cognimem logo, hardwired parallel architecture, scalable, low power, decision space mapping, CogniBlox, CBX, 4K neurons, data mining, analytics, cognitive memory, sensor input, 4 CM1K's, 4096 cognitive memories, CM1K, 1K Neuron Chip, CM1K speed, High-speed pattern recognition, parallel architecture, non-linear classifier, CogniMem Architecture, CPU, HPC architecture, DSP, storage, CogniMem, Cognitive Computing, Artificial intelligence,clustering,data mining,expandable neural network,global sensing,Machine learning,neural network chip,neuron chip,neuronal processor,neurons,parallel neural network,pattern matching,pattern recognition chip,Restricted Coulomb Energy,smart sensors,super computing,trainable neural network
  • View Cart
  • Contact Us

Bibliography and other links

In 2011, CogniMem Technologies Inc. is established and headquartered in Folsom, California; its mission is to market the hardware based neural network technology and develop the next generation of pattern recognition chips targeting usage models in sensory devices as well as cognitive computing systems.

In 2010 - 1993, Intel/Nestor with DARPA co-develop the NI1000 utlizing the RBF non-linear classifier. DARPA Report Around the same time, IBM validates and patents a similar architecture with a project utilizing a RBF non-linear classifier.

US Patent # Patent Description
US-5717832 Improved neuron circuit architecture
US-5710869 Daisy chain circuit for serial connection of neuron circuits
US-5701397 Circuit for pre-charging a free neuron circuit
US-5740326 Circuit for searching/sorting data in neural networks

In 1992, M. Holler et al of Intel Corp publishes in conjunction with Nestor and DARPA
"A High Performance Adaptive Classifier using Radial Basis Functions", submitted to the Government Microcircuit Applications Conference in Las Vegas, Nevada. Wherein a 1024 neuron RBF/RCE VLSI hardware component is proposed.