User Tools

Site Tools

Action unknown: copypageplugin__copy

8a.004.UVA - General Techniques for Explaining/Interpreting Deep Neural Networks

Project - Summary

  • Much of the recent advancement in machine learning has been through neural networks (NNs), and in particular deep learning (DL) has shown record-breaking performance across many domains
  • In some fields where there is a human element involved, e.g. in medicine and finance, there can be legal or ethical considerations that require decisions to be interpretable/explainable. This requirement often prohibits the use of NNs and other so-called 'black-box' models.
  • There have been some proposed techniques to help explain the internal workings of NNs, many focused on image/visual recognition.
  • One of these methods known as feature occlusion shows promise for non-image and multi-modal data.
  • We propose to investigate the use of feature occlusion and potentially develop other techniques for use as general methods for explaining and interpreting deep learning models.

Project - Team

Team Member Role Email Phone Number Academic Site/IAB
Peter Beling PI (804) 982-2066 University of Virginia
Stephen Adams Research Scientist (757) 870-4954 University of Virginia
Alex Langevin Student Not available University of Virginia
Funded by: U-ANGEL

Project - Novelty of Approach

This project seeks to determine the feasibility and performance of feature occlusion and other techniques as general methods of explainable AI with test case implementation in IAB member defined settings

Project - Deliverables

1 Source code
2 Report of results and list of significant predictive features and interactions

Project - Benefits to IAB

  • Improved model accuracy while maintaining interpretability
  • Ability to explain model results where legally or ethically required
  • In commercial or engineering settings, improved business intelligence and diagnostic data gathering

Project - Documents

FilenameFilesizeLast modified
8a.004.uva_final_report.docx599.7 KiB2020/10/31 11:16
8a.004.uva_year_8_project_proposal_revised_09.27.2019.docx54.1 KiB2019/09/27 11:38
8a.004.uva_explainable_ai_year_8_quad_chart_beling.pptx59.5 KiB2019/09/25 17:12

Life Form Feedback

Year 8 Project Poster Session Feedback (Fall 2019)
Real name Great Progress On Course Needs Change Off Course Abstain
Kimmo Valtonen (kimmo.valtonen)     
Sumit Shah (sumit.shah)     
projects/year8/8a.004.uva.txt · Last modified: 2019/10/21 17:12 by matt.delcambre