Samsung Electronics secures grants for high-speed CNN processing and low-light media enhancement.
The filing
On November 10, 2026, Samsung Electronics Co., Ltd. received three US patents covering methods to accelerate convolutional neural network operations and enhance media in low-light conditions. One patent describes a system that selects a survival network based on a high-speed convolution result to perform precision convolution operations (USPTO 12,792,708 (opens in a new tab)), another covers a post-training quantization approach that topologically sorts neural network layers and selects quantization modes based on the quantization of previous layers (USPTO 12,792,869 (opens in a new tab)), and a third details a method for low-light media enhancement that aligns frames, corrects brightness, and selects from multiple neural networks based on parameters like shot boundary detection and artificial light flickering (USPTO 12,792,880 (opens in a new tab)). The company also published a US application on November 12, 2026, for an apparatus and method with frequency detection (USPTO 20260359695A1 (opens in a new tab)).
This filing’s full text wasn’t read for this issue, so its claims and the applicant’s background aren’t shown. Both are in the record, under Sources.