US Patent Application 18010426. CONDITIONAL OUTPUT GENERATION THROUGH DATA DENSITY GRADIENT ESTIMATION simplified abstract
Contents
CONDITIONAL OUTPUT GENERATION THROUGH DATA DENSITY GRADIENT ESTIMATION
Organization Name
Inventor(s)
Nanxin Chen of Baltimore MD (US)
Ron J. Weiss of New York NY (US)
Mohammad Norouzi of Toronto (CA)
Yu Zhang of Mountain View CA (US)
CONDITIONAL OUTPUT GENERATION THROUGH DATA DENSITY GRADIENT ESTIMATION - A simplified explanation of the abstract
- This abstract for appeared for US patent application number 18010426 Titled 'CONDITIONAL OUTPUT GENERATION THROUGH DATA DENSITY GRADIENT ESTIMATION'
Simplified Explanation
This abstract describes a method for generating outputs based on inputs from a network using neural networks. The method involves obtaining the network input, initializing a current network output, and generating the final network output through a series of iterations. Each iteration corresponds to a different noise level, and the current network output is updated at each iteration. This updating process involves using a noise estimation neural network to process the current network output and the network input, generating a noise output that provides a noise estimate for each value in the current network output. The current network output is then updated using the noise estimate and the noise level for that iteration.
Original Abstract Submitted
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating outputs conditioned on network inputs using neural networks. In one aspect, a method comprises obtaining the network input; initializing a current network output; and generating the final network output by updating the current network output at each of a plurality of iterations, wherein each iteration corresponds to a respective noise level, and wherein the updating comprises, at each iteration: processing a model input for the iteration comprising (i) the current network output and (ii) the network input using a noise estimation neural network that is configured to process the model input to generate a noise output, wherein the noise output comprises a respective noise estimate for each value in the current network output; and updating the current network output using the noise estimate and the noise level for the iteration.
