Visualization Method Using Conversion Vectors for Time Series Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data visualization techniques, such as multidimensional scaling, struggle to maintain a consistent positional relationship between data points when new data is added, making it difficult to continuously check the relation between newly acquired and already acquired data, and fail to distinguish between different types of abnormalities in time series data.
Innovation Solution
A computer-readable medium stores a visualization program that generates conversion vectors through dimensional compression of input data, allowing for the plotting of these vectors while maintaining their positional relation, using techniques like autoencoders or main component analysis to stabilize the positional relationship and facilitate the differentiation of data types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multidimensional scaling is used to visualize time series data, then different types of abnormalities can be distinguished in a multidimensional space, but the positional relationship between data points changes when new data is added, making it difficult to continuously monitor data relationships
Solution Approach 1:
The patent pre-calculates and stores conversion vectors for reference data before monitoring begins. When new data arrives, the system compares it against these pre-established conversion vectors rather than recalculating all positional relationships, thus maintaining stability while enabling continuous monitoring
Solution Approach 2:
The patent transforms time series data into a different dimensional representation using conversion vectors that encode multiple temporal dimensions. This dimensional transformation allows abnormality differentiation while the conversion process itself maintains positional relationship stability through consistent transformation rules
2Reliability
If traditional abnormality detection methods are used to determine whether abnormalities occur, then abnormal states can be detected, but it is difficult to determine whether occurrence causes of different abnormalities are equal
Solution Approach 1:
The patent applies dimensional transformation to encode temporal patterns and cause information into the conversion vectors. By representing data in this transformed dimensional space, the system preserves cause differentiation information that would be lost in traditional scalar anomaly detection, enabling both reliable detection and cause identification
Data Source
AI summary
A non-transitory computer-readable recording medium stores therein a visualization program that causes a computer to execute a process including: generating a plurality of conversion vectors, from a plurality of vectors generated from plural pieces of input data, by a dimensional compression in a positional relation between the plurality of vectors; and plotting the plurality of conversion vectors.