C4'S VERY INFORMAL SEMINAR SERIES


The VIS is intended to be an informal forum in which to present new, conceptually and / or analytically challenging ideas. Goals include lots of free wheeling discussion and new collaborations. Scientists and scholars in the UW Madison community are welcome to attend. The VIS is not open to the public. Information about our public lectures can be found
here.

Partial support for the VIS is generously provided by the John Templeton Foundation through a grant to study complexity.

Contact Jessica Flack (jcflack@wisc.edu) or Tim Taylor (ttaylor4@wisc.edu) for further information.



SCHEDULE 2014
(always being updated)


Friday, November 7, 330PM, WID third floor conference room
Marty Lichtman, Kohler Fellow and Physics Department
Topic: TBA

Friday, December 5, 330PM, WID third floor conference room OR teaching lab (TBD)
John Krakauer, Johns Hopkins Neuroscience.
Topic: TBA


PAST SEMINARS

Friday, October 10, 330PM, WID third floor conference room
Matthew Banks, Department of Anesthesiology & also Neuroscience Training Program
Topic: Cortical network activity induced by sensory stimulation: Detection and implications for coding

Background:

The ‘Neuron Doctrine’, a venerable pillar of current neuroscience dogma, holds that the fundamental processing unit of the brain is the neuron, whose output currency consists of action potentials or ‘spikes’. Motivated by this doctrine, electrophysiological recordings from single neurons have provided enormous insight into the organization of sensory and motor neocortex, and have elucidated the molecular and cellular mechanisms of memory formation, sensory coding and effects of psychoactive pharmacological agents. We know, however, that cells in the brain rarely operate independently, and thus their output will depend on the activity of the network in which they are embedded. The advent of multichannel recordings from the brain has enabled investigation of this network dependence. Data from my lab and others’ suggest cortical networks operate in a highly nonlinear regime, producing ~all-or-none burst responses during the processing of sensory stimuli whose probability is determine by both the stimulus parameters (tuning; ‘bottom-up’) and the current state of the cortical network (context; ‘top-down’). Two challenges in analyzing these multi-electrode recordings that I will discuss in this presentation are the detection of these network bursts amidst ongoing (‘noisy’) activity, and the implications of the occurrence and propagation of these events for signal coding.
day, September 5, 330PM, WID third floor conference room

Josh Pultorak, Kohler Fellow & Psychology Department

Topic: Open Science, Publication Bias, and Reproducibility: 21st Century Approaches to a Persisting Problem

Background:

Science is done by people. Despite the critical emphasis placed on objectivity in scientific research, researchers themselves often fall victim to biases arising from internal and external factors, including pressure to publish “positive” results. This, in turn, leads to the perpetuation of scientific practices that fall short of our scientific values. Luckily, there are a number of new initiatives and growing organizations, such as the Center for Open Science, that seek to reduce these biases and strengthen 21st century science through the use of 21st century technology. By fostering both incentive and technical infrastructure support for open data sharing, issues concerning reproducibility and publication bias may be substantially addressed in the coming decades. Challenges and hopes of these new possibilities will be discussed. 

Friday, May 9, 330PM, WID third floor conference room (rescheduled for May 9) CANCELLED

Kevin Jamieson, Optimization & Engineering
Topic:
Provably optimal adaptive data collection in the face of uncertainty: so easy a kid could do it!

Background: As researchers, we are constantly collecting data to answer our questions and each data point comes with a cost of time or money. The aim of my work is to develop practical algorithms with theoretical guarantees that find answers faster by using an adaptive sampling strategy to minimize the amount of data collected. But instead of allowing these algorithms to languish in pseudo-code in papers read only by other experts in the field, I am working with others to build a web interface to allow anyone - researchers, engineers, teachers, kids - to ask questions that can only be answered by collecting data, lots of data, from many people with little time. The system I'll discuss automatically decides which questions will be asked to which people based on all the data seen so far,  and will be no harder to setup than starting a Doodle poll or a survey on Survey Monkey. I will review the basic theory of adaptive sampling, our progress towards the web interface goal, and also discuss problems that we have no idea how to solve. 

