Complex Systems Speaker and Seminar Series

September 30, 2014.

Dr. Steven Shladover. UC Berkeley.

TUESDAY, SEPTEMBER 30

7 p.m. RASMUSON HALL 110

Road Vehicle Automation: History, Opportunities & Challenges

Abstract:

Road vehicle automation has recently attracted intense interest from the media, the general public and the transportation community, largely based on misconceptions about the level of automation achievable in the foreseeable future. This presentation addresses those misconceptions. Communication and cooperation among automated vehicles and between vehicles and the roadway infrastructure is illustrated from experiments conducted at the PATH Program.

DR. STEVEN SHLADOVER has been researching road vehicle automation systems for 40 years, beginning with his masters and doctoral theses at M.I.T. He is the program manager, Mobility at the California PATH Program of the Institute of Transportation Studies of the University of California at Berkeley. He led PATH’s pioneering research on automated highway systems, including its participation in the National Automated Highway Systems Consortium from 1994-98, and has continued research on fully and partially automated vehicle systems since then. This work has included definition of operating concepts, modeling of automated system operations and benefits, and design, development and testing of full-scale prototype vehicle systems. His target applications have included cooperative adaptive cruise control, automated truck platoons, automated buses and fully-automated vehicles in an automated highway system.

October 3, 2014.

Dr. George I Kamberov. UAA Associate Vice Provost for Research.

Unsupervised Detection of Video Sub-Scenes

Abstract:

The development of methods for event segmentation in video streams is keenly pursued by computer scientists, psychologists, roboticists, educators, neuroscientists, social scientists, and data scientists. There is a substantial evidence that humans perform automatic segmentation at multiple scales. The state of the art machine methods for event segmentation apply only to: 1. Low resolution segmentation of structured videos Hollywood style videos, news casts, and edited home videos following well known and established grammars. 2.Detecting and segmenting a small number of pre-determined event types in surveillance video and uncluttered activity videos.

The automatic event segmentation of videos taken by active operators recording human interactions and activities in the field is of high interest but there has been little progress made by the research community until now. For brevity in we will call such subject centric field grade videos ad hoc videos of events. Human test subjects readily segment ad hoc videos of events into scene-like segments. These segmentations cannot be replicated by the state of the art automatic video segmentation algorithms.

I will describe a new method to segment ad hoc videos of events into atomic semantics units. Motivated by Bellour I call these units sub-scenes. Crowd-sourced experiments show that the (low resolution level) segments detected by human subjects are sequences of sub-scenes. Thus the sub-scenes appear to be a mid-level semantic version of the video shots that are used to piece together scenes by state of the art video segmentation algorithms.

In this talk I will discuss the mathematical and cognitive science underpinnings of the proposed approach; the evaluation of the method; and the applications and projects that motivated the development of this new mid level video segmentation method.

October 17, 2014:.

Dr. Frank Witmer. UAA. Computer Science and Engineering.

Spatially Explicit Modeling of Human-Environment Interactions.

Abstract:

Modeling complex human-environment interactions can take many forms. Most of the data we use to inform our models has a spatial dimension to it, even if it is not recorded as an attribute in the dataset. This presentation discusses the importance of explicitly incorporating the spatial dimension when modeling human-environment relationships. Some common modeling approaches using simulation and regression will be discussed before looking at an example from my research modeling climate variability and violence in sub-Saharan Africa.

October 24, 2014.

ICAN Student Talks. UAA

Dr. Cenek's Complex Systems Researcher Group Projects.

Abstract:

Frazer Tee & Cody McWilliams: EEG Authentication in Noisy and Distracted Environments

Maxwell Franklin: Coupled Dynamics between Fish and Fishermen on Kenai Peninsula - Stochastic Model

Sean Southern: Neuromorhpic Model of the Computer Vision System on Low Power Hardware.

Matthew Devins: Anomaly Detection for Irrigation Canals

October 31, 2014.

Dr. Jamie Trammell. UAA. Geography and Environmental Studies.

