WSU Vancouver Mathematics and Statistics Seminar
 WSU Vancouver Mathematics and Statistics Seminar (Spring 2019) Welcome to the WSU Vancouver Seminar in Mathematics and Statistics! The Seminar meets on Wednesdays at 1:10-2:00 PM in VUB 221, unless mentioned otherwise. This is the Undergraduate Building (marked "N" in the campus map), where all Math/Stat faculty sit. The seminar is open to the public, and here is some information for visitors. Students could sign up for Math 592 (titled Seminar in Analysis) for 1 credit. Talks will be given by external speakers, as well as by WSUV faculty and students. Contact the organizer Bala Krishnamoorthy if you want to invite a speaker, or to give a talk.
Date Speaker Topic Slides
Jan   9 Organizational meeting
Jan 16 Andrew Fowler, Nuance and OHSU Improved inference and autotyping in EEG-based brain computer interface typing systems

The RSVP Keyboard is a typing system for people with severe physical disabilities, specifically those with locked-in syndrome (LIS). It uses signals from an electroencephalogram (EEG) combined with information from an n-gram language model to select letters to be typed. One characteristic of the system as currently configured is that it does not keep track of past EEG observations, i.e., observations of user intent made while the user was in a different part of a typed message. We present a principled approach for taking all past observations into account, and show that this method results in a 20% increase in simulated typing speed under a variety of conditions on realistic stimuli. We also show that this method allows for a principled and improved estimate of the probability of the backspace symbol, by which mis-typed symbols are corrected. Finally, we demonstrate the utility of automatically typing likely letters in certain contexts, a technique that achieves increased typing speed under our new method, though not under the baseline approach.

slides
Jan 23 Michael Jordan, UC Berkeley (MBI seminar) Machine Learning: Dynamical, Economic and Stochastic Perspectives
Jan 31 Caroline Uhler, MIT SIAM PNW Seminar at 3 PM in VUB 221:
From Causal Inference to Gene Regulation

A recent break-through in genomics makes it possible to perform perturbation experiments at a very large scale. The availability of such data motivates the development of a causal inference framework that is based on observational and interventional data. We first characterize the causal relationships that are identifiable from interventional data. In particular, we show that imperfect interventions, which only modify (i.e., without necessarily eliminating) the dependencies between targeted variables and their causes, provide the same causal information as perfect interventions, despite being less invasive. Second, we present the first provably consistent algorithm for learning a causal network from a mix of observational and interventional data. This requires us to develop new results in geometric combinatorics. In particular, we introduce DAG associahedra, a family of polytopes that extend the prominent graph associahedra to the directed setting. We end by discussing applications of this causal inference framework to the estimation of gene regulatory networks.

Join on GotoMeeting.

Feb   6 Adam Erickson, WSUV Validation of a simple biogeochemistry variant of SORTIE-PPA in two temperate forests using the Erde modeling framework

While vegetation model development has accelerated over the past three decades, less progress has been made in software interfaces that bind models and data. Such interfaces become necessary as model complexity increases, making models more cumbersome for new users to parameterize, run, and analyze. Furthermore, model wrappers may provide new modeling capabilities, such as Bayesian optimization. Toward this end, we have developed a simple geoscientific simulation model API and toolkit in R and Python known as the Earth-systems Research and Development Environment (Erde). While Erde is primarily intended for vegetation models, its data structures and algorithms are applicable across a range of geoscientific modeling domains.

We demonstrate an application of Erde with a simple biogeochemistry variant of SORTIE-PPA known as PPA-SiBGC. The PPA-SiBGC model combines the Perfect Plasticity Approximation with explicit above-and-below-ground biogeochemical pools and simple flux models. We parameterized, ran, and validated PPA-SiBGC at two research forests in the Eastern United States: (1) Harvard Forest, Massachusetts (HF-EMS) and (2) Jones Ecological Research Center, Georgia (JERC-RD). We assessed model fitness using these temporal metrics: net ecosystem exchange, aboveground net primary production, aboveground biomass, C, and N, belowground biomass, C, and N, soil respiration, and, species total biomass and relative abundance. Without applying any optimization, we found that a simple biogeochemistry variant of SORTIE-PPA was able to outperform an established forest landscape model (LANDIS-II NECN) across the metrics tested. While LANDIS-II NECN showed better NEE fit, PPA-SiBGC demonstrated better overall correspondence to field data for both sites (HF-EMS: $$\overline{R^2} = 0.73, +0.07$$, $$\overline{RMSE} = 4.68, -9.96$$; JERC-RD: $$\overline{R^2} = 0.73, +0.01$$, $$\overline{RMSE} = 2.18, -1.64$$)

