About

Sha Xin Wei Ph.D. is a Professor at the School of Arts, Media + Engineering at Arizona State University, where he directs the Synthesis atelier for transversal art, philosophy, and technology. He also holds a professorship at the European Graduate School in Philosophy, Art and Critical Theory, is a Fellow of the ASU-Santa Fe Institute Center for Biosocial Complex Systems, and a Senior Fellow of Building21 at McGill University. Dr. Sha's research focuses on topological approaches to poiesis, play, and process, integrating speculative philosophy, experimental art, and visionary technologies in a reciprocal manner. His work spans gestural media, movement arts, realtime media installation, interaction design, and critical studies and philosophy of technology. Trained in mathematics at Harvard and Stanford, he has published extensively in experimental philosophy, media arts, science and technology studies, and computer science, including the book "Poiesis and Enchantment in Topological Matter" published by MIT. Dr. Sha's artistic practice includes innovative projects such as TGarden playspaces, Hubbub speech-sensitive urban spaces, and kinetic and light sculptures responsive to movement and gesture. His work has been exhibited in prominent venues worldwide and recognized by major cultural foundations. His technical research involves realtime, continuous mapping of features from environmental or ensemble activity into parameters that modulate expressive improvisation of gesture in dense fields of sound, structured light, and animated materials. He has contributed to fields including tangible interfaces, human-computer interaction, wearables, calligraphic media, movement, and responsive environments. With a background in mathematics and over a decade of experience in scientific computation, mathematical modeling, and visualization, Dr. Sha transitioned to media theory and network media authoring systems in the mid-1990s. He co-founded Pliant Research and worked at Interval Research on machine learning and computer vision. After earning an interdisciplinary Ph.D. at Stanford in 2001, he held academic positions at Georgia Institute of Technology and Concordia University, where he founded the Topological Media Lab and directed research-creation initiatives in wearables and active textiles. His work has been supported by numerous prestigious funding agencies and foundations. In 2020, he established Weightlesstudio, an atelier bridging transversal research-creation with private and public sectors. Dr. Sha also serves as an editor for AI & Society Journal and is on the governing board of Leonardo.

Research topics

• Computer Science
• Engineering
• Artificial Intelligence
• Human–computer interaction
• Psychology
• Optics
• Architectural engineering
• Multimedia
• Visual arts
• Computer vision
• Physics
• History
• Mathematics education

Selected publications

• Noisiness, the Stuff of Thought

  2025-08-13

  book-chapter1st authorCorresponding

  Michel Serres said that history is the propagation of effects, saying in his conversations with Bruno Latour, “we experience time as much in our inner senses as externally in nature, as much as le temps of history as le temps of weather” (Conversations 61), characterized more by turbulence than by Euclidean geometry. He elaborated: “the development of history truly resembles what chaos theory describes. Once you understand this, it’s not hard to accept the fact that time doesn’t always develop according to a line” (57). And elsewhere, in Genesis, he meditated on noise as the incessant murmur and hiss under the sea of atoms, below all discernibility. In one of the most beautiful passages of the book, he writes: Noise and nausea, noise and the nautical, noise and navy belong to the same family […] We never hear what we call background noise so well established there for all eternity […] The silence of the sea is mere appearance. Background noise may well be the ground of our being. It may be that our being is not at rest, it may be that it is not in motion, it may be that our being is disturbed. The background noise never ceases; it is limitless, continuous, unending, unchanging. It has itself no background, no contradictory […] Noise cannot be a phenomenon; every phenomenon is separated from it […] As soon as a phenomenon appears, it leaves the noise; as soon as a form looms up or pokes through, it reveals itself by veiling noise. So noise is not a matter of phenomenology, so it is a matter of being itself. It settles in subjects as well as in objects, in hearing as well as in space, in the observers as well as the observed, it moves through the means and tools of observation whether material or logical, hardware or software, constructed channels or languages; it is part of the in-itself, part of the for-itself; it cuts across the oldest and surest philosophical divisions, yes, noise is metaphysical. It is the complement to physics, in the broadest sense. One hears its subliminal huffing and soughing on the high seas. (Serres 13)

  PublisherDOI

• Metastable Venice

  Leonardo · 2025-07-21

  article1st authorCorresponding

  Abstract This meditation on how being alive is being vulnerable, and on the senses in which a person, a city, a people can be vulnerable to one another and to their milieus, draws on the medieval history of Venice and its sister cities, as well as recent work in theoretical biology. It touches on the drive toward invulnerability via mechanizing and replacing judgment.

  PublisherDOI

• Generating Transferable 3D Adversarial Point Clouds Based on Multi-Scale Features

  2024-07-15

  articleSenior author

  Recent studies have shown that deep neural networks are vulnerable to adversarial attacks. In the task of 3D point cloud classification, transfer-based adversarial attack methods have attracted much attention. However, the adversarial point clouds generated by existing methods are usually overly dependent on the network structure of the surrogate model, resulting in poor transferability. Additionally, they generally adopt the strategy of directly deforming and optimizing the point cloud data, which limits their transferability. To address the above problems, we propose a novel 3D point cloud adversarial attack method based on multi-scale features of data, named MSFA. The method employs an autoencoder designed to extract multi-scale features from the point clouds and reconstruct them effectively. Furthermore, it efficiently identifies positive and negative features within the point cloud data by calculating the Shapley value and employs a feature loss to attack the features of the point cloud. This makes the adversarial point clouds not overly dependent on the network structure of the surrogate model and more concerned with their own features, thus enhancing the transferability. Extensive experiments are conducted on the well-recognized benchmark dataset, demonstrating the effectiveness of the proposed MSFA method in generating highly transferable adversarial point clouds.

