Friday, April 13, 2018 - 14:00 , Location: Skiles 006 , H. Weiss, P. Cvitanovic, L. Dieci, F. Bonetto, H. Zhou , (GT Math and Physics , Organizer: Haomin Zhou
Friday, March 30, 2018 - 15:00 , Location: Skiles 006 , Chethan Pandarinath , GT BME , Organizer: Sung Ha Kang
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]
Friday, March 9, 2018 - 15:00 , Location: Skiles 006 , Evangelos Theodorou , GT AE , Organizer: Sung Ha Kang
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.
Friday, December 1, 2017 - 14:00 , Location: Skiles 006 , Bunimovich, Fathi, Grigoriev, de la Llave and Zeng , GT Math and Physics , Organizer: Sung Ha Kang
Friday, November 10, 2017 - 15:00 , Location: Skiles 006 , Prof. Fumin Zhang , GT ECE , Organizer: Sung Ha Kang
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.
Modeling and predicting urban crime – How data assimilation helps bridge the gap between stochastic and continuous modelsFriday, October 20, 2017 - 15:00 , Location: Skiles 006 , Prof. Martin Short , GT Math , Organizer: Sung Ha Kang
Data assimilation is a powerful tool for combining mathematical models with real-world data to make better predictions and estimate the state and/or parameters of dynamical systems. In this talk I will give an overview of some work on models for predicting urban crime patterns, ranging from stochastic models to differential equations. I will then present some work on data assimilation techniques that have been developed and applied for this problem, so that these models can be joined with real data for purposes of model fitting and crime forecasting.
Thursday, August 17, 2017 - 09:00 , Location: Klaus 2447 , Various Speaker , Different units of GT , Organizer: Sung Ha Kang
The workshop will launch the thematic semesters on Dynamics (Fall 2017) and Control (Spring 2018) for GT-MAP activities. This is a two-day workshop, the first day focusing on the theme of Dynamics, and the second day focusing on the theme of Control. There will be light refreshments throughout the event. The workshop will be held in the Klaus building Room 2447. More information at http://gtmap.gatech.edu/events/gt-map-workshop-dynamics-and-control
Thursday, August 10, 2017 - 10:54 , Location: Klaus 1447 , Various Speakers , From various places , Organizer: Sung Ha Kang
GT MAP sponsored "Workshop on Dynamical Systems" to mark the retirement of Prof. Shui Nee Chow. Full day August 10- 11. After nearly 30 years at Georgia Tech, Prof. Shui Nee Chow has officially retired. This workshop will see several of his former students, post-docs, and friends, coming together to thank Shui Nee for his vision, service, and research, that so greatly impacted the School of Mathematics at Georgia Tech. The workshop will be held in the Klaus building Room 1447. More information at http://gtmap.gatech.edu/events/workshop-dynamical-system
Tuesday, May 9, 2017 - 10:00 , Location: Skiles 006 , Speaker list and schedule can be found at http://www.math.gatech.edu/hg/item/589661 , Organizers: Shui-Nee Chow, Wilfrid Gangbo, Prasad Tetali, and Haomin Zhou , Organizer: Haomin Zhou
This workshop is sponsored by College of Science, School of Mathematics, GT-MAP and NSF.
The goal of this workshop is to bring together experts in various aspects of optimal transport and related topics on graphs (e.g., PDE/Numerics, Computational and Analytic/Probabilistic aspects).
Friday, April 14, 2017 - 16:00 , Location: Skiles 006 , Alexander H. Chang , GT ECE , Organizer: Sung Ha Kang
Robotic snakes have the potential to navigate areas or environments that would be more challenging for traditionally engineered robots. To realize their potential requires deriving feedback control and path planning algorithms applicable to the diverse gait modalities possible. In turn, this requires equations of motion for snake movement that generalize across the gait types and their interaction dynamics. This talk will discuss efforts towards both obtaining general control equations for snake robots, and controlling them along planned trajectories. We model three-dimensional time- and spatially-varying locomotion gaits, utilized by snake-like robots, as planar continuous body curves. In so doing, quantities relevant to computing system dynamics are expressed conveniently and geometrically with respect to the planar body, thereby facilitating derivation of governing equations of motion. Simulations using the derived dynamics characterize the averaged, steady-behavior as a function of the gait parameters. These then inform an optimal trajectory planner tasked to generate viable paths through obstacle-strewn terrain. Discrete-time feedback control successfully guides the snake-like robot along the planned paths.