Patent assignee · US · COMPANY

Perceive Corporation

83Patents
83Active
83Granted
55Portfolio score

Filing activity: Jul 31, 2016 → Oct 10, 2022

Most-cited patents

PatentTitleAreaCited byStatus
US11250326B1 Splitting neural network filters for implementation by neural network inference circuit Physics 46 Active
US10453220B1 Machine-trained network for misalignment-insensitive depth perception Electricity 34 Active
US10740434B1 Reduced dot product computation circuit Physics 30 Active
US10592732B1 Probabilistic loss function for training network with triplets Physics 21 Active
US11049013B1 Encoding of weight values stored on neural network inference circuit Physics 17 Active
US11610154B1 Preventing overfitting of hyperparameters during training of network Physics 14 Active
US11468145B1 Storage of input values within core of neural network inference circuit Physics 14 Active
US11170289B1 Computation of neural network node by neural network inference circuit Physics 13 Active
US11210586B1 Weight value decoder of neural network inference circuit Physics 12 Active
US11347297B1 Neural network inference circuit employing dynamic memory sleep Physics 11 Active
US10586151B1 Mitigating overfitting in training machine trained networks Physics 8 Active
US11615322B1 Compiler for implementing memory shutdown for neural network implementation configuration Physics 8 Active
US11568227B1 Neural network inference circuit read controller with multiple operational modes Physics 7 Active
US11531879B1 Iterative transfer of machine-trained network inputs from validation set to training set Physics 6 Active
US10671888B1 Using batches of training items for training a network Physics 6 Active
US11586910B1 Write cache for neural network inference circuit Physics 5 Active
US11222257B1 Non-dot product computations on neural network inference circuit Physics 5 Active
US11625585B1 Compiler for optimizing filter sparsity for neural network implementation configuration Physics 5 Active
US11295200B1 Time-multiplexed dot products for neural network inference circuit Physics 4 Active
US10867247B1 Machine learning through multiple layers of novel machine trained processing nodes Physics 4 Active
US11113603B1 Training network with discrete weight values Physics 4 Active
US11205115B1 Neural network inference circuit Physics 4 Active
US11475310B1 Training network to minimize worst-case error Physics 4 Active
US10742959B1 Use of machine-trained network for misalignment-insensitive depth perception Electricity 4 Active
US10863127B1 Compressive sensing based image capture using multi-lens array Electricity 4 Active

Source: USPTO / EPO open patent data. Counts and citation impact are objective bibliographic measures.