Dataset Information

Improving Deep Interactive Evolution with a Style-Based Generator for Artistic Expression and Creative Exploration.

ABSTRACT: Deep interactive evolution (DeepIE) combines the capacity of interactive evolutionary computation (IEC) to capture a user's preference with the domain-specific robustness of a trained generative adversarial network (GAN) generator, allowing the user to control the GAN output through evolutionary exploration of the latent space. However, the traditional GAN latent space presents feature entanglement, which limits the practicability of possible applications of DeepIE. In this paper, we implement DeepIE within a style-based generator from a StyleGAN model trained on the WikiArt dataset and propose StyleIE, a variation of DeepIE that takes advantage of the secondary disentangled latent space in the style-based generator. We performed two AB/BA crossover user tests that compared the performance of DeepIE against StyleIE for art generation. Self-rated evaluations of the performance were collected through a questionnaire. Findings from the tests suggest that StyleIE and DeepIE perform equally in tasks with open-ended goals with relaxed constraints, but StyleIE performs better in close-ended and more constrained tasks.

SUBMITTER: Tejeda-Ocampo C

PROVIDER: S-EPMC7823808 | biostudies-literature | 2020 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Improving Deep Interactive Evolution with a Style-Based Generator for Artistic Expression and Creative Exploration.

Tejeda-Ocampo Carlos C López-Cuevas Armando A Terashima-Marin Hugo H

Entropy (Basel, Switzerland) 20201224 1

Deep interactive evolution (DeepIE) combines the capacity of interactive evolutionary computation (IEC) to capture a user's preference with the domain-specific robustness of a trained generative adversarial network (GAN) generator, allowing the user to control the GAN output through evolutionary exploration of the latent space. However, the traditional GAN latent space presents feature entanglement, which limits the practicability of possible applications of DeepIE. In this paper, we implement D ...[more]

PMID: 33374104

Similar Datasets

Project description:There is a rich literature exploring emotional responses to engaging in artistic creative activities such as making music, writing, dancing and crafts. However, it remains unclear how such activities affect our emotions; specifically which mental processes ('strategies') are used to regulate our emotional responses. This paper therefore describes the design and validation of a novel instrument measuring types of emotional regulation strategies (ERSs) used when engaging in artistic creative activities: the Emotion Regulation Strategies for Artistic Creative Activities Scale (ERS-ACA). Using data from an initial pilot study (n = 740 adults, 80.4% female, median age 25-34) and a follow-up large internet sample (n = 47,924, 56.7% female, average age 47.3 ± 14.6 years), we followed a theory-driven iterative factor analysis process. Our analyses converged on a final 18-item scale comprising an overall 'general' factor of ERSs alongside three subscales: a 7-item factor comprising 'avoidance strategies' (such as distraction, suppression and detachment), a 6-item factor comprising 'approach strategies' (such as acceptance, reappraisal and problem solving), and a 5-item factor comprising 'self-development strategies' (such as enhanced self-identify, improved self-esteem and increased agency). All factors showed strong internal reliability (Cronbach's alpha: General Factor = 0.93, Factor 1 = 0.9, Factor 2 = 0.88, Factor 3 = 0.88). We confirmed strong convergent and divergent validity, construct validity, consistency of internal reliability and test-retest reliability of the scale in a third study (n = 165, 82.2% female, average age 46.3 ± 12.2 years). In conclusion, artistic creative activities appear to affect our emotions via a number of ERSs that can be broadly classified into three categories: avoidance, approach and self-development. The ERS-ACA scale presented and validated here should support further research into the use of ERSs when engaging in artistic creative activities and enhance our understanding about how these activities affect mental health.

Project description:During the COVID-19 pandemic some Australians turned to artistic creative activities (ACAs) as a way of managing their own mental health and well-being. This study examined the role of ACAs in regulating emotion and supporting mental health and well-being during the COVID-19 pandemic, and also attempted to identify at-risk populations. We proposed that (1) participants would use ACAs as avoidance-based emotion regulation strategies; and (2) music engagement would be used for emotion regulation. Australian participants (N = 653) recruited from the general public completed an online survey, which included scales targeting anxiety (GAD7 scale), depression (PHQ9 scale) and loneliness (two UCLA Loneliness Scales, referring to "Before" and "Since" COVID-19). Participants reported which ACAs they had undertaken and ceased during the pandemic using an established list and ranked their undertaken ACAs in terms of effectiveness at making them "feel better." For their top-ranked ACA, participants then completed the Emotion Regulation Scale for Artistic Creative Activities (ERS-ACA), and if participants had undertaken any musical ACAs, also the Musical Engagement Questionnaire (MusEQ). The results supported both hypotheses. ANOVAs indicated that participants ranked significantly higher on the "avoidance" ERS-ACA subscale than the other subscales, and that participants ranked significantly higher on the emotion regulation and musical preference MusEQ subscales than the other subscales. Additionally, while ACAs such as "Watching films or TV shows" and "Cookery or baking" were common, they ranked poorly as effective methods of emotion regulation, whereas "Listening to music" was the second-most frequently undertaken ACA and also the most effective. "Singing" and "Dancing" were among the most ceased ACAs but also ranked among the most effective for emotion regulation, suggesting that support for developing pandemic-safe approaches to these ACAs may provide well-being benefits in future crises. Additionally, correlation analyses showed that younger participants, those who took less exercise during the pandemic, and those with the highest musical engagement reported the poorest well-being. We conclude that ACAs provided an important resource for supporting mental health and well-being during the COVID-19 pandemic in Australia and could potentially support mental health and well-being in future crises.

Dataset Information

Improving Deep Interactive Evolution with a Style-Based Generator for Artistic Expression and Creative Exploration.

Publications

Improving Deep Interactive Evolution with a Style-Based Generator for Artistic Expression and Creative Exploration.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets