Pre-September 5th Applicants
|Name||Affiliation||Perspective and Expertise||White Paper||Biography|
|Massimiliano Espositoemail@example.com||University of Luxembourg||Nonequilibrium statistical physics, Fluctuation relations, Quantum thermodynamics, Biochemical reaction networks, Quantum and chemical computing||Massimiliano position paper||Massimiliano bio|
|Robert Fryfirstname.lastname@example.org||Johns Hopkins University||I have been applying a hybrid combination of information theory and statistical mechanics in my research in an area now called thermodynamic computing (TC). My work began almost 30 years ago when I develop and published a model for cortical neurons. This research taught me much about TC with the cortical neuron model being both the simplest TC possible and providing a powerful pathfinder.||Robert position paper||Robert bio|
|Massimiliano Di Ventraemail@example.com||UC San Diego||I would discuss about a physics-inspired computing approach, memcomputing, and its application to a variety of combinatorial optimization problems.||Massimiliano position paper||Massimiliano bio|
|Alex Nugentfirstname.lastname@example.org||Knowm Inc||In ~2001 I conceived of dielectrophoretic assembling nanoparticles acting as synaptic elements in learning processors. Key to the work was understanding how such a processor could ‘build and repair itself’ as a living brain does. My proposed solution was to build physical circuits that implemented a specific form of unsupervised anti-Hebbian and Hebbian plasticity which I demonstrated could spontaneously heal a neural network that was damaged. While working with Todd Hylton on the DARPA Physical Intelligence and (ill-fated) Thermodynamic Computing program, I realized that the same circuit I was using to achieve Anti-Hebbian and Hebbian (AHaH) synaptic plasticity in memristive circuits could also be seen as an instance of a theoretical element that Todd and I were calling a ‘Thermodynamic-Bit’. I have since found a number of ways to relate these AHaH circuits to solutions in machine learning and logic. Between ~2010 and 2014 my world view was dramatically changed as I began to wrap my mind around how the natural world self-organizes and found the work of others who were seeing and saying the same things. I have spent most of my adult life trying to acquire the pieces, technological and conceptual, required to build a self-organizing ‘thermodynamic’ or ‘physical’ processor. The idea of Thermodynamic Computing haunts me and I suspect that I will either help to build a TC or die trying.||Alex position paper||Alex bio|
|Yan Yufikemail@example.com||Virtual Structures Research, Inc.||Yan position paper||Yan bio|
|Joshua Yangfirstname.lastname@example.org||University of Massachusetts Amherst||I will show how to design and build artificial synapses and neurons functioning based on thermodynamics principles. Then based on those devices, the first integrated fully memristive neuron network was achieved and used to demonstrate unsupervised learning via self-orgazniation.||Joshua position paper||Joshua bio|
|Jeff Krichmaremail@example.com||University of California, Irvine||I bring a biological perspective to thermodynamic computing. My research interests include developing efficient coding algorithms, neuromorphic engineering and neurorobotics.||Jeff position paper||Jeff bio|
|Christof Teuscherfirstname.lastname@example.org||Portland State University||My research focuses on new computing paradigms and architectures. I’m particularly interested in computation at the thermodynamic limit by harnessing the inherent dynamics a physical system offers. I’d be very excited to contribute to this workshop. I think it’s fair to say that I’m a outside-of-the-box thinker who likes radically new ideas that have the potential to get the computing sciences to the next level.||Christof position paper||Christof bio|
|Joseph Friedmanemail@example.com||The University of Texas at Dallas||Joseph S. Friedman has experience in the design of circuits that process information through unconventional mechanisms and with unconventional devices. His research objective is to unlock the potential of nanotechnology by integrating nanodevices into efficient computing/information processing systems, particularly with spintronics and low-dimensional carbon.||Joseph position paper (same as Yiorgos)||Joseph bio|
|Yiorgos Makrisfirstname.lastname@example.org||The University of Texas at Dallas||Yiorgos Makris has extensive expertise in CMOS reconfigurable fabrics and on-die machine learning solutions. He has demonstrated various analog neural network implementations using floating-gate transistors, along with their use in general learning tasks and in designing self-testable and self-tunable analog/RF integrated circuits.||Yiorgos position paper (same as Joseph)||Yiorgos bio|
