Generative Network Imputation for Time Series Data Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Time series data often contains missing values due to measurement failures, incomplete capture, or intentional deletion, which hinders analysis, particularly for machine learning algorithms that require complete datasets.
Innovation Solution
A machine learning-based approach using a generative network, specifically a convolutional neural network (CNN), is employed to impute missing values in time series datasets, enabling the reconstruction of complete datasets for effective pattern classification and remedial action initiation in IT infrastructure monitoring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional imputation methods (Lasso, mean imputation) are used to handle missing values, then the process is simple and fast, but the reconstruction accuracy and reliability are insufficient
Solution Approach 1:
The patent replaces conventional statistical imputation methods (Lasso, mean imputation) with a deep learning-based generative network. This substitution transitions from traditional mechanical/statistical approaches to an intelligent system that can capture complex temporal dependencies and patterns in time series data, thereby improving imputation accuracy while accepting increased algorithmic complexity
Solution Approach 2:
The patent transforms the imputation problem by changing the approach from direct value prediction to generative modeling. The generative network learns the underlying data distribution and generates plausible missing values based on learned patterns, rather than relying on simple statistical formulas. This parameter change in the methodological approach enables superior reconstruction performance
2Adaptability or versatility
If machine learning algorithms are applied to time series data with missing values, then analysis capabilities are enhanced, but the algorithms fail to produce reliable results due to incomplete data
Solution Approach 1:
The patent applies preliminary action by reconstructing the complete time series dataset before applying machine learning analysis algorithms. The generative network pre-processes the incomplete data by imputing missing values based on learned temporal patterns, ensuring that subsequent analysis algorithms receive complete and reliable input data, thereby enabling both enhanced analysis capability and maintained reliability
3Measurement precision
If complete datasets are required for machine learning analysis, then analysis accuracy improves, but data loss occurs due to inability to process incomplete data
Solution Approach 1:
The patent introduces an intermediary solution by using a generative network as a bridge between incomplete raw data and the requirements of machine learning analysis algorithms. The network acts as a mediator that transforms incomplete data into complete reconstructed datasets, preventing data loss while maintaining analysis accuracy by generating plausible missing values rather than discarding incomplete records
Data Source
AI summary
An apparatus comprises a processing device configured to obtain a time series dataset having missing values, the time series dataset comprising monitoring data associated with one or more assets. The processing device is also configured to generate, utilizing a machine learning algorithm, a reconstructed time series dataset having imputed values for the missing values in the obtained time series dataset, the machine learning algorithm comprising a generative network implementing inverse network parameter determination for network parameters of the generative network. The processing device is further configured to classify patterns in the obtained time series dataset utilizing the reconstructed time series dataset, to select remedial actions for controlling at least one of the one or more assets based at least in part on the classified patterns in the obtained time series dataset, and to initiate the selected remedial actions to control the at least one asset.