首页|Patent Issued for Methods for efficient 3D SRAM-based computein- memory (USPTO 1 2159683)
Patent Issued for Methods for efficient 3D SRAM-based computein- memory (USPTO 1 2159683)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
The patent’s inventors are Fouda, Mohammed Elneanaei Abdelmoneem (Irvine, CA, US ).This patent was filed on May 2, 2024 and was published online on December 3, 202 4.From the background information supplied by the inventors, news correspondents o btained the followingquote: “Artificial intelligence (AI), or machine learning, utilizes learning networks (e.g. deep neuralnetworks) loosely inspired by the brain in order to solve problems. Learning networks typically include layersof weights that weight signals (mimicking synapses) interleaved with activation lay ers that apply activationfunctions to the signals (mimicking neurons). Thus, a weight layer provides weighted input signals to anactivation layer. Neurons in the activation layer operate on the weighted input signals by applying someacti vation function to the input signals and provide output signals corresponding to the statuses of theneurons. The output signals from the activation layer are p rovided as input signals to the next weightlayer, if any. This process may be r epeated for the layers of the network. Learning networks are thusable to reduce complex problems to a set of weights and the applied activation functions. The structure ofthe network (e.g., number of layers, connectivity among the layers, dimensionality of the layers, the typeof activation function, the weights or p arameters for the network, etc.) are together known as a model.The values of th e parameters (e.g. the weights used for particular tasks) for the model are iden tified viatraining of the learning network. Moreover, learning networks can lev erage hardware, such as graphicsprocessing units (GPUs) and/or AI accelerators, which perform operations usable in machine learning inparallel. Such tools can dramatically improve the speed and efficiency with which data-heavy and other tasks can be accomplished by the learning network.
NETL
NSTL
