All Seminars

Show:
Title: Compositional Models for Information Extraction
Seminar: N/A
Speaker: Mark Dredze of Johns Hopkins University
Contact: Eugene Agichtein, eugene@mathcs.emory.edu
Date: 2017-03-27 at 4:00PM
Venue: White Hall 207
Download Flyer
Abstract:
Information extraction systems are the backbone of many end-user applications, including question answering, web search and clinical text analysis. These applications depend on underlying technologies that can identify entities and relations as expressed in natural language text. For example, Amazon Echo may answer a user question based on a relation extracted from a news article. A clinical decision support system may offer a physician suggestions based on a symptom identified in the clinical notes from a previous patient visit. In political science, we may seek to aggregate opinions expressed in public comments about a new public policy. Advances in machine learning have led to new neural models for learning effective representations directly from data that improve information extraction tasks. Yet for many tasks, years of research have created hand-engineered features that yield state of the art performance. I will present feature-rich compositional models that combine both hand-engineered features with learned text representations to achieve new state-of-the-art results for relation extraction. These models are widely applicable to problems within natural language processing and beyond. Additionally, I will survey how these models fit into my broader research program by highlighting work by my group on developing new machine learning methods for extracting public health information from clinical and social media text.
Title: TBD
Seminar: N/A
Speaker: TBD of
Contact: TBA
Date: 2017-03-23 at 0:00AM
Venue: TBA
Download Flyer
Abstract:
TBA
Title: Human-centered Data Science for Crisis Informatics
Seminar: Computer Science
Speaker: Marina Kogan of University of Colorado
Contact: Li Xiong, lxiong@emory.edu
Date: 2017-03-23 at 4:00PM
Venue: W201
Download Flyer
Abstract:
Disasters arising from natural hazards are associated with the disruption of existing social structures, but they also result in the creation of new social ties by those affected as they problem-solve alone and together. With social media now being a site for some of this interaction, there is much to learn about the nature of those changing social structures, including how and why they shift. However, the study of this social arena is challenging, because the high-tempo, high-volume convergent nature of crisis events produces vast amounts of social media data, necessitating the use of the data science methods. On the other hand, to glean meaningful insight from the crisis-related social media activity, it is necessary to use methods that account for the complex social context of the user activity, including qualitative analysis.\\ \\In this talk Kogan will show how Human-Centered Data Science provides methodological approaches that both harness the power of computation methods and account for the highly situated nature of social media activity in disaster. Utilizing these methodological approaches, she will show how disaster-related coordination and distributed problem solving take shape on two social media platforms: Twitter and OpenStreetMap.
Title: Congruence of Galois representations
Seminar: Algebra
Speaker: Sujatha Ramdorai of University of British Columbia
Contact: David Zureick-Brown, dzb@mathcs.emory.edu
Date: 2017-03-21 at 4:00PM
Venue: W306
Download Flyer
Abstract:
We consider Galois representations whose residual representations are isomorphic and study what this implies for invariants associated to such representations.
Title: Extremal Problems for Graphs and Hypergraphs
Defense: Dissertation
Speaker: Bill Kay of Emory University
Contact: Bill Kay, w.w.kay@emory.edu
Date: 2017-03-21 at 4:00PM
Venue: W301
Download Flyer
Abstract:
We discuss a pair of papers in extremal combinatorics. One establishes asymptotically the chromatic number of the so-called type graphs, and the other investigates a certain property of oriented hypergraphs (Property O).
Title: Designing Abstract Meaning Representations
Seminar: Computer Science
Speaker: Martha Palmer of University of Colorado
Contact: Jinho Choi, choi@mathcs.emory.edu
Date: 2017-03-17 at 3:00PM
Venue: W301
Download Flyer
Abstract:
Abstract Meaning Representations (AMRs) provide a single, graph-based semantic representation that abstracts away from the word order and syntactic structure of a sentence, resulting in a more language-neutral representation of its meaning. AMRs implements a simplified, standard neo-Davidsonian semantics. A word in a sentence either maps to a concept or a relation or is omitted if it is already inherent in the representation or it conveys inter-personal attitude (e.g., stance or distancing). The basis of AMR is PropBank’s lexicon of coarse-grained senses of verb, noun and adjective relations as well as the roles associated with each sense (each lexicon entry is a ‘roleset’). By marking the appropriate roles for each sense, this level of annotation provides information regarding who is doing what to whom. However, unlike PropBank, AMR also provides a deeper level of representation of discourse relations, non-relational noun phrases, prepositional phrases, quantities and time expressions (which PropBank largely leaves unanalyzed), as well as Named Entity tags with Wikipedia links. Additionally, AMR makes a greater effort to abstract away from language-particular syntactic facts. The latest version of AMR includes adding coreference links across sentences, including links to implicit arguments. This talk will explore the differences between PropBank and AMR, the current and future plans for AMR annotation, and the potential of AMR as a basis for machine translation. It will end with a discussion of areas of semantic representation that AMR is not currently addressing, which remain as open challenges.\\ \\ Martha Palmer is a Professor at the University of Colorado in Linguistics, Computer Science and Cognitive Science, and a Fellow of the Association of Computational Linguistics.. She works on trying to capture elements of the meanings of words that can comprise automatic representations of complex sentences and documents. Supervised machine learning techniques rely on vast amounts of annotated training data so she and her students are engaged in providing data with word sense tags, semantic role labels and AMRs for English, Chinese, Arabic, Hindi, and Urdu, both manually and automatically, funded by DARPA and NSF. These methods have also recently been applied to biomedical journal articles, clinical notes, and geo-science documents, funded by NIH and NSF. She is a co-editor of LiLT, Linguistic Issues in Language Technology, and has been on the CLJ Editorial Board and a co-editor of JNLE. She is a past President of the Association for Computational Linguistics, past Chair of SIGLEX and SIGHAN, co-organizer of the first few Sensevals, and was the Director of the 2011 Linguistics Institute held in Boulder, Colorado.
