Convolutional neural net face recognition works in non-human-like ways
Open Access
- 7 October 2020
- journal article
- research article
- Published by The Royal Society in Royal Society Open Science
- Vol. 7 (10)
Abstract
Convolutional neural networks (CNNs) give the state-of-the-art performance in many pattern recognition problems but can be fooled by carefully crafted patterns of noise. We report that CNN face recognition systems also make surprising ‘errors'. We tested six commercial face recognition CNNs and found that they outperform typical human participants on standard face-matching tasks. However, they also declare matches that humans would not, where one image from the pair has been transformed to appear a different sex or race. This is not due to poor performance; the best CNNs perform almost perfectly on the human face-matching tasks, but also declare the most matches for faces of a different apparent race or sex. Although differing on the salience of sex and race, humans and computer systems are not working in completely different ways. They tend to find the same pairs of images difficult, suggesting some agreement about the underlying similarity space.Keywords
Funding Information
- Engineering and Physical Sciences Research Council (EP/N007743/1)
This publication has 23 references indexed in Scilit:
- Deep neural networks are easily fooled: High confidence predictions for unrecognizable images2015
- Unfamiliar face matching: Pairs out‐perform individuals and provide a route to trainingBritish Journal of Psychology, 2014
- Variability in photos of the same faceCognition, 2011
- The processing of invariant and variant face cues in the Garner Paradigm.Emotion, 2011
- The Glasgow Face Matching TestBehavior Research Methods, 2010
- Soft Biometric Traits for Personal Recognition Systems2004
- Categorical perception of facial gender information: Behavioural evidence and the face-space metaphorVisual Cognition, 2001
- Prototyping and transforming facial textures for perception researchIEEE Computer Graphics and Applications, 2001
- Low-dimensional representation of faces in higher dimensions of the face spaceJournal of the Optical Society of America A, 1993
- Sex Discrimination: How Do We Tell the Difference between Male and Female Faces?Perception, 1993