MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202638104250 A) filed by Fraunhofer-Gesellschaft Zur Frderung Der on August 29, 2026, for Concepts For Coding Neural Networks Parameters.
Inventors include Haase, Paul; Kirchhoffer, Heiner; Schwarz, Heiko; Marpe, Detlev; and Wiegand, Thomas Filed On.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: Embodiments according to a first aspect of the present invention are based on the idea, that neural network parameters may be compressed more efficiently by using a non-constant quantizer, but varying same during coding the neural network parameters, namely by selecting a set of reconstruction levels depending on quantization indices decoded from, or respectively encoded, into the data stream for previous or respectively previously encoded neural network parameters. Embodiments according to a second aspect of the present invention are based on the idea that a more efficient neural network coding may be achieved when done in stages – called reconstruction layers to distinguish them from the layered composition of the neural network in neural layers – and if the parametrizations provided in these stages are then, neural network parameter-wise combined to yield a neural network parametrization improved compared to any of the stages.
Disclaimer: Curated by HT Syndication.
