Seminars and Colloquia by Series

Artificial Intelligence Techniques for Design and Knowledge Discovery in Nanophotonics

Series
GT-MAP Seminar
Time
Friday, November 1, 2024 - 15:00 for 2 hours
Location
Skiles 006
Speaker
Prof. Ali Adibi – Georgia Tech

Please Note: Ali Adibi is the director of Bio and Environmental Sensing Technologies (BEST) and a professor and Joseph M. Pettit chair in the School of Electrical and Computer Engineering, Georgia Institute of Technology. His research group has pioneered several structures in the field of integrated nanophotonics for information processing, sensing, and quantum photonic applications. He is the author of more than 230 journal papers and 550 conference papers. He is the editor-in-chief of the Journal of Nanophotonics, and the nanophotonic program track chair of the Photonics West meeting. He is the recipient of several awards including Presidential Early Career Award for Scientists and Engineers, Packard Fellowship, NSF CAREER Award, and the SPIE Technology Achievement Award. He is also a fellow of OSA, SPIE, and AAAS.

A survey of the new artificial-intelligence (AI)-based approaches for analysis, design, optimization, and knowledge discovery in electromagnetic nanostructures will be presented. Recent advances in using both deep-learning (DL) techniques and machine-learning (ML) techniques and their application to practical problems will be covered. These techniques will not only enable more efficient designs of the electromagnetic nanostructures (e.g., metasurfaces), but also provide valuable insight about the physics of light-matter interactions in such structures. Details of the training process for these algorithms as well as the challenges and limitations of these techniques for different classes of nanostructures will be discussed. Knowledge discovery using these techniques includes the study of feasibility of a certain response from a given nanostructure and comparing the roles of different design parameters to facilitate the training process.

Digital Twins in the era of generative AI — Application to Geological CO2 Storage

Series
GT-MAP Seminar
Time
Friday, September 20, 2024 - 15:00 for 2 hours
Location
Skiles 006
Speaker
Felix J. Herrmann – GT CSE, ECE, and EAS –

Please Note: Felix J. Herrmann Georgia Research Alliance Eminent Scholar Chair in Energy Seismic Laboratory for Imaging and Modeling Schools of Earth & Atmospheric Sciences, Computational Science & Engineering, Electrical and Computer Engineering Georgia Institute of Technology https://slim.gatech.edu Felix J. Herrmann is a professor with appointments at the College of Sciences (EAS), Computing (CSE), and Engineering (ECE) at the Georgia Institute of Technology. He leads the Seismic Laboratory for Imaging and modeling (SLIM) and he is co-founder/director of the Center for Machine Learning for Seismic (ML4Seismic). This Center is designed to foster industrial research partnerships and drive innovations in artificial-intelligence assisted seismic imaging, interpretation, analysis, and time-lapse monitoring. In 2019, he toured the world presenting the SEG Distinguished Lecture. In 2020, he was the recipient of the SEG Reginald Fessenden Award for his contributions to seismic data acquisition with compressive sensing. Since his arrival at Georgia Tech in 2017, he expanded his research program to include machine learning for Bayesian wave-equation based inference using techniques from simulation-based inference. More recently, he started a research program on seismic monitoring of Geological Carbon Storage, which includes the development of an uncertainty-aware Digital Twin. In 2023, the manuscript entitled “Learned multiphysics inversion with differentiable programming and machine learning” was the most downloaded paper of 2023 in Society of Exploration Geophysicist’s The Leading Edge.

As a society, we are faced with important challenges to combat climate change. Geological Carbon Storage, during which gigatonnes of super-critical CO2 are stored underground, is arguably the only scalable net-negative negative CO2-emission technology that is available. Recent advances in generative AI offer unique opportunities—especially in the context of Digital Twins for subsurface CO2-storage monitoring, decision making, and control—to help scale this technology, optimize its operations, lower its costs, and reduce its risks, so assurances can be made whether storage projects proceed as expected and whether CO2 remains underground.

During this talk, it is shown how techniques from Simulation-Based Inference and Ensemble Bayesian Filtering can be extended to establish probabilistic baselines and assimilate multimodal data for problems challenged by large degrees of freedom, nonlinear multiphysics, and computationally expensive to evaluate simulations. Key concepts that will be reviewed include neural Wave-Based Inference with Amortized Uncertainty Quantification and physics-based Summary Statistics, Ensemble Bayesian Filtering with Conditional Neural Networks, and learned multiphysics inversion with Differentiable Programming.

This is joint work with Rafael Orozco.

 

Computing High-Dimensional Optimal Transport by Flow Neural Networks

Series
GT-MAP Seminar
Time
Friday, April 26, 2024 - 15:00 for 2 hours
Location
Skiles 005 and https://gatech.zoom.us/j/98355006347
Speaker
Yao Xie – H. Milton Stewart School of Industrial and Systems Engineering at Georgia Tech –

Flow-based models are widely used in generative tasks, including normalizing flow, where a neural network transports from a data distribution P to a normal distribution. This work develops a flow-based model that transports from P to an arbitrary Q (which can be pre-determined or induced as the solution to an optimization problem), where both distributions are only accessible via finite samples. We propose to learn the dynamic optimal transport between P and Q by training a flow neural network. The model is trained to optimally find an invertible transport map between P and Q by minimizing the transport cost. The trained optimal transport flow subsequently allows for performing many downstream tasks, including infinitesimal density ratio estimation (DRE) and distribution interpolation in the latent space for generative models. The effectiveness of the proposed model on high-dimensional data is demonstrated by strong empirical performance on high-dimensional DRE, OT baselines, and image-to-image translation.

