Sequential Learning Using Image-Based Classification

Abstract

Machine learning for time series, or sequential learning, has been a growing field due to interests in medicine, weather, stocks, and more. We present an image-based scenario classification solution to a dataset with highly nonuniformly sampled data. Scenario data are obtained through software and are used as a starting point for data processing. First, we fill in data using a number of samples determined from averaging adjacent sampling rates of groups of data, we call this data the "dead zone." Next, groupings of output data or dead zones are then given a temporal encoding, denoting dead zones with zeros and output data with a linear encoding. Finally, we transform the scenario by feature into 2D channels of a full image using signal processing techniques such as the Constant Q-Transform (CQT).

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 2022
Accession Number
AD1167189

Entities

People

  • Kevin M. Lee

Organizations

  • Ohio State University

Tags

DTIC Thesaurus Topics

  • Accuracy
  • Air Force
  • Air Force Facilities
  • Air Force Research Laboratories
  • Algorithms
  • Artificial Intelligence Software
  • Classification
  • Coding
  • Computers
  • Convolutional Neural Networks
  • Data Processing
  • Data Visualization
  • Deep Learning
  • Dimensionality Reduction
  • Image Classification
  • Image Processing
  • Information Science
  • Learning
  • Machine Learning
  • Neural Networks
  • Sampling
  • Signal Processing

Fields of Study

  • Computer science

Readers

  • Approximation Theory.
  • Computer Vision.
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Neural Networks