Memorism Processor

(Set Operating Processor)

The Set Operating Processor (SOP) is a processor specialized in detecting the desired pattern from images or texts.

What is SOP?

The SOP is a processor specialized in detecting the desired pattern from images or texts. It can be used in any kind of situation, such as defect and abnormality detection, face recognition, and on-vehicle image recognition systems.

About pattern recognition

Pattern recognition

Pattern recognition is the starting point of knowledge processing.
The most basic method of pattern recognition is pattern matching, but CPUs (GPUs) are not good (i.e., are inefficient) at pattern matching.

Source: Nikkei Electronics

Hardware pattern matching

Consequently, pattern recognition must rely on complex algorithms or deep learning.
However, through high-speed pattern matching processing by hardware, it is possible to revolutionize recognition technology.

SOP’s pattern matching and chip configuration

The SOP’s pattern matching executes set operations that include spatial conditions (data location).
By making a color and its position ambiguous, it conducts human-like recognition.
With this concept (definition), the SOP can execute pattern matching of any kind of information by hardware.

In the pattern matching with the SOP, operations ① to ③ are executed in parallel for all pixels, resulting in an ultra high-speed operation.

Pattern matching with the SOP is executed as a set operation of all pixels fully in parallel (pattern matching in the broad sense).
A process that takes dozens of milliseconds in software processing can reduce in 0.5 to a few microseconds.

SOP application examples

The SOP can be applied to detect various objects, as well as character and face recognition.
With the SOP, it is possible to execute operations up to 10,000 times faster than a laptop while consuming as little as 1W of power.
It can be used as a new AI technology that can replace machine learning and deep learning in the future.

Application examples

Object detection: Edge detection, corner detection, area detection
Character recognition: Pattern detection
Face recognition: Pattern detection
General recognition: Combination detection

Object detection: Edge detection
Face recognition: Pattern detection
Object detection: Corner detection
Object detection: Area detection
Character recognition: Pattern detection
SOP’s characteristic detection, recognition, and extraction capabilities

Color and grayscale: Arbitrary edges, arbitrary corners, arbitrary areas, arbitrary patterns