The discussion will derive from a series of problems I have worked on, some solved but most not:  a) what is the best way to assign judges to science projects so that more eyes get on the projects that have a chance of winning? b)  you had the best beer of your life last night but forgot its name; by asking you questions like "was it more similar to Spotted Cow or Hopaliscious" how can I find your beer (or recommend you a beer) using as few questions as possible? c) psychologists are interested in how people perceive the structured world by asking questions like "Is a frog more similar to a fish or kangaroo?" Structure implies that queries are not equally informative, which questions should be asked to maximize information gain? Surprisingly, the solution of this problem might even help physicists improve quantum computing systems. d) Sociologists are interested in discovering what cues people pay attention to when they decide that one street looks "safer" than an alternative. How should pairs of street images be chosen to discover the right cues as quickly as possible? Can the solution of this problem help biologists design experiments to understand which genes influence fitness?  


Friday, April 11, 330PM, WID third floor conference room
Tony Ives, Zoology
Topic: How do we use theory in biology?

Background: Theory is often associated with laws of nature: there are simple theoretical equations for Newton’s law of gravity and Einstein’s laws of special relativity. But theory, or at least mathematics, is also used for the more-mundane, day-to-day business of Science. How is this theory being used? What are the epistemological underpinnings of this day-to-day theory? I want to discuss the use of theory in Biology in a mundane way, asking for examples of useful biological models. Are analyzing these models any more or less abstract than doing an experiment?

Friday, March 7, 330PM, WID third floor conference room
Jerome Gaillard, Physics & C4
Topic: What are string theorists up to? Current issues in high energy physics

Background: The recent discovery of the Higgs boson at the LHC has been one of the biggest experimental result in high energy physics. It plays a very important role in our understanding of the Universe at the most fundamental scales. But it is not by far the end of the story. Many questions still do not have an answer. Why is gravity so different from the other fundamental forces? What is dark matter? What is really happening inside a proton? Are there more fundamental particles that we have not discovered yet? I will talk about those questions and others, discussing current theories and experiments. In particular, I will explain what string theory, by providing a dual description to the standard field-theoretical one, has to say on those issues.


Friday, February 14, 330PM, WID third floor conference room
Bill Sandholm, Economics
Topic: Evolutionary game theory dynamics and applications

Background: Population games provide a general model of strategic interactions among large numbers of agents.  To model the dynamics of behavior in these games, we introduce revision protocols, which are explicit stochastic descriptions of how individual agents make decisions.  Together, a population game and a revision protocol define a Markov process that captures the evolution of aggregate behavior.  When the number of agents is large, evolution over moderate time spans is well approximated by solutions to ordinary differential equations.  Over longer time spans, tools from large deviations theory can be used to analyze transitions between stable regimes.  Allowing for heterogeneity in agents' preferences creates a powerful tool for studying behavior dynamics arising in applications, as we illustrate using a model of congestion pricing in highway networks, and through models of residential segregation a la Schelling.


Friday, December 6, 330PM, WID third floor conference room
Jean-Luc Thiffeault, Department of Mathematics, UW Madison
Topic: How do we characterize complex entanglements?

Background: There are many physical systems that lead to entangled threads: polymers, human hair, fibrous materials, gels...  From a mathematical standpoint, how do we characterize the "degree of entanglement" in a manner that is useful for computing macroscopic physical quantities?  For instance, can we predict the strength of a material from some characteristic of the network of fibers?

