Creative AI newsletter #6 — Art, Design and Music updates over the past 2 months
This is an occasional newsletter with updates on creative applications of artificial intelligence in art, music, design and beyond. Below is issue #6 and you can subscribe for future editions here 😀
Researchers from Rutgers University and the Atelier for Restoration & Research of Paintings used RNNs to identify what stroke features identified the artist based on a dataset of 80,000 brushstrokes from 300 paintings by Picasso, Matisse and others. Meanwhile, Artnome searched 1800+ of Munch’s paintings using object tags to discover new favourite artworks.
Recent artworks include Marco Donnarumma’s work Amygdala, a robot learning the ritual of purification known as “skin-cutting” and Paperclips, a game by Frank Lantz that lets you control an AI running a paperclip company. Hyperallergic also covered Phillipp Schmitt’s “Computed Curation” photobook.
In the project Neural Kubrick, the Interactive Architecture Lab at UCL gave AI the role of Art Director, Film Editor and Director of Photography. Yannick Assogba explored visual motifs in Wes Anderson films.
November was NaNoGenMo and Procjam month. Check out Janelle Shane’s experiments generating first lines of novels with neural networks and GLKT’s Random Character project, which generates characters made from various objects and textures. Mario Klingemann has a transhanced version.
There’s also a podcast about using machine learning to generate a new “super religion”.
There have been a few projects using GANs in fashion lately: Adobe and UCSD aim to find your personal taste and generate designs most consistent with it, while the start-up Vue.ai shows how fashion garments fit a person. Adidas too has been applying data-driven approaches to design specialised shoes for runners in different cities. Kota Yamaguchi has an overview of computer vision applications in fashion.
University of Illinois at Urbana-Champaign released Rico, the largest repository of mobile app designs to date ,while Airbnb developed an AI-powered product development tool.
Back in October, Adobe announced new AI-based image editing tools for removing objects, style transfer and colorization.
The latest music AI experiments include this lullaby from Jukedeck, black metal from Dadabots and the Stromae and Flow Machines collaboration for the album Hello World (in French). Do check out Pixel Records, a fictional label created by Boiler Room and Google Pixel 2.
Tero Parviainen has an interactive explainer on generative music, while Sophia Ciocca revealed how Spotify’s personalised recommendations work and Leon Fedden wrote up his project on Comparative Audio Analysis With Wavenet, MFCCs, UMAP, t-SNE and PCA.
Connection Science have devoted their Issue 4 to Computational Creativity including contributions from Margaret Boden on deep dreaming and collage and from Anna Jordanous on “Beyond the Fence” and musical theatre.
Glass Bead have the issue Site 1: Logic Gate, the Politics of the Artifactual Mind exploring reason and AI with pieces from Hito Steyerl, Ian Cheng and Benjamin Bratton.
There’s a comprehensive overview of AI in game design in Technology Review looking at the work of Mike Cook, Julian Togelius and Mark Riedl. Speaking of Mark Riedl, he considered the past, present and future of computational narrative intelligence here.
If you’re looking for an introduction to neural networks, Gene Kogan’s ml4acourse have released their first two book chapters here and here. There’s also a great collection of resources for neural style transfer by Kailash Ahirwar here and a playground to explore GANs by Reiichiro Nakano here.
NVIDIA’s been doing some impressive work recently with GANs including this high-resolution image synthesis from semantic label maps following their progressive growing of GANs ICLR submission that is able to generate hyper-realistic images.
Chris Olah, Alex Mordvintsev (researchers behind the original DeepDream!) and Ludwig Schubert published an article on feature visualisation with a comparison of techniques and tips to improve optimisation. Make sure to check out the extra visualisations in the appendix too.
David Ha has written up a visual guide to evolution strategies.
In case you missed it, there’s the fooling Neural Networks in the Physical World with 3D Adversarial Objects. Always fun to get a cat recognised as guacamole.
There’s a few conference open calls for papers going including Cybernetic Serendipity Reimagined on 4–6 April in Liverpool (deadline 5th Jan), Politics of Machines — Art and After on May 15–17 in Copenhagen (deadline 22ndDec) and Intelligent Music Interfaces for Listening and Creation on 11 March in Tokyo (deadline 17th Dec).
Google Arts and Culture is looking for artists using machine learning for n-dimensions, its first remote “artist-in-residence” programme at Somerset House Studios. Apply before 7th December. If that’s you, consider submitting your art to Singularity Now, the 14th Athens Digital Festival, taking place in May 2018. Open call for artworks until 31st December.
Dartmouth have opened applications for their 2018 Literary and Music Creative Turing Tests. Literary Categories include machine-generated sonnets, limericks, poetry and children’s stories. For music, it’s style or free composition and improvisation with a human performer. Apply before May 15th2018.
The Meetup 🤖🎨
I’m still planning next year’s events, so if you have speakers or themes to recommend, please let me know. The dates for the next meetups are set as 11th January and 6th February. Follow the meetup group for updates.
Oh and slides from Hooman Shayani’s talk at the event on AI and Design are now online.
Cool things to do 😃
I’m organising a NIPS workshop on Machine Learning for Creativity and Design on 8th December in Long Beach. Our schedule including accepted papers can be found here. There will also be an online gallery of accepted art soon. Come drop by if you’re at NIPS.
Karlsruhe: The Open Codes exhibition at ZKM presents art and scientific works that visualize and explain codes and the way it shapes the way we live and perceive the world. Artists include Constant Dullaart, James Bridle and Matthew Plummer-Fernandez. Until 5th August 2018.
New York: Thinking Machines at MOMA looks at art and design in the computer age between 1959–1989, bringing artworks produced using computers and computational thinking together with notable examples of computer and component design. Until 8th April 2018.
Manchester: The Lowry is hosting a humansbeingdigital exhibition that looks at the touch-point between human beings and being digital including Max Dovey’s Hipster Bar. Until 25th Feb 2018.