Towards High Performance Network Training with Noisy Label Datasets

Abstract

Creating large amounts of labeled data to train neural networks is an obstacle to applying deep learning to new applications. Heuristic methodsfor labeling data typically produce a significant fraction of mislabeled samples. This report describes some methods in the literature that find thefraction of noisy labeled datasets that are probably labeled correctly and our efforts to improve on these methods. The method we describe and testis called NoisyLabel Correcting Cross Validation. The results of this method proved inferior to the INCV method in the literature but the newunderstandings learned from this effort inspired two new methods: the generalized sensitivity analysis and the soft lables approaches. Our futureplans include testing these methods.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 10, 2020
Accession Number
AD1100433

Entities

People

  • Elizabeth A. Gilmour
  • Leslie N. Smith

Organizations

  • United States Naval Research Laboratory

Tags

Communities of Interest

  • Autonomy
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Accuracy
  • Algorithms
  • Artificial Intelligence
  • Artificial Intelligence Software
  • Computing System Architectures
  • Deep Learning
  • Errors
  • Heuristic Methods
  • Information Systems
  • Iterations
  • Learning
  • Machine Learning
  • Military Research
  • Network Architecture
  • Neural Networks
  • Neurobehavioral Manifestations
  • Training

Fields of Study

  • Computer science

Readers

  • Aerospace Research.
  • Neural Network Machine Learning.

Technology Areas

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