Reservoir Computing for Hardware-Friendly Time-Series Outlier Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection models for time-series signals have complex algorithms and hardware configurations, making them difficult to realize as hardware for outlier detection.
Innovation Solution
A reservoir computer with an input layer, reservoir main unit, read-out, learning unit, norm calculation unit, and determination unit is used to calculate the norm of a weight vector and determine outliers based on adaptive filtering and systolic array processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pattern matching between observed signal and teacher data is executed in time region or frequency region, then outlier detection capability is improved, but algorithm complexity and hardware configuration complexity increase
Solution Approach 1:
The patent replaces complex pattern matching algorithms with a reservoir computer system that uses adaptive filtering and norm calculation. Instead of executing complicated pattern recognition in time or frequency domains, the system substitutes a hardware-realizable architecture comprising a reservoir main unit with neurons and synapses, a learning unit with adaptive filter, and a norm calculation unit that computes the norm of weight vectors to detect outliers.
Solution Approach 2:
The patent changes the detection parameter from pattern matching results to the norm of weight vectors. By monitoring changes in the norm value of weight vectors during adaptive filtering, the system can detect outliers without requiring complex pattern recognition algorithms. This parameter transformation simplifies the hardware configuration while maintaining detection capability.
2Measurement precision
If complex anomaly detection algorithms are implemented, then detection accuracy is improved, but ease of manufacture and hardware realization deteriorate
Solution Approach 1:
The patent segments the anomaly detection function into distinct modular components: a reservoir main unit containing neurons and synapses for signal processing, a learning unit with adaptive filter for weight adjustment, and a norm calculation unit for outlier detection. This segmentation allows each module to be independently manufactured and tested, improving ease of hardware realization while maintaining overall detection accuracy.
Solution Approach 2:
The reservoir computer system performs self-adjustment through adaptive filtering where the learning unit automatically updates weight vectors based on input signals. The norm calculation unit continuously monitors weight vector norms to detect outliers without requiring external complex control systems. This self-service capability reduces manufacturing complexity by eliminating the need for sophisticated external control hardware.
Data Source
AI summary
An outlier detection device includes a reservoir computer having an input layer, a reservoir main unit including neurons connected by synapses, and a read-out configured to calculate and output an inner product of a weight vector and an activity value vector, each element of which is an activity value output from each of neurons based on an input to the input layer, a learning unit configured to acquire an observed signal, calculate an error between the inner product and the observed signal, and update the weight vector using a value obtained by applying an adaptive filter to the error, a norm calculation unit configured to sequentially calculate a norm of the weight vector updated by the learning unit, and a determination unit configured to determine whether an outlier is included in the observed signal based on at least one of the norms calculated by the norm calculation unit.