Neural Exploration and Research Laboratory (NERL)

Image of NERL_Tri2.png

Sandia’s Neural Exploration & Research Laboratory enables researchers to explore the boundaries of neural computation. The research conducted in the lab evaluates what is possible with neural hardware and software for national security benefit and the advancement of basic research.

Current areas of NERL research include:

  • Machine Learning and Deep Learning
  • Neural Computing and Applications to High Performance Computing (HPC)
  • Computational and Mathematical Neuroscience
  • Reinforcement Learning
  • Remote Sensing
  • Hyperspectral Imaging

NERL consists of a variety of neuromorphic hardware & neural algorithms providing a testbed facility for comparative benchmarking and new architecture exploration. The lab’s hardware resources include:

Cognimem CM1K
Cognimem CM1K
NVidia Jetson TX1
NVidia Jetson TX1
High-end GPU workstations
iniLabs DAVIS240C DVS
iniLabs DAVIS240C DVS
Intel / Movidius Neural Compute Stick
Intel / Movidius Neural Compute Stick
Xilinx PYNQ FPGA
Xilinx PYNQ FPGA
SpiNNaker 48 Node Board
SNL STPU (on Nallatech FPGA)
IBM TrueNorth (remote access)
IBM TrueNorth NS16e (remote access)
IBM TrueNorth NS16e (remote access)
Xilinx ZYNQ-7000 ZC706 FPGA
Xilinx ZYNQ-7000 ZC706 FPGA

Featured Project

Mapping particle tracking algorithms onto IBM TrueNorth

The IBM TrueNorth Chip offers a million.  To complement the efforts putting deep learning on TrueNorth, we have developed an approach to map a spiking cross-correlation algorithm useful for tracking the movement of objects, such as particle image velocimetry (Severa et al., ICRC 2016).