|Adam Stiegemail@example.com||University of California, Los Angeles||My research seeks to bridge the gap between our fundamental understanding of nanomaterials and how these systems tend toward complexity at mesoscopic scales through the rational design of functional nanosystems and architectures. Over the last seven years, my colleagues and I have been developing unconventional computing architectures that well onto the concepts of Thermodynamic Computing. Participation in the workshop would be a tremendous opportunity to do my part in pushing this burgeoning field forward.||Adam position paper||Adam bio|
|Alex Alemifirstname.lastname@example.org||Google Research||I am a former physicist turned Machine Learning researcher. I am currently a Research Scientist at Google Research. I have been studying the utility of information bottleneck style objectives on large scale problems by means of tractable variational approximations. Lately, I have been exploring a formal analogy between a wide class of machine learning objectives and thermodynamics.||Alex position paper||Alex bio|
|Peter Sadowskiemail@example.com||University of Hawai’i at Manoa||I am conducting research in deep learning with artificial neural networks. I have multiple publications proposing and analyzing new randomized learning algorithms for such models, and I am very interested in discussing this work with the Thermodynamic Computing community, finding new collaborators, and getting ideas.||Peter position paper||Peter bio|
|Suhas Kumarfirstname.lastname@example.org||Hewlett Packard Labs||I bring the expertise that bridges device physics and physics-driven computing architectures. My research over the last 4 years has heavily focused on how to exploit thermodynamics-driven device behavior in computing networks to solve NP-hard problems. I have written the attached white paper together with my colleague, which details a system-level platform (namely the Hopfield Network) that can aid in studying any computing process that mimics thermodynamic energy minimization. The following two recent papers of mine illustrate the bridge between thermodynamics of devices and using them to construct computing systems.
1. S. Kumar, J. P. Strachan, R. S. Williams, “Chaotic Dynamics in Nanoscale NbO2 Mott Memristors for Analogue Computing”, Nature, 548, 318 (2017)
2. S. Kumar, R. S. Williams, “Nonlinear Dynamics and Imaging of Current Density and Electric Field Bifurcations Caused by Electronic Instabilities”, Nature Communications, 9, 2030 (2018)
|Suhas position paper||Suhas bio|
Post-September 15th Applicants
|Name||Affiliation||Perspective and Expertise||White Paper||Biography|
|Christopher Kelloemail@example.com||UC Merced||Collaborated on Adam Stieg’s paper and would like to attend||Adam position paper||Chris bio|
|Michael Frankfirstname.lastname@example.org||Sandia National Laboratories||In my research, I have specialized in the area of the fundamental physical limits of computation, with particular attention to the thermodynamics of computation, for more than twenty years. In 1995 I designed one of the first protocols for universal computing using DNA chemistry, and found that the method was required to be logically reversible for fundamental thermochemical reasons. This led to a fascination with the field of logically and thermodynamically reversible computing which continues to this day. Much of my work has involved the design of electronic circuits which utilize nonequilibrium dynamical phenomena but nevertheless approach the ideal of thermodynamically reversible operation.||Michael position paper||Michael bio|
|Elan Stopnitzkyemail@example.com||University of Hawaii at Manoa||I am Dr. Susanna Still’s PhD student. I specialize in non-equilibrium thermodynamics and information theory.||Elan position paper||Elan bio|
|Norman Packardfirstname.lastname@example.org||ProtoLife Inc||I study far from equilibrium states that arise from interacting shapes at the molecular level. One example is flow of ions through pores in membranes. Theses states may play a role in computation, and are strongly affected by thermodynamic parameters that lead to the formation of the states, e.g. temperature, chemical potential.||Norman position paper||Norman bio|
|Robert Shawemail@example.com||ProtoLife Inc||Long-time interest in dynamical systems, and a more recent interest in biological topics, in particular the dynamics of molecular shapes diffusing through narrow membrane pores. These processes are an example of “thermodynamic computation”, in that they perform sorting, rectification, and simple logical functions, and are driven only by simple thermodynamic gradients.||Robert position paper||Robert bio|