Feb 13 No seminar
Feb 20 3-Minute Thesis Contest prep
Feb 28 Rebecca Doerge, Carnegie Mellon Math Colloquium at 4:10 PM in VECS 309:
The Future of Statistical Bioinformatics and Genomics in the Automated World of Agriculture

World population is expected to reach 9.6 billion by 2050, and crop yields are not keeping pace fast enough to avoid widespread food shortages. Modern molecular breeding programs are extremely effective, and can generate annual yield increases of around 4%. However, it is not scalable due to both the human labor and expertise required. Employing robots in agriculture has great potential to address chronic issues that are challenging food systems in the developing world. While quantitative trait locus (QTL) mapping associates molecular with variation in traits, there are new challenges for the statistical bioinformatics and genomics communities in how data are collected, stored and analyzed with respect to understanding relationships between genetic composition and phenotype. Using automation, the expectation is a dramatic increase, factor of 100+, in the number of individuals in the breeding population that can be evaluated. Artificial intelligence based technologies and analysis allow rapid evaluation, and thus make molecular breeding adaptive to environmental and other changes. Automated phenotyping is a transformational technology that can be applied worldwide.

Mar   6 Nathaniel Saul, WSUV Stitch-fix for Mapper

Mapper is a powerful construction from topological data analysis that is designed for the analysis and visualization of multivariate data. We investigate a method for stitching a pair of mappers together, and study a topological notion of information gain as well as a general measure of correlation during such a process. We are inspired by the ideas of stepwise regression for model selection and of scatterplot matrices for visualization, and introduce a topological analogue of the scatterplot matrix for the mapper, and study the degree of topological correlation between filter functions.

This is joint work with Bei Wang and Bala Krishnamoorthy.

Mar 13 Spring break (no seminar)
Mar 20 Andreas Wild, Intel labs Loihi - Intel's neuromorphic computing platform

Loihi is a massively parallel, event-driven, in-memory hardware architecture with on-chip learning for accelerating energy-efficient neural network computations which has been developed by Intel Labs' Neuromorphic Computing Lab. Possible application areas span from small edge devices, autonomous driving and robotics to the cloud.

This talk will give an introduction to neuromorphic computing, the Loihi architecture and provide an overview over our latest algorithmic research results to stimulate discussion on potential application areas.

Mar 21 Viktoria Taroudaki, EWU. Math Colloquium at 4:10 PM in VECS 309:
Using optimization and statistics to improve images and acoustical signals

Images that imaging devices record and sounds that microphones record are corrupted by blur and noise. In this talk, we will see how a filtering method can help eliminate the effects of blur and noise and under what conditions. The results of the filtering method will be compared in both the two- and the one-dimensional cases and we will see one uncertainty quantification measure.

Apr   3 Nella Ludlow, WSU EECS Quantum Computing—Its Time is Now

Most technologists have some basic understanding of quantum computing. Quantum bits are represented by qubits, and information isn't simply a zero or one, it can be both at the same time, representing much more information. A quantum effect known as entanglement linking two particles allows instantaneous communication.

Ultimately, quantum computing promises to be a disruptive technology with such dramatic speed improvements that real tractable solutions to hard problems could be solved in hours and days. These same problems would take hundreds of years to solve on our best supercomputers, known as a classical computer, by brute-forcing their way searching through all possible solutions.

This presentation we will look at Quantum Computing with an overview of Quantum Mechanics, qubits and quantum logic, linear algebra and quantum gates and transformations, quantum algorithms such as Shor's Factoring and Grover's search algorithms, and an evaluation of the level of performance of current quantum computers and software platforms.

