Signal analysis systems and methods for features extraction and interpretation thereof

Abstract

The invention relates to the development of descriptive and normative systems based on sensor data, and relates to machine learning tasks such as classification and regression. Any such system development requires participation of different interested parties to obtain features. These typically obtained features are uninterpretable for 1-D sensor signals. Embodiments of the present disclosure provide systems and methods capable of performing signal analysis for feature extraction and interpretation thereof, wherein an input is original signal data. In the original signal data, the origin of the feature is depicted as signal data and mapped to domain/application knowledge. Features are extracted using a deep learning network, and a machine learning (ML) model is implemented for sensor dataanalysis to perform predicted causal relationship analysis. A layer (e.g., the last layer) of the deep network includes automated export features that can be used for ML tasks. Parameter adjustment isperformed according to a set of features recommended by the system to determine the performance of the considered system (or application).

Description

[0001] Cross References to Related Applications [0002] This application claims priority based on Indian Patent Application No. 201721039304 filed on November 3, 2017. technical field [0003] The present disclosure relates generally to data analysis systems and, more particularly, to signal analysis systems and methods of feature extraction and interpretation for recommendation. Background technique [0004] The development of sensor data based descriptive and prescriptive systems involves machine learning tasks such as classification and regression. Any such system development requires the participation of different stakeholders to understand the problem domain and generate models for performing causal analysis, signal processing (SP), machine learning (ML) techniques for data analysis and finally deployment by developers solution. Now, the problem with developing such a system is that each stakeholder speaks its own language and terminology. In a related survey, it wa...

AI Technical Summary

Problems solved by technology

In a related survey, it was found that the most difficult task in the above workflow is feature engineering (combination of feature enumeration/extraction and feature selection),

Images