Difference between revisions of "SOCR News APS 2025"
(→SOCR Presentations) |
(→Numerical and Analytical Complex-time Transformations of Longitudinal Processes and Spacekime Analytics) |
||
| (One intermediate revision by the same user not shown) | |||
| Line 29: | Line 29: | ||
* '''Presentation Method''': Invited in-person lecture | * '''Presentation Method''': Invited in-person lecture | ||
* '''Session''': [https://summit.aps.org/events/MAR-C37 MAR-C37 (Machine Learning for Complex Systems)] | * '''Session''': [https://summit.aps.org/events/MAR-C37 MAR-C37 (Machine Learning for Complex Systems)] | ||
| + | * '''Sponsor''': [https://engage.aps.org/gds/home GDS] | ||
* '''Session Chair''': [https://www.socr.umich.edu/people/dinov/ Ivo Dinov (Michigan)] | * '''Session Chair''': [https://www.socr.umich.edu/people/dinov/ Ivo Dinov (Michigan)] | ||
* '''Date/Time/Place''': [https://summit.aps.org/events/MAR-C37 3:00 pm - 6:00 pm Monday, March 17, 2025, Session: MAR-C37, Anaheim Convention Center, 258B (Level 2))] | * '''Date/Time/Place''': [https://summit.aps.org/events/MAR-C37 3:00 pm - 6:00 pm Monday, March 17, 2025, Session: MAR-C37, Anaheim Convention Center, 258B (Level 2))] | ||
* [https://summit.aps.org/events/MAR-C37/3 '''Abstract''']: | * [https://summit.aps.org/events/MAR-C37/3 '''Abstract''']: | ||
| − | : Complex-time (kime) representation of repeated measurement longitudinal processes paves the way for advanced spacekime statistical inference and artificial intelligence (AI) applications. Extending time into the complex plane offers a unified framework connecting fundamental quantum mechanics principles, statistical dynamics, and machine learning. Kime representation enhances both model-based statistical inference techniques – utilizing classical probability distributions – and model-free AI prediction and classification algorithms – relying on data and generalized functions. Many open mathematical-physics problems emerge from this formulation, including definition and interpretation of a consistent spacekime-metric tensor and classification of alternative time-series to kime-surfaces transformations. Simulations and observed neuroimaging data demonstrate the utility of complex-time representation and the induced spacekime analytics. These methods enable forward prediction by extrapolating processes beyond their observed timespan and facilitate group comparisons based on corresponding kime surfaces. Additionally, they allow for statistical quantification of differences between experimental groups and conditions, support topological kime surface analysis, and enhance AI prediction for repeated measurement longitudinal data | + | : Complex-time (kime) representation of repeated measurement longitudinal processes paves the way for advanced spacekime statistical inference and artificial intelligence (AI) applications. Extending time into the complex plane offers a unified framework connecting fundamental quantum mechanics principles, statistical dynamics, and machine learning. Kime representation enhances both model-based statistical inference techniques – utilizing classical probability distributions – and model-free AI prediction and classification algorithms – relying on data and generalized functions. Many open mathematical-physics problems emerge from this formulation, including definition and interpretation of a consistent spacekime-metric tensor and classification of alternative time-series to kime-surfaces transformations. Simulations and observed neuroimaging data demonstrate the utility of complex-time representation and the induced spacekime analytics. These methods enable forward prediction by extrapolating processes beyond their observed timespan and facilitate group comparisons based on corresponding kime surfaces. Additionally, they allow for statistical quantification of differences between experimental groups and conditions, support topological kime surface analysis, and enhance AI prediction for repeated measurement longitudinal data. |
==Resources== | ==Resources== | ||
Latest revision as of 19:53, 20 December 2024
Contents
SOCR News & Events: 2025 APS Global Physics Summit
2025 APS Global Physics Summit focused on AI
The annual American Physical Society's March Meetings are scientific research conferences convening over 13,000 physicists and students from around the globe to connect and collaborate across academia, industry, and major labs. The 2025 APS Global Physics Summit is aligned with the United Nations-proclaimed 2025 International Year of Quantum Science and Technology. There are 3 venues for the APS Global Physics Summit campus — Anaheim Convention Center, Anaheim Marriott, and Hilton Anaheim. The APS 2025 Summit Program is here.
The SOCR group will present two research papers as oral invited presentations.
SOCR Presentations
Probabilistic Symmetry, Variable Exchangeability, and Deep Network Learning Invariance and Equivariance
- Presenter: Yueyang Shen (co-authors Yupeng Zhang, and Ivo Dinov)
- Title: Probabilistic Symmetry, Variable Exchangeability, and Deep Network Learning Invariance and Equivariance
- Session Chair: Carlos Benavides-Riveros, University of Trento
- Presentation Method: In-person Oral Presentation
- Session: (MAR-N37) AI/ML-Driven Quantum Simulations, Quantum Monte Carlo and Variational II
- Date/Time/Place: 258B (Level 2) Wednesday March 19, 2025, 3:00p - 6:00p, Anaheim Convention Center
- Sponsor: GDS
- Abstract:
- This talk will describe a mathematical-statistics framework for representing, modeling, and utilizing invariance and equivariance properties of deep neural networks. By drawing direct parallels between topological characterizations of invariance and equivariance principles, probabilistic symmetry, and statistical inference, we explore the foundational properties underpinning reliability in deep learning models. We examine the group-theoretic invariance in a number of deep neural networks including, multilayer perceptrons, convolutional networks, transformers, variational autoencoders, and steerable neural networks. Some biomedical and imaging applications are discussed at the end. Understanding the theoretical foundation underpinning deep neural network invariance is critical for reliable estimation of prior-predictive distributions, accurate calculations of posterior inference, and consistent AI prediction, classification, and forecasting.
Numerical and Analytical Complex-time Transformations of Longitudinal Processes and Spacekime Analytics
- Presenter: Ivo Dinov (Co-author: Yueyang Shen)
- Title: 2630589 (Numerical and Analytical Complex-time Transformations of Longitudinal Processes and Spacekime Analytics)
- Presentation Method: Invited in-person lecture
- Session: MAR-C37 (Machine Learning for Complex Systems)
- Sponsor: GDS
- Session Chair: Ivo Dinov (Michigan)
- Date/Time/Place: 3:00 pm - 6:00 pm Monday, March 17, 2025, Session: MAR-C37, Anaheim Convention Center, 258B (Level 2))
- Abstract:
- Complex-time (kime) representation of repeated measurement longitudinal processes paves the way for advanced spacekime statistical inference and artificial intelligence (AI) applications. Extending time into the complex plane offers a unified framework connecting fundamental quantum mechanics principles, statistical dynamics, and machine learning. Kime representation enhances both model-based statistical inference techniques – utilizing classical probability distributions – and model-free AI prediction and classification algorithms – relying on data and generalized functions. Many open mathematical-physics problems emerge from this formulation, including definition and interpretation of a consistent spacekime-metric tensor and classification of alternative time-series to kime-surfaces transformations. Simulations and observed neuroimaging data demonstrate the utility of complex-time representation and the induced spacekime analytics. These methods enable forward prediction by extrapolating processes beyond their observed timespan and facilitate group comparisons based on corresponding kime surfaces. Additionally, they allow for statistical quantification of differences between experimental groups and conditions, support topological kime surface analysis, and enhance AI prediction for repeated measurement longitudinal data.
Resources
- ... coming up later ...
Translate this page:
