Data multiplexing for neural networks
AssigneeThe Trustees of Princeton University
InventorsKarthik NARASIMHAN, Vishvak MURAHARI, Carlos JIMENEZ, Runzhe YANG, Ameet DESHPANDE, Yushan SU, Kai LI
Life of this patent
Estimating the term…
Loading patent…
AssigneeThe Trustees of Princeton University
InventorsKarthik NARASIMHAN, Vishvak MURAHARI, Carlos JIMENEZ, Runzhe YANG, Ameet DESHPANDE, Yushan SU, Kai LI
Estimating the term…
Disclosed is a technique for improving the throughput of a neural network, using multiplexing and demultiplexing of information. Specifically, the multiplexing may include receiving a plurality of inputs, generating transformed inputs by performing, via a multiplexing layer, a transformation to each input of the plurality of inputs, and combining the transformed inputs into a single compact representation of the plurality of inputs. The demultiplexing may include receiving an output from a neural network, generating a plurality of values by converting, via a demultiplexing layer, the output back into independent representations, and producing predictions for each input based on the plurality of values. Further improvements may be seen when pretraining of the neural network and/or high-throughput transformers are incorporated.
A method for improving the throughput of a neural network, comprising:
A non-transitory computer-readable storage medium containing instructions that, when executed, cause a processor to perform operations that include:
17-30. (canceled)
A system, comprising:
32-45. (canceled)
A neural network apparatus, comprising:
47-53. (canceled)
Dependencies are inferred from the claim text.