Alternative Landscape Futures: Using Spatially-Explicit Scenarios to Model Landscape Change

Abstract:

The use of scenario analysis to explain and quantify uncertainty in environmental systems has grown over the past two decades. In part, this is because scenario analysis provides an explicit method for quantifying past, present and future uncertainties. Thus, scenario analysis has become a useful tool for examining the complex system of landscape change. I present here the beginning steps of the scenario analysis framework we are using to understand comprehensive landscape change in the Kenai watershed. This includes first looking to the past to quantify how the landscape has changed and identifying the drivers of the change. Then, we ask the question of how well is the landscape currently working, based on what we know about the past. Over the next 2 years, we will construct plausible scenarios of landscape change that integrate knowledge about past and present conditions to identify both the” known unknowns” and some of the “unknown unknowns”.

November 7, 2014.

Drs. Jonathan E Alevy and Lance Howe. UAA. Economics and Public Policy.

Choice, risk, and motivated cognition: contributions from economic models

Abstract:

This talk develops core concepts in theoretical and empirical modeling of economic choices, drawing also on work that incorporates findings from other social and behavioral sciences. Bayesian decision making and utility maximization are traditional tools that provide useful benchmarks for understanding economic choices. We introduce these tools and provide an overview of how models of economic decisions have been broadened to address informational asymmetries, details of cognition (dual process models), and personal identity. Applications to research in progress on recreational demand on the Kenai Peninsula and to the impact of identity on inference (motivated cognition) are discussed. Choice, risk, and motivated cognition: contributions from economic models

November 14, 2014.

Mollie McCarthy. UAA. Biological Sciences.

Unearthing past sockeye salmon populations on the Kenai Peninsula

Abstract:

The Kenai River’s sockeye salmon runs are supported by production from both glacial and clearwater lakes, introducing an opportunity to explore how salmon population dynamics and climatic drivers differ in these distinctive habitat types. For this study, sockeye salmon population fluctuations will be reconstructed on historic (1970-present) and paleo (up to 2000 ybp) time scales using existing ADF&G data and sediment cores, respectively. We hypothesize that productivity will be out of phase, with clearwater lakes being most productive during warm periods and glacial lakes being most productive during cool periods. Since warming impacts on glacial systems include increased suspended silt and decreased productivity, cooling periods would enhance euphotic volume and prey productivity. We constructed brood tables and calculated recruits per spawner (R/S) for the major clearwater (i.e., Russian River) and glacial (the remainder of the run) components of the Kenai sockeye run using escapement and harvest data from 1970-present. Paleo-data is currently being analyzed and age models for cores are being constructed to obtain paleo-salmon population predictions.

November 21, 2014.

Dr. Alan Boraas and Hannah Johnson. UAA/KPC. Anthropology.

Salmon, Complexity, and the Cook Inlet Dena’ina

Abstract:

This presentation will be two parts. Alan Boraas will talk about the invention of Ełnen T’uh, underground cold storage pits, by the Dena’ina about A.D. 1000. Elegantly simple, underground cold storage solved the problem of effectively storing salmon that come in huge abundance in summer for winter consumption. This triggered cultural complexity, including spiritual complexity, that led to the pre-contact Dena’ina being one of the world’s great sustainable cultures. A.D. 1000 coincides with the Medieval Warm Period, so an additional question is “did warming trigger cultural complexity?”

Hannah Johnson will speak on her current research with contemporary Kenaitze Dena’ina asking the question, “How are salmon important in your lives nutritionally, socially, and spiritually?” If salmon, particularly king salmon, are impacted, or if access to salmon are impacted, how will the Kenaitze be affected culturally.

November 5, 2014.

Kacy Krieger. Alaska Hydrography Database Coordinator, Alaska Natural Heritage Program UAA.

Data Driven Integration.

Abstract:

What are the benefits and challenges of data access, management and integration in a “Big Data” world? With so much of what we do and study dependent on information access, it is important that current, accurate and useful data makes it through the right conduits to those that need it. Using examples from the geospatial and hydrographic world, we will explore how data resources are valuable tools used by resource managers, scientists and the public for analysis, mapping and recreation, and how representatives from many organizations recognized the need for a common and accurate dataset. These representatives have come together to coordinate and update key geospatial datasets in Alaska and developed a successful data stewardship model to do so. This collaborative model allows a diverse group of data users, resource managers and scientists the ability to maintain, update and consume an authoritative hydrography dataset. This model can serve as an example of how data needs across a wide range of disciplines can be addressed.