HARDWARE

Today’s flavors of NeuroMem neurons…

 

CM1K chip

  • 1024 neurons
  • 256 bytes of memory per neuron
  • Cascadable
  • I2C controller (optional use)
  • Recognition stage (optional use)

Intel Quark SE

  • 128 neurons
  • 128 bytes of memory per neuron
  • Component of an SOC also combining an x86 MCU and a sensor subsystem

NM500 chip

  • 576 neurons
  • 256 bytes of memory per neuron
  • Cascadable
  • Wafer scale package

Evaluation platforms…

  • CurieNeurons library and…
  • Arduino/Genuino 101
  • Other Intel Curie module based board such as TinyTile
  • BrainCard rev2 (Q4, 2017)

Arduino/Genuino 101

Unleash the neurons of the Arduino/Genuino101 board

Using our CurieNeurons library, you can gain full access to the 128 NeuroMem neurons of the Intel Quark SE and take advantage of the Arduino eco-system to learn and recognize patterns and signals of your choice. These neurons behave exactly like the neurons of the CM1K chip with the exception that they can only process patterns of 128 bytes instead of 256 bytes.

NEUROSTACK

Scalable Pattern Recognition Accelerator, easy to connect via USB

NeuroStack is a pattern learning and recognition accelerator featuring a network of 4 CM1K neurons (4096) which can be expanded by stacking additional NeuroStack boards. Typical applications include text, data and image analytics. Patterns can be learned and classified at high speed whether they come from text, network packets, images, audio signal, biosensors, and more. The on-board Field Programmable Gate Array can be reconfigured to execute specific recognition engines.

BRAINCARD

Optional RaspiCam mounted on BrainCard

The artificial intelligence shield for Arduino and Raspberry PI

Use your Arduino shields to collect audio, video and other signals and the CM1K neurons to learn relevant signatures and give you instantaneous responses to ring alarms, activate relays, control a motor and more.

Turn your Raspberry PI into a smart camera or data analytics device with the BrainCard. Teach the neurons to recognize objects while discarding false alarms. Detect people or events and transfer and record data only when something of interest occurs.

gv