Rhizome - Signal Representation Theory for Artists
Rhizome - Signal Representation Theory for Artists
Saturday, December 2 · 2 - 5pm EST
Rhizome is pleased to announce a new workshop, "Signal Representation Theory for Artists," taught by Max Ardito and held in partnership with Onassis ONX, a hybrid production and exhibition space with a global community of members who create interactive and immersive XR works.
Implicit in generative AI, computer software, programming languages, hardware, and data itself, lies an unspoken ideology to which we consent as both artists and consumers: that of signal representation. Signal representation theory is a field that is concerned with the representation of real-world phenomena—language, image, sound—through ideas and structures in pure mathematics. Signal representations are depictions of the world, no different than paintings, poetry, or sheet music. Likewise, they are painted, written, and composed by engineers and researchers, and the assumptions made in their depictions of sensation have a substantial impact on our collective aesthetic experience, as well as our fundamental understanding of the function of art in society.
This workshop is geared towards artists who want to develop a rudimentary understanding of signal representation theory. The workshop will focus heavily on sound, but sometimes bleed into the domains of image and language processing. Topics covered will include high, low, and latent dimensional audio signal representations, their material histories, and their relationships to various assumptions about auditory perception. A high level walkthrough of papers such as Understanding Deep Convolutional Networks (Mallat, 2016) and DDSP: Differentiable Digital Signal Processing (Engel, 2020) will also be presented. Through these topics we will attempt to demystify a number of multifaceted concepts that are implied in the design of generative AI by analyzing the ways in which engineers aesthetically and phenomenologically represent the world through different mathematical models.
Lecture (~ 45 min):
➔ Background (5 min)
◆ Imagining Utopias — Adorno
◆ Historically Informed Performance
◆ IEEE, NeurIPS, ICASSP
➔ Representation Theory (5 min)
◆ in Math — (e.g. Bourbaki)
◆ in Music — (e.g. Cella)
◆ in Engineering — (e.g. Daubechies, Mallat)
➔ High Dimensional Representations (15 min)
◆ Fourier Analysis and Uncertainty
◆ Bases and Dictionaries
◆ Wavelet Decomposition
◆ Scattering Transforms
➔ Low Dimensional Representations (10 Min)
◆ Classical Descriptors
◆ Spectral Descriptors
◆ Conditional Representations and Parameter Spaces
➔ Latent Representations (10 min)
◆ Manifolds and Differential Geometry
◆ Dimensionality Reduction and Convolutional Neural Networks
◆ Conditional Variational Autoencoders
◆ Wavelet Scattering as CNNs
Workshop (~ 1 hr)
● Basic Time-Frequency Analysis (15 min)
● Multiresolution Analysis and Wavelet Scattering (15 min)
● Classical Feature Extraction (15-30 min)
● Questions (15 min)
Bios:
Max Ardito is a composer and researcher who studies signal processing, mathematics, media theory, and the history of technology. Currently he lives in Montréal where he’s a member of the Centre for Interdisciplinary Research in Music Media and Technology (CIRMMT), and McGill University's Sound Processing and Control Lab (SPCL). His academic work centers around various applications of time-frequency analysis, wavelet scattering, group theory, and representation theory to both music composition and computational musicology. During the summer, he teaches a Max/MSP/Jitter course in Paris at l'Institut de Recherche et Coordination Acoustique/Musique (IRCAM) on computer music composition and electronic music history. In Brooklyn—where he was born and raised—he’s one of the founders of synth manufacturer, community space, and experimental music venue CuteLab.
Rhizome is a born-digital, non-profit organization that arose from early net art communities' growing need to both advocate for themselves as artists and preserve their work. Rhizome's experimental programs and writing can be traced along the pathways of internet history. Now, as Rhizome continues to engage with new ideas and artists, we find ourselves in partnership with many others who also explore the depths of the web. Since 2003, Rhizome has been an affiliate in residence at the New Museum of Contemporary Art in New York City.
Onassis ONX is an accelerator for a global community of member artists who create immersive XR and AI works. This hybrid experimentation, production and exhibition space is located in the Olympic Tower in Midtown Manhattan and was founded in 2020 in partnership between the Onassis Foundation and NEW INC. Onassis ONX now exists as a project of Onassis Culture, working in partnership with organizations in New York, including New York University’s Tandon School, Games for Change, NEW INC, Rhizome, and MAX, as well as with global institutions such as IDFA, CPH:DOX, DiMoDA, and others. Offering development funding, state-of-the-art facilities, advocacy, and community for an international community of artists from a broad range of disciplines, Onassis ONX has become an influential hub and manifestation of the dynamism of a new global interdisciplinary artistic ecosystem.