Demis Hassabis
Research Homepage
http://demishassabis.com/
DEEP REINFORCEMENT LEARNING
FRIDAY, 17TH JUNE, 2016
by David Silver, Google DeepMind
DEEP REINFORCEMENT LEARNING
FRIDAY, 17TH JUNE, 2016
by David Silver, Google DeepMind
https://deepmind.com/blog
DECOUPLED NEURAL INTERFACES USING SYNTHETIC GRADIENTS
MONDAY, 29TH AUGUST, 2016
https://deepmind.com/blog#decoupled-neural-interfaces-using-synthetic-gradients
by Max Jaderberg, DeepMind
Neural networks are the workhorse of many of the algorithms developed at DeepMind. For example, AlphaGo uses convolutional neural networks to evaluate board positions in the game of Go and DQN and Deep Reinforcement Learning algorithms use neural networks to choose actions to play at super-human level on video games.
This post introduces some of our latest research in progressing the capabilities and training procedures of neural networks called Decoupled Neural Interfaces using Synthetic Gradients. This work gives us a way to allow neural networks to communicate, to learn to send messages between themselves, in a decoupled, scalable manner paving the way for multiple neural networks to communicate with each other or improving the long term temporal dependency of recurrent networks. This is achieved by using a model to approximate error gradients, rather than by computing error gradients explicitly with backpropagation. The rest of this post assumes some familiarity with neural networks and how to train them. If you’re new to this area we highly recommend Nando de Freitas lecture series on Youtube on deep learning and neural networks.
Neural networks are the workhorse of many of the algorithms developed at DeepMind. For example, AlphaGo uses convolutional neural networks to evaluate board positions in the game of Go and DQN and Deep Reinforcement Learning algorithms use neural networks to choose actions to play at super-human level on video games.
This post introduces some of our latest research in progressing the capabilities and training procedures of neural networks called Decoupled Neural Interfaces using Synthetic Gradients. This work gives us a way to allow neural networks to communicate, to learn to send messages between themselves, in a decoupled, scalable manner paving the way for multiple neural networks to communicate with each other or improving the long term temporal dependency of recurrent networks. This is achieved by using a model to approximate error gradients, rather than by computing error gradients explicitly with backpropagation. The rest of this post assumes some familiarity with neural networks and how to train them. If you’re new to this area we highly recommend Nando de Freitas lecture series on Youtube on deep learning and neural networks.
https://scholar.google.com/citations
| What Learning Systems do Intelligent Agents Need? Complementary Learning Systems Theory Updated
D Kumaran, D Hassabis, JL McClelland
Trends in Cognitive Sciences 20 (7), 512-534
| 2016 | |
| Model-Free Episodic Control
C Blundell, B Uria, A Pritzel, Y Li, A Ruderman, JZ Leibo, J Rae, ...
arXiv preprint arXiv:1606.04460
| 2016 | |
| Neural Mechanisms of Hierarchical Planning in a Virtual Subway Network
J Balaguer, H Spiers, D Hassabis, C Summerfield
Neuron 90 (4), 893-903
| 1 | 2016 |
| Mastering the game of Go with deep neural networks and tree search
D Silver, A Huang, CJ Maddison, A Guez, L Sifre, G Van Den Driessche, ...
Nature 529 (7587), 484-489
| 120 | 2016 |
| Approximate Hubel-Wiesel Modules and the Data Structures of Neural Computation
JZ Leibo, J Cornebise, S Gómez, D Hassabis
arXiv preprint arXiv:1512.08457
| 1 | 2015 |
| Hippocampal place cells construct reward related sequences through unexplored space
HF Olafsdottir, C Barry, AB Saleem, D Hassabis, HJ Spiers
Elife 4, e06063
| 18 | 2015 |
| Human-level control through deep reinforcement learning
V Mnih, K Kavukcuoglu, D Silver, AA Rusu, J Veness, MG Bellemare, ...
Nature 518 (7540), 529-533
| 358 | 2015 |
| A goal direction signal in the human entorhinal/subicular region
MJ Chadwick, AEJ Jolly, DP Amos, D Hassabis, HJ Spiers
Current Biology 25 (1), 87-92
| 18 | 2015 |
| Foraging under competition: the neural basis of input-matching in humans
D Mobbs, D Hassabis, R Yu, C Chu, M Rushworth, E Boorman, ...
The Journal of Neuroscience 33 (23), 9866-9872
| 10 | 2013 |
| Imagine all the people: how the brain creates and uses personality models to predict behavior
D Hassabis, RN Spreng, AA Rusu, CA Robbins, RA Mar, DL Schacter
Cerebral Cortex, bht042
| 49 | 2013 |
| Detecting representations of recent and remote autobiographical memories in vmPFC and hippocampus
HM Bonnici, MJ Chadwick, A Lutti, D Hassabis, N Weiskopf, EA Maguire
The journal of neuroscience 32 (47), 16982-16991
| 50 | 2012 |
| The future of memory: remembering, imagining, and the brain
DL Schacter, DR Addis, D Hassabis, VC Martin, RN Spreng, KK Szpunar
Neuron 76 (4), 677-694
| 300 | 2012 |
| Multi-voxel pattern analysis in human hippocampal subfields
HB Bonnici, M Chadwick, D Kumaran, D Hassabis, N Weiskopf, ...
Frontiers in human neuroscience 6, 290
| 47 | 2012 |
| Decoding representations of scenes in the medial temporal lobes
HM Bonnici, D Kumaran, MJ Chadwick, N Weiskopf, D Hassabis, ...
Hippocampus 22 (5), 1143-1153
| 48 | 2012 |
| Scene construction in amnesia: An fMRI study
SL Mullally, D Hassabis, EA Maguire
The Journal of Neuroscience 32 (16), 5646-5653
| 53 | 2012 |
| Is the brain a good model for machine intelligence?
R Brooks, D Hassabis, D Bray, A Shashua
Nature 482 (7386), 462-463
| 2 | 2012 |
| Decoding overlapping memories in the medial temporal lobes using high-resolution fMRI
MJ Chadwick, D Hassabis, EA Maguire
Learning & Memory 18 (12), 742-746
| 31 | 2011 |
| Role of the hippocampus in imagination and future thinking
EA Maguire, D Hassabis
Proceedings of the National Academy of Sciences 108 (11), E39-E39
| 51 | 2011 |
| Imagining fictitious and future experiences: Evidence from developmental amnesia
EA Maguire, F Vargha-Khadem, D Hassabis
Neuropsychologia 48 (11), 3187-3192
| 75 | 2010 |
| Differential engagement of brain regions within a ‘core’network during scene construction
JJ Summerfield, D Hassabis, EA Maguire
Neuropsychologia 48 (5), 1501-1509
| 66 | 2010 |
