CAN: Creative Adversarial Networks, Generating "Art" by Learning About\n Styles and Deviating from Style Norms

2017/06/21 by Ahmed Elgammal, Bingchen Liu, Elgammal, Ahmed +5 · 4 voices · 22 citations
Computer Science · Neuroscience · #Generative Adversarial Networks and Image Synthesis #Aesthetic Perception and Analysis #Music Technology and Sound Studies

paper · pdf · doi:10.48550/arxiv.1706.07068

Abstract

We propose a new system for generating art. The system generates art by\nlooking at art and learning about style; and becomes creative by increasing the\narousal potential of the generated art by deviating from the learned styles. We\nbuild over Generative Adversarial Networks (GAN), which have shown the ability\nto learn to generate novel images simulating a given distribution. We argue\nthat such networks are limited in their ability to generate creative products\nin their original design. We propose modifications to its objective to make it\ncapable of generating creative art by maximizing deviation from established\nstyles and minimizing deviation from art distribution. We conducted experiments\nto compare the response of human subjects to the generated art with their\nresponse to art created by artists. The results show that human subjects could\nnot distinguish art generated by the proposed system from art generated by\ncontemporary artists and shown in top art fairs. Human subjects even rated the\ngenerated images higher on various scales.\n

Citations

• Quantifying Creativity in Art Networks
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

Cited by

• Simple Lines, Big Ideas: Towards Interpretable Assessment of Human Creativity from Drawings
• Active Divergence with Generative Deep Learning -- A Survey and Taxonomy
• Ownership and Creativity in Generative Models
• Disentangling Action Sequences: Discovering Correlated Samples
• Machine learning based co-creative design framework
• Seeking the Unfamiliar but Memorable: Conceptual Creativity as Meta-Learning
• Generalization and Memorization: The Bias Potential Model
• Generative AI in Depth: A Survey of Recent Advances, Model Variants, and Real-World Applications
• Cultural Alien Sampler: Open-ended art generation balancing originality and coherence
• A Multimodal Approach to Heritage Preservation in the Context of Climate Change
• VLM-Guided Adaptive Negative Prompting for Creative Generation
• Evolutionary Computation as Natural Generative AI
• Echoes of Humanity: Exploring the Perceived Humanness of AI Music
• High-Resolution Mammogram Synthesis using Progressive Generative\n Adversarial Networks
• Beethoven and Divergent Thinking
• CMTS: Conditional Multiple Trajectory Synthesizer for Generating Safety-critical Driving Scenarios
• Improving image generative models with human interactions
• From Sound to Sight: Towards AI-authored Music Videos
• Autonomy, Authenticity, Authorship and Intention in computer generated art
• WP-CLIP: Leveraging CLIP to Predict Wölfflin's Principles in Visual Art
• Discovering Visual Patterns in Art Collections with Spatially-consistent Feature Learning
• Predicting Opioid Relapse Using Social Media Data

Discussions

• CAN: Generating Art by Learning About Styles and Deviating from Style Norms [hn, 37 points, 9 comments]
• The loss function could be biased in other ways. I talk about this and other examples in my book. arxiv.org/abs/1706.07068 [bsky, 2 points, 1 comments]
• Generating “Art” by Learning About Styles and Deviating from Style Norms [hn, 2 points, 0 comments]
• Creative Adversarial Networks Generating “Art” [hn, 1 points, 0 comments]

Related

• Creative Portraiture: Exploring Creative Adversarial Networks and Conditional Creative Adversarial Networks
• GAN Computers Generate Arts? A Survey on Visual Arts, Music, and Literary Text Generation using Generative Adversarial Network
• Art Creation with Multi-Conditional StyleGANs
• Collaborative Interactive Evolution of Art in the Latent Space of Deep Generative Models
• Words as Art Materials: Generating Paintings with Sequential GANs