Aaron Ng
24Patents
6h-index
35Co-inventors
65Inventor score
Filing activity: Feb 1, 2008 → Aug 8, 2022
Most-cited inventions
| Patent | Title | Area | Cited by | Status |
|---|---|---|---|---|
| US10192016B2 | Neural network based physical synthesis for circuit designs | Physics | 42 | Active |
| US10354733B1 | Software-defined memory bandwidth reduction by hierarchical stream buffering for general matrix multiplication in a programmable IC | Physics | 17 | Active |
| US11204747B1 | Re-targetable interface for data exchange between heterogeneous systems and accelerator abstraction into software instructions | Physics | 12 | Active |
| US10515135B1 | Data format suitable for fast massively parallel general matrix multiplication in a programmable IC | Physics | 11 | Active |
| US10460416B1 | Inline image preprocessing for convolution operations using a matrix multiplier on an integrated circuit | Physics | 8 | Active |
| US11222256B2 | Neural network processing system having multiple processors and a neural network accelerator | Physics | 7 | Active |
| US10943039B1 | Software-driven design optimization for fixed-point multiply-accumulate circuitry | Physics | 6 | Active |
| US10678509B1 | Software-driven design optimization for mapping between floating-point and fixed-point multiply accumulators | Physics | 5 | Active |
| US9836568B1 | Programmable integrated circuit design flow using timing-driven pipeline analysis | Physics | 5 | Active |
| US9646126B1 | Post-routing structural netlist optimization for circuit designs | Physics | 5 | Active |
| US10936311B1 | Sparse matrix processing circuitry | Physics | 5 | Active |
| US11568218B2 | Neural network processing system having host controlled kernel acclerators | Physics | 4 | Active |
| US7840919B1 | Resource mapping of functional areas on an integrated circuit | Physics | 3 | Active |
| US11620490B2 | Multi-layer neural network processing by a neural network accelerator using host communicated merged weights and a package of per-layer instructions | Physics | 3 | Active |
| US10366201B1 | Timing closure of circuit designs for integrated circuits | Physics | 2 | Active |
| US9965581B1 | Fanout optimization to facilitate timing improvement in circuit designs | Physics | 2 | Active |
| US10984500B1 | Inline image preprocessing for convolution operations using a matrix multiplier on an integrated circuit | Physics | 2 | Active |
| US11036827B1 | Software-defined buffer/transposer for general matrix multiplication in a programmable IC | Physics | 2 | Active |
| US11429848B2 | Host-directed multi-layer neural network processing via per-layer work requests | Electricity | 2 | Active |
| US11386644B2 | Image preprocessing for generalized image processing | Physics | 1 | Active |
| US12079158B2 | Reconfigurable neural engine with extensible instruction set architecture | Physics | 0 | Active |
| US12248786B2 | Instruction set architecture for data processing array control | Physics | 0 | Active |
| US11694066B2 | Machine learning runtime library for neural network acceleration | Physics | 0 | Active |
| US12412109B2 | Machine learning deployment platform | Physics | 0 | Active |
Source: USPTO / EPO open patent data. Inventor disambiguation is heuristic; counts are objective bibliographic measures.