Measure-valued splines and matrix optimal transport

Series
GT-MAP Seminar
Time
Friday, March 8, 2019 - 15:00 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Prof. Yongxin Chen – GT AE

Two recent extensions of optimal mass transport theory will be covered. In the first part of the talk, we will discuss measure-valued spline, which generalizes the notion of cubic spline to the space of distributions. It addresses the problem to smoothly interpolate (empirical) probability measures. Potential applications include time sequence interpolation or regression of images, histograms or aggregated datas. In the second part of the talk, we will introduce matrix-valued optimal transport. It extends the optimal transport theory to handle matrix-valued densities. Several instances are quantum states, color images, diffusion tensor images and multi-variate power spectra. The new tool is expected to have applications in these domains. We will focus on theoretical side of the stories in both parts of the talk.

Efficient Prediction of User Activity using Mass Transport Equation

Series
GT-MAP Seminar
Time
Tuesday, November 27, 2018 - 15:00 for 2 hours
Location
Skiles 005
Speaker
Prof. Le Song – GT CSE

Please Note: This is a part of GT MAP seminar. See gtmap.gatech.edu for more information.

Point processes such as Hawkes processes are powerful tools to model user activities and have a plethora of applications in social sciences. Predicting user activities based on point processes is a central problem which is typically solved via sampling. In this talk, I will describe an efficient method based on a differential-difference equation to compute the conditional probability mass function of point processes. This framework is applicable to general point processes prediction tasks, and achieves marked efficiency improvement in diverse real-world applications compared to existing methods.

Unsupervised discovery of ensemble dynamics in the brain using deep learning techniques

Series
GT-MAP Seminar
Time
Friday, March 30, 2018 - 15:00 for 2 hours
Location
Skiles 006
Speaker
Chethan Pandarinath – GT BME
Since its inception, neuroscience has largely focused on the neuron as the functional unit of the nervous system. However, recent evidence demonstrates that populations of neurons within a brain area collectively show emergent functional properties ("dynamics"), properties that are not apparent at the level of individual neurons. These emergent dynamics likely serve as the brain’s fundamental computational mechanism. This shift compels neuroscientists to characterize emergent properties – that is, interactions between neurons – to understand computation in brain networks. Yet this introduces a daunting challenge – with millions of neurons in any given brain area, characterizing interactions within an area, and further, between brain areas, rapidly becomes intractable.I will demonstrate a novel unsupervised tool, Latent Factor Analysis via Dynamical Systems ("LFADS"), that can accurately and succinctly capture the emergent dynamics of large neural populations from limited sampling. LFADS is based around deep learning architectures (variational sequential auto-encoders), and builds a model of an observed neural population's dynamics using a nonlinear dynamical system (a recurrent neural network). When applied to neuronal ensemble recordings (~200 neurons) from macaque primary motor cortex (M1), we find that modeling population dynamics yields accurate estimates of the state of M1, as well as accurate predictions of the animal's motor behavior, on millisecond timescales. I will also demonstrate how our approach allows us to infer perturbations to the dynamical system (i.e., unobserved inputs to the neural population), and further allows us to leverage population recordings across long timescales (months) to build more accurate models of M1's dynamics.This approach demonstrates the power of deep learning tools to model nonlinear dynamical systems and infer accurate estimates of the states of large biological networks. In addition, we will discuss future directions, where we aim to pry open the "black box" of the trained recurrent neural networks, in order to understand the computations being performed by the modeled neural populations.pre-print available: lfads.github.io [lfads.github.io]

The science of autonomy: A "happy" symbiosis between learning, control and physics.

Series
GT-MAP Seminar
Time
Friday, March 9, 2018 - 15:00 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Evangelos Theodorou – GT AE
In this talk I will present an information theoretic approach to stochastic optimal control and inference that has advantages over classical methodologies and theories for decision making under uncertainty. The main idea is that there are certain connections between optimality principles in control and information theoretic inequalities in statistical physics that allow us to solve hard decision making problems in robotics, autonomous systems and beyond. There are essentially two different points of view of the same "thing" and these two different points of view overlap for a fairly general class of dynamical systems that undergo stochastic effects. I will also present a holistic view of autonomy that collapses planning, perception and control into one computational engine, and ask questions such as how organization and structure relates to computation and performance. The last part of my talk includes computational frameworks for uncertainty representation and suggests ways to incorporate these representations within decision making and control.

Bio-Inspired Autonomy for Mobile Sensor Network

Series
GT-MAP Seminar
Time
Friday, November 10, 2017 - 15:00 for 2 hours
Location
Skiles 006
Speaker
Prof. Fumin Zhang – GT ECE
There is an increasing trend for robots to serve as networked mobile sensing platforms that are able to collect data and interact with humans in various types of environment in unprecedented ways. The need for undisturbed operation posts higher goals for autonomy. This talk reviews recent developments in autonomous collective foraging in a complex environment that explicitly integrates insights from biology with models and provable strategies from control theory and robotics. The methods are rigorously developed and tightly integrated with experimental effort with promising results achieved.

Pages