|Eva Delifirstname.lastname@example.org||University of Minnesota; Institute for Consciousness Research||A hypothesis of consciousness, based on thermodynamic principles was introduced in my 2015 book, ‘The Science of Consciousness.’ Based on that work, we published the second manuscript in the literature on the thermodynamic consideration of neural systems, which we expanded for the whole brain.*
Self-regulation is an essential quality of neural systems that perform computations with thermodynamic efficiency in orders of magnitude better than current supercomputers. The subtle regulation is based on electric flows, minute potential differences and the maintenance of the high entropy state. Subjective perception of environmental information unbalances the stability, which triggers a homeostatic motivation. As material systems observe the principle of least action when moving in space, intelligent systems might realize a dynamic balance between the past and the future. The same concepts can apply to deep learning. Building computers and artificial intelligence based on neural systems would introduce a subtle regulation of computation with significantly lower the thermodynamic cost.
*Deli, E., Peters, J., and Tozzi, A. (2018) The Thermodynamic Analysis of Neural Computation. J Neurosci Clin Res 3:1.
|Eva position paper||Eva bio|
|Daewon Seoemail@example.com||UIUC, ECE||I will discuss integrated advantage of nanofunctions over classical logic gate functions in terms of Landauer’s limit using Shannon’s entropy, as an information theorist.||Daewon position paper||Daewon bio|
|Andrew Pinedafirstname.lastname@example.org||US Air Force Research Laboratory||We in the Space Electronics Technology program at AFRL are extremely interested in exploring technologies that can enable greater computing capabilities in the energy constrained environment of space. We are also interested in machine learning techniques and algorithms that can take us beyond current state of the art to enable a much greater degree of autonomy in future space missions.||Andrew position paper||Andrew bio|
|Sadasivan Shankaremail@example.com||Harvard University||Led the effort on Materials Design at Intel was part of SIA Roadmap working groups on Emerging Research on Materials, Emerging Research on Devices, Emerging Research n Architectures.
I had worked with Semiconductor Research Corporation scientists to extend their single switch formalism to larger computer systems. This is still on ongoing work and working on a publication including extension to 3D architectures.
I was also the technology representative within Intel to the high-level invited-only Power Forum in which we were examining the possible disruptions to evolutionary aspects of Moore’s law. Since my team was involved in actually designing new materials, thermodynamical efficiencies were a key part of my work with designers and architects. My work was one of the finalists from about 200 entries for Long Range Technology Forum.
My role in this meeting would be to work with several collaborators in developing the system-level thermodynamics including some of the work referred in my white paper. In addition, I would welcome a discussion to synthesize the 5-dimensional thinking for assessing thermodynamic efficiency of computing systems.
|Sadasivan position paper||Sadasivan bio|
|Karpur Shuklafirstname.lastname@example.org||Centre for Mathematical Modelling, Flame University||My main research interest, and the perspective from which I intend to approach problems relevant to the thermodynamic computing programme, lies in non-equilibrium aspects of quantum field theory and quantum many-body theory, and their applications to physical models of computing systems. My position paper describes the topic I’m particularly interested in discussing: namely, physical models for so-called adiabatic ballistic reversible computing (ABRC), originally developed by Michael Frank, and the extensions of these models to non-equilibrium regimes.
I’m also currently working on, and interested in discussing, extensions of non-equilibrium fluctuation theorems in quantum field theory to models that support topological quantum computation. My expertise further extends to non-equilibrium Green function approaches to problems in quantum many-body theory. Previously, I also worked on models of optical-electronic charge transport on the surfaces of exotic materials.
|Karpur position paper||Karpur bio|