Friday, November 8, 330PM, WID third floor conference room
Randy Ashton, Bionates Group, WID
Topic: Interdisciplinary Challenges in Tissue Engineering Morphogensis

Background: Can we recapitulate developmental processes in vitro to generate anatomically biomimetic human tissues from human pluripotent stem cells (hPSCs)? Recent advancements in organoid culture suggest that hPSCs contain innate self-organizing capabilities that can be exploited with minimal exogenous interventions to recapitulate tissue morphogenesis ex vivo. This VIS will present these advancements and seek to develop an interdisciplinary framework for modeling and predicting morphogenic events in differentiating hSPC cultures. Such models could be experimentally tested using engineered culture systems, and ultimately, create systems that instruct ex vivo morphogenesis of anatomically biomimetic human tissues. 


Friday, October 11, 330PM, WID third floor conference room
John Denu, Epigenetics Group, WID
Topic: Technical and Theoretical Hurdles Toward Understanding the Epigenome

Background: Our laboratory studies the complex interaction of metabolism and epigenetics. Specifically, we seek to elucidate important paradigms in the dynamic regulation of chromatin structure and function, which controls gene expression, within the context of cellular nutrient status and metabolism. To this end, we have traditionally used a combination of classic biochemistry and molecular biology to explore important mechanisms of chromatin regulation and metabolic control. More recently, we (and others) have embraced the use of state-of-the-art techniques in genomics, proteomics, and metabolomics. These large, information rich datasets provide a unique opportunity to view cellular processes in a global, synergistic manner. However, providing a coherent structure to these largely disparate datasets is very difficult. We believe the synthesis of these datasets to reveal emergent patterns of basic physiological and biochemical truths is an incredibly important endeavor with the potential to impact all areas of biology and medicine. However, how does one begin to tackle this problem? How does one integrate and interpret these data sets? Ultimately, we seek unifying concepts/models that are testable experimentally. In this presentation, we will provide a general introduction to chromatin-regulated gene expression and the basic links to metabolism. We will describe the kinds of data that can be extracted from cells/animals. In the end, we seek insight into both the practical and theoretic aspects required to interrogate the connections between metabolism and chromatin-controlled gene expression.




Friday, September 13, 330PM, WID third floor conference room
Steven Durlauf, UW Economics Department
Topic: The Statistical Mechanics of Choice

Background: Steven will discuss how one can start with standard economic assumptions about individual decision making, enrich these assumptions by including social influences, and derive a description of choices across a population of actors that is mathematically equivalent to a standard model in statistical mechanics.  Steven will then describe some economic environments that build upon this framework which, if successfully analyzed, would be important in the study of a range of socioeconomic outcomes. This discussion will be based on joint work with William Brock.

Friday, August 9, 330PM, WID third floor conference room
Sebastien Roch, UW Mathematics Department
Topic: The Tree of Life: More Data, More Problems

Background: In a classical molecular phylogenetic analysis, a single evolutionary tree - hopefully depicting parts of the true Tree of Life - is reconstructed from copies of a single gene in several species. One might expect that the increasing availability of large-scale datasets comprising multiple genes would necessarily contribute to more accurate inferences. However, in addition to obvious scalability issues, genomic data presents new challenges. For instance, many evolutionary mechanisms cause different genes to support different trees. In this context, the Tree of Life may be better thought of as a forest, or cloud, of gene histories. As an example I will describe one such mechanism, deep coalescence, and discuss the difficulties it presents for the reconstruction of the history of life.

Friday, July 19, 2013, 3rd floor WID conference room
Matthew Berland, UW Madison Department of Curriculum and Instruction
Topic: What is tinkering & why is it so effective?

Background: A variety of work in the learning sciences suggests that people come to understand complex content effectively when they make some artifact that requires that knowledge to work. My own work suggests that a lot of the benefit from that creative process comes not necessarily from planning and thinking through problems but by ‘tinkering’ during the building of the artifact. For instance, in computer programming, tinkering might characterize the quick (<1m) cycle of ‘change a few letters, recompile, retest’. But what is going on there? Does the feedback from changing minor features, however insignificant or ancillary, reify existing broader knowledge? What data or experiments might help us understand what about tinkering is so effective in teaching complex content? What features of tinkering resonate across or outside of computer programming or electrical engineering?