Title: A Method for Landscape Exploration in Global Optimization.
Seminar: N/A
Speaker: Manuela Manetta of Emory University
Contact: Bree Ettinger, bettinger@mathcs.emory.edu
Date: 2017-03-16 at 1:00PM
Venue: W303
Download Flyer
Abstract:
Finding global minima for a general smooth objective function is a fundamental yet challenging problem arising in applied mathematics. Nevertheless, the most reliable techniques to converge to a minimum are local (gradient descent and Newton's method), and remain trapped in the basins of attraction of the minima found, which could be either local or global. Even in optimization courses, the students are often left with questions, such as, "how do we know that we found the global minimum? How do we know that we have visited the interesting regions of configuration space?" The purpose of this talk is to present methods and ideas that could help the students formulate answers to the questions above. In particular, I will present a new descent technique and a way to explore the landscape of the objective function, with no pretense that this is the answer to the problem, but with the hope of engaging the students. Interested students can begin to take small steps in the right direction toward the development of their own methods.
Title: Curvature through Cubes
Seminar: N/A
Speaker: Michael Carr of Emory University
Contact: Bree Ettinger, bettinger@mathcs.emory.edu
Date: 2017-03-15 at 4:00PM
Venue: W303
Download Flyer
Abstract:
Everyone knows the shortest distance between two points in Euclidean space is a straight line, but what about in more exotic spaces? Mathematicians have been studying paths on curved spaces for as long as we have known we lived on one. More recently the work of Gromov showed that notions of curvature can be extended to spaces that seem to have no curves at all: complexes made from ordinary cubes glued together. We will look at applications of these spaces from robotics to chemistry to recent advances in topology.
Title: Bias and Uncertainty in Information Visualization
Type: Computer Science
Speaker: Michael Correll of University of Washington
Contact: TBA
Date: 2017-03-13 at 4:00PM
Venue: W303
Download Flyer
Abstract:
We often turn to data to help us make sense of an uncertain world. However, the uncertainty in our data is often esoteric, complex, or counter-intuitive. It can be challenging to present this uncertainty, especially to audiences without backgrounds in statistics. Charts, graphs, and other visualizations of data address this issue by making people into “visual statisticians:” we can estimate statistical properties through visual inspection. However, just as statistical measures can be subject to bias, visualizations can also introduce bias. In this talk, I show how designers can intervene to create new visualizations that correct these biases, and improve the judgments of visual statisticians. From this perspective of designing for de-biasing, I focus on two common visualizations: error bars and thematic maps. I present visual alternatives for error bars that avoid “within-the-bar” bias while also promoting statistically grounded comparisons between means. I also present “Surprise Maps,” a technique for thematic maps that relies on Bayesian reasoning to highlight interesting regions that might otherwise be hidden in traditional maps. I conclude with a discussion of remaining challenges for visual de-biasing, and how we might use visualizations to encourage better, data-driven decision-making.
Title: The Distribution Of The Number Of Prime Factors With Restrictions - Variations Of The Classical Theme
Seminar: Algebra
Speaker: Krishna Alladi of University of Florida
Contact: David Zureick-Brown, dzb@mathcs.emory.edu
Date: 2017-02-28 at 4:00PM
Venue: W306
Download Flyer
Abstract:
The study of $\nu(n)$ the number of prime factors of $n$ began with Hardy and Ramanujan in 1917 who showed that $\nu(n)$ has normal order $log\,log\,n$ regardless of whether the prime factors are counted singly or with multiplicity. Their ingenious proof of this utilized uniform upper bounds for $N_k(x)$, the number of integers up to $x$ with $\nu(n)=k$. Two major results followed a few decades later - the Erd\"os-Kac theorem on the distribution more generally of additive functions, and the Sathe-Selberg theorems on the asymptotic behavior of $N_k(x)$ as $k$ varies with $x$ - a significant improvement of Landau's asymptotic estimate for $N_k(x)$ for fixed $k$. We shall consider the distribution of the number of prime factors by imposing certain restrictions - such as (i) requiring all prime factors of $n$ to be $