  PublisherDOI

• Defending Adversarial Patches via Joint Region Localizing and Inpainting

  Lecture notes in computer science · 2024 · 1 citations

  Senior authorCorresponding
  • Computer Science
  • Artificial Intelligence
  • Computer Science
  PublisherDOI

• Parametric modeling aided design for star sensor lens

  Chinese Optics · 2021 · 2 citations

  • Computer Science
  • Optics
  • Physics

  The predecessor of China Optics was China Optics and Applied Optics Digest, which was founded in 1985. At that time, it was the only retrieval journal in the field of optics in China. At the end of 2008, China Optics and Applied Optics Digest was renamed

  PublisherDOI

• Processual and Experiential Design in Wearable Music Workshopping

  2020 · 3 citations

  Senior authorCorresponding
  • Computer Science
  • Computer Science
  • Human–computer interaction

  This paper describes a series of workshops in which dancers participated in the design of collectively-playable, computational, wearable musical instruments, exploring ensemble experience in improvised events. We describe the progressive phases of layering and computationally creative learning that occurred during our workshops with epistemically and culturally diverse groups of participants. Our approach is motivated by processual and experiential design, as well as a relational approach using collective sonic mappings. The media instruments we constructed during our workshops are based on trial-and-error development of signal processing symmetrizing action and perception, designed without pre-schematizing, abstract models of user intention or semantics. Hence we outline a pre-reflective, pre-individual, sub-semantic approach to relational media synthesis.

  PublisherDOI

• Instruments of Articulation

  2019-09-25 · 5 citations

  articleSenior author

  Building on the first author's hybrid/augmented violin practice and the second author's work with responsive media environments, we build and reflect on collectively-played room-scale instruments that afford the precision and nuance of an individually-played real-time gestural media system. We consider gestural instruments designed for the interplay of action and perception at the sensorimotor level bypassing tokenization of features of activity and sensors. Our gesturally-modulated media instruments are based not on models or a priori schemata but driven by continuous adaptation to contingent activity and state of the event, as well as compositional intent. We think of such performable, expressive systems as instruments of articulation rather than of representation. Our work is motivated by a progression from phenomenological interpretations of individually-played instruments through non-anthropocentric notions of lived experience, to ecosystemic approaches to ensembles of real-time instruments, people and processes concurrently co-articulating an event.

  PublisherDOI

• Composing Ecosystemically in Responsive Environments with Gestural Media, Objects and Textures

  2019-03-15

  articleSenior author

  In this workshop, participants will try their hand at a variety of tangible, embodied, and embedded sensing and feedback technologies including vibrotactile instruments, expressive mechatronics, gesturally modulated fields of light, sound, mist; realtime steerable immersive atmospheres. Working through hands-on experience by theme, participants will be introduced to compositional and experimental methodologies. In the second half of the workshop, participants will compose together some simple "ecosystems" using the Synthesis Center's hardware-software media choreography architecture (sc), in the iStage experimental theater-scale blackbox space.

  PublisherDOI

• Rich State Transitions in a Media Choreography Framework Using an Idealized Model of Cloud Dynamics

  2017-10-23 · 3 citations

  articleSenior author

  This paper presents "An Experiential Model of the Atmosphere," a responsive media environment including multi-surface video projection and ambisonic audio feedback in which participants can fluidly interact with an idealized model of atmospheric dynamics, using the presence and motion of their bodies to force convection, condense and evaporate moisture, and manipulate atmospheric pressure gradients. We use this model to explore steering of computational models of complex systems that can span gamuts of single to multi-body interaction, local to global spatial scales, short- to long-term time scales, and abstract representations to immersive, felt environments.

  PublisherDOI

• Towards active annotation for detection of numerous and scattered objects

  2015-06-01

  article

  Object detection is an active study area in the field of computer vision and image understanding. In this paper, we propose an active annotation algorithm by addressing the detection of numerous and scattered objects in a view, e.g., hundreds of cells in microscopy images. In particular, object detection is implemented by classifying pixels into specific classes with graph-based semi-supervised learning and grouping neighboring pixels with the same label. Sample or seed selection is conducted based on a novel annotation criterion that minimizes the expected prediction error. The most informative samples are therefore annotated actively, which are subsequently propagated to the unlabeled samples via a pairwise affinity graph. Experimental results conducted on two real world datasets validate that our proposed scheme quickly reaches high quality results and reduces human efforts significantly.

  PublisherDOI

Frequent coauthors

• Hua Yang

  Shanghai Jiao Tong University

  3 shared

• Seth Dominicus Thorn

  3 shared

• Bonnie Nardi

  3 shared

• Hang Su

  Université Paris-Saclay

  3 shared

• Yawen Fan

  South China University of Technology

  3 shared

• Shibao Zheng

  3 shared

• Doug Van Nort

  York University

  2 shared

• Brandon Mechtley

  2 shared

Education

• Ph.D.

  Harvard University

• Ph.D.

  Stanford University

Awards & honors

• Canada Research Chair in media arts and sciences (2005-2013)
• Daniel Langlois Foundation for Art, Science and Technology
• LEF Foundation
• Canada Fund for Innovation
• Creative Work Fund in New York