DeepMind Technologies Limited
London, GB· latest recorded location
DeepMind Technologies Limited is named on 304 US grants and 532 published applications in this corpus. Its most frequent classification is G06N (AI and machine learning).
- Granted patents
- 304
- Published applications
- 532
- Inventors
- 751
- Granted in 2026 so far
- 4
- Most recent grant
- 24 Feb 2026
Filings by year
| Year | Granted patents | Published applications |
|---|---|---|
| 2014 | 1 | 2 |
| 2015 | 0 | 0 |
| 2016 | 0 | 1 |
| 2017 | 0 | 1 |
| 2018 | 1 | 11 |
| 2019 | 13 | 38 |
| 2020 | 51 | 75 |
| 2021 | 25 | 74 |
| 2022 | 37 | 38 |
| 2023 | 64 | 74 |
| 2024 | 63 | 113 |
| 2025 | 45 | 70 |
| 2026 | 4 | 35 |
By publication year. 2026 runs to 8 Oct.
Technology areas
Share of 836 documents
- G06NAI and machine learning778
- G06FComputing and data processing321
- G06VImage and video recognition144
- G06TImage data processing or generation, in general71
- G10LSpeech analysis techniques or speech synthesis67
- H04NPictorial communication34
- G05BControl or regulating systems in general29
- B25JManipulators25
- G16BBioinformatics, i.e. information and communication technology [ict] specially adapted for genetic or21
- G10HElectrophonic musical instruments17
Recent patents
- Continuous control with deep reinforcement learning24 Feb 2026US 12,561,573 B2
- Recurrent unit for generating or processing a sequence of images20 Jan 2026US 12,530,814 B2
- Artificial intelligence-based image search refinement13 Jan 2026US 12,524,459 B1
- Generative neural network systems for generating instruction sequences to control an agent performing a task6 Jan 2026US 12,518,128 B2
- Aligning entities using neural networks2 Dec 2025US 12,488,579 B2
- Graph neural networks that model face-face interactions between meshes10 Sep 2026US 2026/0268042 A1
- Generating output sequences with inline evidence using language model neural networks3 Sep 2026US 2026/0260120 A1
- Model-free reinforcement learning with regularized nash dynamics3 Sep 2026US 2026/0260121 A1
- Reinforcement learning using density estimation with online clustering for exploration27 Aug 2026US 2026/0252897 A1
- Controlling agents using proto-goal pruning13 Aug 2026US 2026/0236773 A1
Most cited
By later US patents
- Training variational autoencoders to generate disentangled latent factors20US 10,373,055 B12019
- Reinforcement learning using agent curricula20US 11,113,605 B22021
- Processing sequences using convolutional neural networks19US 11,080,591 B22021
- Selecting reinforcement learning actions using goals and observations18US 10,628,733 B22020
- Selecting actions from large discrete action sets using reinforcement learning17US 10,885,432 B12021