Evaluating Deep Learning Biases Based on Grey-Box Testing Results

J. Jenny Li, Thayssa Silva, Mira Franke, Moushume Hai, Patricia Morreale

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

6 Scopus citations

Abstract

The very exciting and promising approaches of deep learning are immensely successful in processing large real world data sets, such as image recognition, speech recognition, and language translation. However, much research discovered that it has biases that arise in the design, production, deployment, and use of AI/ML technologies. In this paper, we first explain mathematically the causes of biases and then propose a way to evaluate biases based on testing results of neurons and auto-encoders in deep learning. Our interpretation views each neuron or autoencoder as an approximation of similarity measurement, of which grey-box testing results can be used to measure biases and finding ways to reduce them. We argue that monitoring deep learning network structures and parameters is an effective way to catch the sources of biases in deep learning.

Original languageEnglish
Title of host publicationIntelligent Systems and Applications - Proceedings of the 2020 Intelligent Systems Conference IntelliSys Volume 1
EditorsKohei Arai, Supriya Kapoor, Rahul Bhatia
PublisherSpringer
Pages641-651
Number of pages11
ISBN (Print)9783030551797
DOIs
StatePublished - 2021
EventIntelligent Systems Conference, IntelliSys 2020 - London, United Kingdom
Duration: 3 Sep 20204 Sep 2020

Publication series

NameAdvances in Intelligent Systems and Computing
Volume1250 AISC
ISSN (Print)2194-5357
ISSN (Electronic)2194-5365

Conference

ConferenceIntelligent Systems Conference, IntelliSys 2020
Country/TerritoryUnited Kingdom
CityLondon
Period3/09/204/09/20

Keywords

  • Bias measurement
  • Deep learning
  • Deep learning evaluation
  • Mathematical interpretation
  • Neural Networks

Fingerprint

Dive into the research topics of 'Evaluating Deep Learning Biases Based on Grey-Box Testing Results'. Together they form a unique fingerprint.

Cite this