Producing flexible behaviours in simulated environments
In three new papers, we demonstrate about how simulated agents can master sophisticated motor control - a hallmark of physical intelligence, and a crucial part of AI research. Specifically, we show ways to produce flexible and natural behaviours that can be reused and adapted to solve tasks.
作者：DeepMind 来源：DeepMind
发布时间：2017-07-10 19:00:00
DeepMind expands to Canada with new research office in Edmonton, Alberta
We're thrilled to announce the opening of DeepMind’s first ever international AI research office in Edmonton, Canada, in close collaboration with the University of Alberta. ‘DeepMind Alberta’ will be led by the pioneer of reinforcement learning - and DeepMind’s first ever advisor from back in 2010 - Rich Sutton, together with Michael Bowling and Patrick Pilarski. All three will maintain their professorships at the UAlberta, and continue to teach and contribute to the academic community. They’ll be joined by Adam White, who will be returning to Canada to join the university as an adjunct professor, and six more researchers who co-authored the influential DeepStack paper published earlier this year in Science.
作者：DeepMind 来源：DeepMind
发布时间：2017-07-05 22:00:00
Independent Reviewers release first annual report on DeepMind Health
Today, a panel of Independent Reviewers has published its first annual report into DeepMind Health.
作者：DeepMind 来源：DeepMind
发布时间：2017-07-05 07:01:00
The Information Commissioner, the Royal Free, and what we’ve learned
The Information Commissioner's Office has concluded a year-long investigation that focused on the Royal Free’s clinical testing of Streams in late 2015 and 2016, which was intended to guarantee that the service could be deployed safely at the hospital. Although the findings are about the decisions made by the Royal Free, we need to reflect on our own actions too.
作者：DeepMind 来源：DeepMind
发布时间：2017-07-03 19:47:00
