-
Unexplored Faces of Robustness and Out-of-Distribution: Covariate Shifts in Environment and Sensor Domains
Authors:
Eunsu Baek,
Keondo Park,
Jiyoon Kim,
Hyung-Sin Kim
Abstract:
Computer vision applications predict on digital images acquired by a camera from physical scenes through light. However, conventional robustness benchmarks rely on perturbations in digitized images, diverging from distribution shifts occurring in the image acquisition process. To bridge this gap, we introduce a new distribution shift dataset, ImageNet-ES, comprising variations in environmental and…
▽ More
Computer vision applications predict on digital images acquired by a camera from physical scenes through light. However, conventional robustness benchmarks rely on perturbations in digitized images, diverging from distribution shifts occurring in the image acquisition process. To bridge this gap, we introduce a new distribution shift dataset, ImageNet-ES, comprising variations in environmental and camera sensor factors by directly capturing 202k images with a real camera in a controllable testbed. With the new dataset, we evaluate out-of-distribution (OOD) detection and model robustness. We find that existing OOD detection methods do not cope with the covariate shifts in ImageNet-ES, implying that the definition and detection of OOD should be revisited to embrace real-world distribution shifts. We also observe that the model becomes more robust in both ImageNet-C and -ES by learning environment and sensor variations in addition to existing digital augmentations. Lastly, our results suggest that effective shift mitigation via camera sensor control can significantly improve performance without increasing model size. With these findings, our benchmark may aid future research on robustness, OOD, and camera sensor control for computer vision. Our code and dataset are available at https://github.com/Edw2n/ImageNet-ES.
△ Less
Submitted 25 April, 2024; v1 submitted 24 April, 2024;
originally announced April 2024.
-
Perceived community alignment increases information sharing
Authors:
Elisa C. Baek,
Ryan Hyon,
Karina López,
Mason A. Porter,
Carolyn Parkinson
Abstract:
Information sharing is a ubiquitous and consequential behavior that has been proposed to play a critical role in cultivating and maintaining a sense of shared reality. Across three studies, we tested this theory by investigating whether or not people are especially likely to share information that they believe will be interpreted similarly by others in their social circles. Using neuroimaging whil…
▽ More
Information sharing is a ubiquitous and consequential behavior that has been proposed to play a critical role in cultivating and maintaining a sense of shared reality. Across three studies, we tested this theory by investigating whether or not people are especially likely to share information that they believe will be interpreted similarly by others in their social circles. Using neuroimaging while members of the same community viewed brief film clips, we found that more similar neural responding of participants was associated with a greater likelihood to share content. We then tested this relationship using behavioral studies and found (1) that people were particularly likely to share content about which they believed others in their social circles would share their viewpoints and (2) that this relationship is causal. In concert, our findings support the idea that people are driven to share information to create and reinforce shared understanding, which is critical to social connection.
△ Less
Submitted 26 April, 2023;
originally announced April 2023.
-
Lonely individuals process the world in idiosyncratic ways
Authors:
Elisa C. Baek,
Ryan Hyon,
Karina López,
Meng Du,
Mason A. Porter,
Carolyn Parkinson
Abstract:
Loneliness is detrimental to well-being and is often accompanied by self-reported feelings of not being understood by others. What contributes to such feelings in lonely people? We used functional magnetic resonance imaging (fMRI) of 66 participants to unobtrusively measure the relative alignment of people's mental processing of naturalistic stimuli and tested whether or not lonely people actually…
▽ More
Loneliness is detrimental to well-being and is often accompanied by self-reported feelings of not being understood by others. What contributes to such feelings in lonely people? We used functional magnetic resonance imaging (fMRI) of 66 participants to unobtrusively measure the relative alignment of people's mental processing of naturalistic stimuli and tested whether or not lonely people actually process the world in idiosyncratic ways. We found evidence for such idiosyncrasy: lonely individuals' neural responses were dissimilar to their peers, particularly in regions of the default-mode network in which similar responses have been associated with shared perspectives and subjective understanding. These relationships persisted when controlling for demographic similarities, objective social isolation, and participants' friendships with each other. Our findings suggest the possibility that being surrounded by people who see the world differently from oneself, even if one is friends with them, may be a risk factor for loneliness.
△ Less
Submitted 16 August, 2022; v1 submitted 2 July, 2021;
originally announced July 2021.
-
Popular individuals process the world in particularly normative ways
Authors:
Elisa C. Baek,
Ryan Hyon,
Karina López,
Emily S. Finn,
Mason A. Porter,
Carolyn Parkinson
Abstract:
People differ in how they attend to, interpret, and respond to their surroundings. Convergent processing of the world may be one factor that contributes to social connections between individuals. We used neuroimaging and network analysis to investigate whether the most central individuals in their communities (as measured by in-degree centrality, a notion of popularity) process the world in a part…
▽ More
People differ in how they attend to, interpret, and respond to their surroundings. Convergent processing of the world may be one factor that contributes to social connections between individuals. We used neuroimaging and network analysis to investigate whether the most central individuals in their communities (as measured by in-degree centrality, a notion of popularity) process the world in a particularly normative way. We found that more central individuals had exceptionally similar neural responses to their peers and especially to each other in brain regions that are associated with high-level interpretations and social cognition (e.g., in the default-mode network), whereas less-central individuals exhibited more idiosyncratic responses. Self-reported enjoyment of and interest in stimuli followed a similar pattern, but accounting for these data did not change our main results. These findings suggest that highly-central individuals process the world in exceptionally similar ways, whereas less-central individuals process the world in idiosyncratic ways.
△ Less
Submitted 23 September, 2021; v1 submitted 4 June, 2021;
originally announced June 2021.
-
Social Network Analysis for Social Neuroscientists
Authors:
Elisa C. Baek,
Mason A. Porter,
Carolyn Parkinson
Abstract:
Although social neuroscience is concerned with understanding how the brain interacts with its social environment, prevailing research in the field has primarily considered the human brain in isolation, deprived of its rich social context. Emerging work in social neuroscience that leverages tools from network analysis has begun to pursue this issue, advancing knowledge of how the human brain influe…
▽ More
Although social neuroscience is concerned with understanding how the brain interacts with its social environment, prevailing research in the field has primarily considered the human brain in isolation, deprived of its rich social context. Emerging work in social neuroscience that leverages tools from network analysis has begun to pursue this issue, advancing knowledge of how the human brain influences and is influenced by the structures of its social environment. In this paper, we provide an overview of key theory and methods in network analysis (especially for social systems) as an introduction for social neuroscientists who are interested in relating individual cognition to the structures of an individual's social environments. We also highlight some exciting new work as examples of how to productively use these tools to investigate questions of relevance to social neuroscientists. We include tutorials to help with practical implementation of the concepts that we discuss. We conclude by highlighting a broad range of exciting research opportunities for social neuroscientists who are interested in using network analysis to study social systems.
△ Less
Submitted 11 February, 2020; v1 submitted 26 September, 2019;
originally announced September 2019.
