Office of Technology Transfer – University of Michigan

Sparse Coding with RRAM Crossbar

Technology #7427

Questions about this technology? Ask a Technology Manager

Download Printable PDF

Wei Lu
Managed By
Joohee Kim
Licensing Specialist, Physical Sciences & Engineering 734-764-8202
Patent Protection
US Patent Pending

Memristors, also referred to as resistive random access memory (RRAM), are resistors that contain memory in the sense that the properties of the resistor can be changed by applying different voltages. The global market for circuit devices with memory, e.g., RRAM, is rapidly growing, and it is projected to be $529.6M in 2021 [1]. RRAM-based devices are promising for computing applications, in particular, machine learning applications and neuromorphic computing applications. Machine learning can roughly be described as the statistical analysis of existing data to make predictions or to learn characteristics of the existing dataset. Neuromorphic computing is the concept of using integrated circuits to mimic systems present in a biological nervous system, e.g., a portion of a brain. Machine learning algorithms have been implemented using neuromorphic computing systems containing standard computing devices, which can require significant power to execute the algorithm. However, developing specialized computing hardware, i.e., application specific integrated circuits (ASIC), for machine learning algorithms using RRAM can reduce the required power.

Neuromorphic Computing for Machine Learning using RRAM

A specialized neuromorphic computing device has been designed and constructed to implement machine learning algorithms using arrays of RRAM memory. This device is able to achieve faster execution of machine learning algorithms with reduced power usage compared to digital computing resources. The supporting data includes testing on a prototype to demonstrate the device’s efficacy as well as benchmarking of the speed and energy uses versus a digital device. The RRAM-based design has a wide range of applications including driverless vehicles, computer vision, image processing, control systems, and signal processing. The device impacts the defense, automotive, industrial process control, healthcare, and other sectors.


  • Defense
  • Driverless vehicles
  • Computer vision
  • Control systems
  • Image processing
  • Signal processing


  • Higher computation speed
  • Reduced energy usage

[1] Global Markets for Circuit Elements with Memory: Memresistors, Memcapacitors, Meminductors. SMC095B. BCC Research. 2016.