Li Deng
30Patents
10h-index
61Co-inventors
74Inventor score
Filing activity: Dec 10, 2009 → Jun 18, 2019
Most-cited inventions
| Patent | Title | Area | Cited by | Status |
|---|---|---|---|---|
| US9846836B2 | Modeling interestingness with deep neural networks | Physics | 106 | Active |
| US9836671B2 | Discovery of semantic similarities between images and text | Physics | 27 | Active |
| US9535960B2 | Context-sensitive search using a deep learning model | Physics | 25 | Active |
| US9519859B2 | Deep structured semantic model produced using click-through data | Physics | 25 | Active |
| US10089576B2 | Representation learning using multi-task deep neural networks | Physics | 23 | Active |
| US9842585B2 | Multilingual deep neural network | Physics | 16 | Active |
| US10133729B2 | Semantically-relevant discovery of solutions | Physics | 14 | Active |
| US8972253B2 | Deep belief network for large vocabulary continuous speech recognition | Physics | 13 | Active |
| US9031844B2 | Full-sequence training of deep structures for speech recognition | Physics | 12 | Active |
| US10546066B2 | End-to-end learning of dialogue agents for information access | Physics | 11 | Active |
| US10055686B2 | Dimensionally reduction of linguistics information | Physics | 5 | Active |
| US9070360B2 | Confidence calibration in automatic speech recognition systems | Physics | 4 | Active |
| US11170293B2 | Multi-model controller | Physics | 4 | Active |
| US10366163B2 | Knowledge-guided structural attention processing | Physics | 4 | Active |
| US10909450B2 | Multiple-action computational model training and operation | Physics | 4 | Active |
| US9165243B2 | Tensor deep stacked neural network | Physics | 4 | Active |
| US9099083B2 | Kernel deep convex networks and end-to-end learning | Physics | 3 | Active |
| US9390371B2 | Deep convex network with joint use of nonlinear random projection, restricted boltzmann machine and batch-based parallelizable optimization | Physics | 3 | Active |
| US11449744B2 | End-to-end memory networks for contextual language understanding | Physics | 3 | Active |
| US10776716B2 | Unsupervised learning utilizing sequential output statistics | Physics | 2 | Active |
| US10474950B2 | Training and operation of computational models | Physics | 2 | Active |
| US11783173B2 | Multi-domain joint semantic frame parsing | Physics | 2 | Active |
| US10204097B2 | Efficient dialogue policy learning | Physics | 2 | Active |
| US10997233B2 | Multi-stage image querying | Physics | 2 | Active |
| US9292787B2 | Computer-implemented deep tensor neural network | Physics | 2 | Active |
Source: USPTO / EPO open patent data. Inventor disambiguation is heuristic; counts are objective bibliographic measures.