Analog Neural Computing with Super-resolution Memristor Crossbars
This addresses a key bottleneck in neuromorphic hardware by improving analog neural network implementation, though it is an incremental advancement in memristor-based computing.
The paper tackles the problem of limited conductance levels in memristor crossbars due to device non-idealities and aging, by introducing a super-resolution technique using multiple memristors per node to generate a wider range of unique conductance values, which enhances analog neural network layers for neuromorphic computing.
Memristor crossbar arrays are used in a wide range of in-memory and neuromorphic computing applications. However, memristor devices suffer from non-idealities that result in the variability of conductive states, making programming them to a desired analog conductance value extremely difficult as the device ages. In theory, memristors can be a nonlinear programmable analog resistor with memory properties that can take infinite resistive states. In practice, such memristors are hard to make, and in a crossbar, it is confined to a limited set of stable conductance values. The number of conductance levels available for a node in the crossbar is defined as the crossbar's resolution. This paper presents a technique to improve the resolution by building a super-resolution memristor crossbar with nodes having multiple memristors to generate r-simplicial sequence of unique conductance values. The wider the range and number of conductance values, the higher the crossbar's resolution. This is particularly useful in building analog neural network (ANN) layers, which are proven to be one of the go-to approaches for forming a neural network layer in implementing neuromorphic computations.