Reducing Bias in AI-based Analysis of Visual Artworks

Zhang, Zhuomin; Li, Jia; Stork, David G.; Mansfield, Elizabeth; Russell, John E.; Adams, Catherine E.; Wang, James

doi:10.1109/mbits.2022.3197102

Cited by 5 publications

(3 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2, this method comprises two components: classification of painted content (cloud in this case) and evaluation of painting style. In a nutshell, we evaluate pictorial realism by assessing the similarity between paintings and photographs in terms of both the painted content and painting style, which makes our evaluation system more thorough and unbiased [8]. After obtaining a labeled dataset containing both photographic images of clouds and a collection of sky paintings, we first train a machine learning system using these photographs to classify cloud categories.…”

Section: Overview Of Our Approachmentioning

confidence: 99%

A Machine Learning Paradigm for Studying Pictorial Realism: How Accurate are Constable's Clouds?

Zhang,

Mansfield,

et al. 2024

IEEE Trans. Pattern Anal. Mach. Intell.

Self Cite

View full text Add to dashboard Cite

The British landscape painter John Constable is considered foundational for the Realist movement in 19 th -century European painting. Constable's painted skies, in particular, were seen as remarkably accurate by his contemporaries, an impression shared by many viewers today. Yet, assessing the accuracy of realist paintings like Constable's is subjective or intuitive, even for professional art historians, making it difficult to say with certainty what set Constable's skies apart from those of his contemporaries. Our goal is to contribute to a more objective understanding of Constable's realism. We propose a new machine-learning-based paradigm for studying pictorial realism in an explainable way. Our framework assesses realism by measuring the similarity between clouds painted by artists noted for their skies, like Constable, and photographs of clouds. The experimental results of cloud classification show that Constable approximates more consistently than his contemporaries the formal features of actual clouds in his paintings. The study, as a novel interdisciplinary approach that combines computer vision and machine learning, meteorology, and art history, is a springboard for broader and deeper analyses of pictorial realism.

show abstract

Section: Overview Of Our Approachmentioning

confidence: 99%

A Machine Learning Paradigm for Studying Pictorial Realism: How Accurate are Constable's Clouds?

Zhang,

Mansfield,

et al. 2024

IEEE Trans. Pattern Anal. Mach. Intell.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Advancements are being made to enhance filtering mechanisms, promote responsible AI use, and provide users with more control and transparency in guiding the output of GPT models. Continued research focuses on reducing biases [71], addressing potential harmful content generation, and enabling users to shape the behavior and values of the models. Beyond the aforementioned models, researchers and practitioners have explored variations and specialized versions of GPT.…”

Section: − Techniques For Model Compression and Efficiencymentioning

confidence: 99%

Unleashing the Potential of Conversational AI: Amplifying Chat-GPT’s Capabilities and Tackling Technical Hurdles

Hassija,

Chakrabarti,

Singh

et al. 2023

IEEE Access

View full text Add to dashboard Cite

Conversational AI has seen a growing interest among government, researchers, and industrialists. This comprehensive survey paper provides an in-depth analysis of large language models, specifically focusing on ChatGPT. This paper discusses the architecture, training process, and challenges associated with large language models, including bias, interpretability, and ethics. It explores various applications of ChatGPT and examines future research trends, such as improving model generalization, addressing data scarcity, and integrating multimodal capabilities. This survey also serves as a roadmap for researchers, practitioners, and policymakers, offering valuable insights into the current state and future potential of large language models and ChatGPT.

show abstract

“…An important issue is reducing bias in such statistical analyses of fine art. 14 Recall that bias is a systematic difference between the estimate of a value and its true value. 8 For instance, we saw in Fig.…”

Section: Bias Variance and Analysis Strategiesmentioning

confidence: 99%

Computational tracking of head pose through 500 years of fine-art portraiture

Chou¹,

Stork²

2023

View full text Add to dashboard Cite

We used deep neural network image analysis to automatically extract head pose angles-roll, yaw, pitch (or tilt)-and figure display length (quarter-length, half-length, full-length) from 11,000 digital images of portrait paintings in a wide variety of styles, from the early Renaissance through Modern eras. We tracked trends and exposed anomalies in such formal properties of these portraits, and this sheds light upon the social and aesthetic forces to which portrait artists respond. For example, we find that the so-called Primitive or Naive portraitists favor a highly restricted range of pose angles (primarily frontal) while Expressionist, Mannerist, and Ukiyo-e portraitists employ a far greater range of angles. We also analyzed these formal properties to reveal the different trends throughout the careers of several individual artists, such as Paul Cézanne, Édouard Manet, and Francisco Goya. Our methods can be expanded to incorporate additional computed visual and contextual informationsuch as genders and ages of figures-and thus form a foundation for addressing a large range of problems in the history of art.

show abstract

Reducing Bias in AI-based Analysis of Visual Artworks

Cited by 5 publications

References 33 publications

A Machine Learning Paradigm for Studying Pictorial Realism: How Accurate are Constable's Clouds?

A Machine Learning Paradigm for Studying Pictorial Realism: How Accurate are Constable's Clouds?

Unleashing the Potential of Conversational AI: Amplifying Chat-GPT’s Capabilities and Tackling Technical Hurdles

Computational tracking of head pose through 500 years of fine-art portraiture

Contact Info

Product

Resources

About