Friday, June 21, 330PM, WID third floor conference room
Terry Millar, UW Mathematics Department
Topic: The journey (to) (of) (from)(and back to) the Infinitesimal - An Inconvenient Truth

This will be a quick look at the birth (Democritus 450 B.C.E.), use (Archimedes, Leibniz, Newton, physicists, engineers....), persecution (Eudoxus, Berkeley, Russell,...), death (Bolzano, Cauchy,... Weirstrauss 19th Century), and resurrection (Robinson 1960) of infinitesimals as a mathematical construct.

Friday, May 31, 330PM, WID third floor conference room
Elliot Sober, Philosophy Department
Topic: When should you postulate an intervening variable?


Cognitive ethologists debate what mental abilities should be attributed to chimpanzees to explain their behaviors.  For example, is there evidence that chimps have beliefs about the mental states of other chimps? I'll discuss an argument that has been made in this debate that concludes that the explanation of behavior never requires such attributions.

Friday, April 12, 2013, 3:30PM
David Krakauer, WID Director & Co-Director, Center for Complexity and Collective Computation
Topic: A demonstration of the extreme limitations of memory (this will be amusing).


Friday, March 15, 2013, 4:30PM
Richie Davidson, Center for Investigating Healthy Minds and Laboratory of Affective Neuroscience, UW-Madison
Topic: Change your brain by transforming your mind


This informal talk will consider the possibility of producing functional and structural changes in the brain through mental practices such as meditation.  The import of such alterations in brain function and structure for transforming behavior and experience will be discussed. 


Friday, February 15, 2013, 3:30PM
Chris Ellison, Center for Complexity & Collective Computation
Topic: Fundamental unsolved issues in information theory

Friday, January 4, 2013, 3:30PM
Tim Rogers, Psychology
Topic: Why do people believe crazy things?


Thursday, December 13, 2012, 10AM
Seth Lloyd, Center for Extreme Quantum Information Theory, MIT
Topic:
The Energy-Information-Computation Triarchy

Friday, November 16, 330PM

Eddie Lee
Topic: Principles of Human Behavior: A Physics Perspective


Friday, October 26, 330PM
Bryan Daniels, C4, WID, UW Madison
Topic: Cage War: Ockham, Laplace, Google and the Nature of Science

September 21, 2012, 330PM
Rob Nowak, Optimization Group, WID, UW Madison
Topic: In Vivo Las Vegas: Experimental Design as a Multi-Arm Bandit Problem


SUMMER 2012
August 24 2012, 330PM
David Baum, Botany, UW Madison
Topic: Origin of Life


July 13 2012
Sushmita Roy, WID, UW Madison
Topic: Regulatory Causal Networks
330PM WID 3rd Floor Conference Room

June 15 2012
Pupa Gilbert, Physics, UW Madison & Berkeley
Topic: Mullusk Shells and Temperature
330PM WID 3rd Floor Conference Room

SPRING 2012
May 25 2012
Philip Poon, WID, UW Madison
Topic: Symmetry, Scale Separation, & Chaos
330PM WID 3rd Floor Conference Room

April 20 2012
Giulio Tononi, Psychiatry, UW Madison
Topic: Information Theory and Consciousness
330PM WID 3rd Floor Conference Room

March 16 2012
Laurence Lowe, WID, UW Madison
Topic: Organizing Data (to get the most out of your model)
330PM WID 3rd Floor Conference Room

February 22 2012
Jordan Ellenberg, Mathematics, UW Madison
Topic: Network Structure, Rules, and Behavior
330PM WID 3rd Floor Conference Room

January 20 2012
Ben Recht, WID, UW Madison
Topic: Compressed Sensing
330PM WID 3rd Floor Conference Room

/