We will discuss the recent rapid increase in quantum company growth, new quantum hardware and software offerings, and new government funding, and the push to develop a quantum-literate workforce.

These recent activities suggest the Quantum Information Science "tipping point" is now!

Apr  4 Haonan Wang, Colorado State U. Math Colloquium at 4:10 PM in VMMC 102Q:
Statistical Analysis of Complex and Inhomogeneous Data with Application to Neuroscience

In this talk, we consider two types of data from neuroscience: neuromorphology data and neuron activity data. First, we focus on data extracted from brain neuron cells of rodents and model each neuron as a data object with topological and geometric properties characterizing the branching structure, connectedness and orientation of a neuron. We define the notions of topological and geometric medians as well as quantiles based on newly-developed curve representations. In addition, we take a novel approach to define the Pareto medians and quantiles through a multi-objective optimization problem. In particular, we study two different objective functions which measure the topological variation and geometric variation respectively. Analytical solutions are provided for topological and geometric medians and quantiles, and in general, for Pareto medians and quantiles, the genetic algorithm is implemented. The proposed methods are demonstrated in a simulation study and are also applied to analyze a real data set of pyramidal neurons from the hippocampus. Next, we model the neuron spiking activity through nonlinear dynamical systems. We adapt the Volterra series expansion of an analytic function to account for the point-process nature of multiple inputs and a single output (MISO) in a neural ensemble. Our model describes the transformed spiking probability for the output as the sum of kernel-weighted integrals of the inputs. The kernel functions need to be identified and estimated, and both local sparsity (kernel functions may be zero on part of their support) and global sparsity (some kernel functions may be identically zero) are of interest. The kernel functions are approximated by B-splines and a penalized likelihood-based approach is proposed for estimation. Even for moderately complex brain functionality, the identification and estimation of this sparse functional dynamical model poses major computational challenges, which we address with big data techniques that can be implemented on a single, multi-core server. The performance of the proposed method is demonstrated using neural recordings from the hippocampus of a rat during open field tasks. This is the joint work with Dr. Sienkiewicz, Professor Breidt and Professor Song.

Apr 11 Fred Adler, U. Utah Math Colloquium at 4:10 PM in VECS 309:
Modeling mutualisms in plants and cancer

Essentially every biological interaction has a combination of positive and negative feedbacks, but capturing that in models proves surprisingly difficult. I will discuss two models of so-called mutualisms. Using a resource-mediated interaction of plants with their soil microbiome, we will look at whether mutualisms buffer the system against environmental perturbations. With the second, a model of how cancer cells coopt the behavior of promoters of cell growth, we will try to understand why cancer is rare but possible.

Apr 17 Noah Doss, WSUV Kal-Toh: The Vulcan Game of Triconnected Planar Graphs

Star Trek: Voyager, a television program from the late 90s, featured a game played by the hyper-logical Vulcan race called Kal-Toh. Said by characters to dwarf chess in complexity, the game, which features the placement of a clump of metal rods, can be well-represented by the framework we know as graph theory. I will discuss a paper written by Terry David Anderson from the University of Waterloo in which, as a representation of Kal-Toh, studies the problem of determining the existence of a polyhedral subgraph of some graph G. In the exploration of this problem, Anderson finds that both showing the existence of such a subgraph and obtaining an optimal strategy, in a game of two players, for achieving such a subgraph first are NP-hard problems.

Apr 24 Daniel Taylor-Rodriguez, Portland State U. A Bayesian nonparametric multiple testing procedure for comparing several treatments against a control

We propose a Bayesian nonparametric strategy to test for differences between a control group and several treatment regimes. Many of the existing tests for this type of comparison are based on the differences between location or scale parameters. In contrast, our approach identifies differences across the entire distributions, avoids strong modeling assumptions over the distributions for each treatment, and accounts for multiple testing through the prior distribution on the space of hypotheses. The proposal is compared to other commonly used hypothesis testing procedures under simulated scenarios.

Apr 25 David Gleich, Purdue U. SIAM PNW Seminar at 4 PM:
Higher-order clustering of complex networks