A5017724364- Face recognition and analysis
- Face Recognition and Perception
- Face and Expression Recognition
- Evolutionary Psychology and Human Behavior
- Biometric Identification and Security
- Complex Systems and Decision Making
- Emotion and Mood Recognition
- Forensic Anthropology and Bioarchaeology Studies
- Expert finding and Q&A systems
- Deception detection and forensic psychology
- Team Dynamics and Performance
- Cognitive Science and Mapping
- Conflict Management and Negotiation
- Emotional Intelligence and Performance
- Advanced Text Analysis Techniques
- Names, Identity, and Discrimination Research
- Complex Network Analysis Techniques
University of California, Irvine
2022-2024
The University of Texas at Dallas
2018-2023
Significance This study measures face identification accuracy for an international group of professional forensic facial examiners working under circumstances that apply in real world casework. Examiners and other human “specialists,” including forensically trained reviewers untrained superrecognizers, were more accurate than the control groups on a challenging test identification. Therefore, specialists are best available solution to problem We present data comparing state-of-the-art...
Previous generations of face recognition algorithms differ in accuracy for images different races (race bias). Here, we present the possible underlying factors (data-driven and scenario modeling) methodological considerations assessing race bias algorithms. We discuss data-driven (e.g., image quality, population statistics, algorithm architecture), modeling that consider role "user" threshold decisions demographic constraints). To illustrate how these issues apply, data from four (a...
Face identification is more accurate when people collaborate in social dyads than they work alone (Dowsett & Burton, 2015, Br. J. Psychol., 106, 433). Identification accuracy also increased the responses of two are averaged for each item to create a 'non-social' dyad (White, Kemp, Jenkins, 2013, Appl. Cogn. 27, 769; White et al., Proc. R. Soc. B Biol. Sci., 282, 20151292). Does collaboration add benefits response averaging face identification? We compared individuals, dyads, and non-social...
Abstract Measures of face-identification proficiency are essential to ensure accurate and consistent performance by professional forensic face examiners others who perform tasks in applied scenarios. Current tests rely on static sets stimulus items so cannot be administered validly the same individual multiple times. To create a test, large number “known” difficulty must assembled. Multiple equal can constructed then using subsets items. We introduce Triad Identity Matching (TIM) test...
Face identification is particularly prone to error when individuals identify people of a race other than their own - phenomenon known as the other-race effect (ORE). Here, we show that collaborative "wisdom-of-crowds" decision-making substantially improves face accuracy for own- and faces over working alone. In two online experiments, East Asian White recognized part dyad. Collaboration never proved more beneficial in social setting individual decisions were combined computationally. The...
Deep convolutional neural networks (DCNNs) have achieved human-level accuracy in face identification (Phillips et al., 2018), though it is unclear how accurately they discriminate highly-similar faces. Here, humans and a DCNN performed challenging face-identity matching task that included identical twins. Participants (
Forensic facial professionals have been shown in previous studies to identify people from frontal face images more accurately than untrained participants when given 30 seconds per pair. We tested whether this superiority holds challenging conditions. Two groups of forensic (examiners, reviewers) and were three lab-based tasks: other-race identification, disguised memory. For on same-race faces, examiners superior controls; different-race controls performed comparably. Examiners for...
ABSTRACT Forensic facial professionals have been shown in previous studies to identify people from frontal face images more accurately than untrained participants when given 30 s per pair. We tested whether this superiority holds challenging conditions. Two groups of forensic (examiners, reviewers) and were three lab‐based tasks: other‐race identification, disguised memory. For on same‐race faces, examiners superior controls; different‐race controls performed comparably. Examiners for...
Collaborative problem-solving (CPS) is a vital skill used both in the workplace and educational environments. CPS useful tackling increasingly complex global, economic, political issues considered central 21st century skill. The connected global community presents fruitful opportunity for creative collaborative interactions solutions that involve diverse perspectives. Unfortunately, women underrepresented minorities (URMs) often face obstacles during hinder their key participation these...
Previous generations of face recognition algorithms differ in accuracy for images different races (race bias). Here, we present the possible underlying factors (data-driven and scenario modeling) methodological considerations assessing race bias algorithms. We discuss data driven (e.g., image quality, population statistics, algorithm architecture), modeling that consider role "user" threshold decisions demographic constraints). To illustrate how these issues apply, from four (a...
Measures of face-identification proficiency are essential to ensure accurate and consistent performance by professional forensic face examiners others who perform tasks in applied scenarios. Current tests rely on static sets stimulus items, so, cannot be administered validly the same individual multiple times. To create a test, large number items "known" difficulty must assembled. Multiple equal can constructed then using subsets items. We introduce Triad Identity Matching (TIM) test...
The Other-Race Effect (ORE) refers to the well-known finding that people recognize own-race faces more accurately than other-race faces. Is it possible reduce ORE? Here, we examined role of learning context, in combination with multiple-image training on recognition accuracy for own-and East Asian and Caucasian participants saw images each identity either a contiguous order (multiple an grouped together) or distributed dispersed randomly throughout set). Participants learned from four highly...
Previous generations of face recognition algorithms show differences in accuracy for faces different races (race bias) (O’Toole et al., 1991; Furl 2002; Givens 2004; Phillips 2011; Klare 2012). Whether newer deep convolutional neural networks (DCNNs) are also race biased is less well studied (El Khiyari 2016; Krishnapriya 2019). Here we present methodological considerations measuring underlying bias. We consider two key factors: data-driven and scenario modeling. Data-driven factors driven...
Accurate estimates of face-identification ability are crucial in applied forensic settings. Current datasets often large and uncalibrated, making them sub-optimal for pre- post-training evaluations. To optimize efficient accurate performance assessments, small sets well-labelled test items needed. However, item-wise measures cannot be to the common task identity matching image pairs, because either “matched” or “non-matched” identities. Therefore, this case, an item response confounds...
Deep convolutional neural networks (DCNNs) have achieved human-level accuracy in face identification (Phillips et al., 2018), though it is unclear how accurately they discriminate highly-similar faces. Here, humans and a DCNN performed challenging face-identity matching task that included identical twins. Participants (N=87) viewed pairs of images three types: same-identity, general imposter (different identities from similar demographic groups), twin (identical siblings). The was to...
Deep convolutional neural networks (DCNNs) have achieved human-level accuracy in face identification (Phillips et al., 2018), though it is unclear how accurately they discriminate highly similar faces. Here, we compared human and DCNN performance on a challenging face-identification task that includes identical twins. Participants (N=29) completed which viewed pairs of high-quality, frontal images three conditions: match (same identity, N=40), general imposter (different identities from...
Collaborative "wisdom-of-crowds" decision making improves face identification accuracy over individuals working alone. We examined whether collaboration both own- and other-race identification. In Experiment 1, participants completed an online face-identification task on their own with a same-race partner (East Asian dyads, N = 27; Caucasian dyad, 31). decisions were as part of social dyad (completing the together) non-social (individual scores fused independently). Social improved equally....