Amar Phanishayee
16Patents
3h-index
32Co-inventors
56Inventor score
Filing activity: Oct 22, 2012 → Jan 17, 2024
Most-cited inventions
| Patent | Title | Area | Cited by | Status |
|---|---|---|---|---|
| US10084868B2 | IoT gateway for weakly connected settings | Electricity | 8 | Active |
| US9798631B2 | Block storage by decoupling ordering from durability | Physics | 5 | Active |
| US10114709B2 | Block storage by decoupling ordering from durability | Physics | 3 | Active |
| US9342411B2 | Reducing memory overhead of highly available, distributed, in-memory key-value caches | Physics | 3 | Active |
| US11715002B2 | Efficient data encoding for deep neural network training | Physics | 2 | Active |
| US11868880B2 | Mitigating communication bottlenecks during parameter exchange in data-parallel DNN training | Emerging Cross-Sectional Technologies | 1 | Active |
| US10187292B2 | Data center topology having multiple classes of reliability | Electricity | 1 | Active |
| US10061791B2 | Data management for connected devices | Physics | 0 | Active |
| US10356187B2 | IoT gateway for weakly connected settings | Electricity | 0 | Active |
| US11562247B2 | Neural network activation compression with non-uniform mantissas | Electricity | 0 | Active |
| US12045724B2 | Neural network activation compression with outlier block floating-point | Electricity | 0 | Active |
| US12165038B2 | Adjusting activation compression for neural network training | Physics | 0 | Active |
| US12067495B2 | Neural network activation compression with non-uniform mantissas | Electricity | 0 | Active |
| US12056604B2 | Highly performant pipeline parallel deep neural network training | Physics | 0 | Active |
| US12277502B2 | Neural network activation compression with non-uniform mantissas | Electricity | 0 | Active |
| US10740195B2 | Block storage by decoupling ordering from durability | Physics | 0 | Active |
Source: USPTO / EPO open patent data. Inventor disambiguation is heuristic; counts are objective bibliographic measures.