arXiv subjects

Christopher M. Homan

Publications and source records attributed to Christopher M. Homan.

32 records · Page 2Linked to original sources

arXiv2021

Domain-specific MT for Low-resource Languages: The case of Bambara-French

Translating to and from low-resource languages is a challenge for machine translation (MT) systems due to a lack of parallel data. In this paper we address the issue of domain-specific MT for Bambara, an under-resourced Mande language spoken in Mali. We present the first domain-specific parallel dataset for MT of Bambara into and from French. We discuss challenges in working with small quantities of domain-specific data for a low-resource language and we present the results of machine learning experiments on this data.

cs.CL↗

arXiv2020

Neighborhood-based Pooling for Population-level Label Distribution Learning

Supervised machine learning often requires human-annotated data. While annotator disagreement is typically interpreted as evidence of noise, population-level label distribution learning (PLDL) treats the collection of annotations for each data item as a sample of the opinions of a population of human annotators, among whom disagreement may be proper and expected, even with no noise present. From this perspective, a typical training set may contain a large number of very small-sized samples, one for each data item, none of which, by itself, is large enough to be considered representative of the underlying population's beliefs about that item. We propose an algorithmic framework and new statistical tests for PLDL that account for sampling size. We apply them to previously proposed methods for sharing labels across similar data items. We also propose new approaches for label sharing, which we call neighborhood-based pooling.

cs.LG↗

arXiv2020

Assessing Human Translations from French to Bambara for Machine Learning: a Pilot Study

We present novel methods for assessing the quality of human-translated aligned texts for learning machine translation models of under-resourced languages. Malian university students translated French texts, producing either written or oral translations to Bambara. Our results suggest that similar quality can be obtained from either written or spoken translations for certain kinds of texts. They also suggest specific instructions that human translators should be given in order to improve the quality of their work.

cs.CL↗

arXiv2019

Twitter Job/Employment Corpus: A Dataset of Job-Related Discourse Built with Humans in the Loop

We present the Twitter Job/Employment Corpus, a collection of tweets annotated by a humans-in-the-loop supervised learning framework that integrates crowdsourcing contributions and expertise on the local community and employment environment. Previous computational studies of job-related phenomena have used corpora collected from workplace social media that are hosted internally by the employers, and so lacks independence from latent job-related coercion and the broader context that an open domain, general-purpose medium such as Twitter provides. Our new corpus promises to be a benchmark for the extraction of job-related topics and advanced analysis and modeling, and can potentially benefit a wide range of research communities in the future.

cs.CL↗

arXiv2017

Learning from various labeling strategies for suicide-related messages on social media: An experimental study

Suicide is an important but often misunderstood problem, one that researchers are now seeking to better understand through social media. Due in large part to the fuzzy nature of what constitutes suicidal risks, most supervised approaches for learning to automatically detect suicide-related activity in social media require a great deal of human labor to train. However, humans themselves have diverse or conflicting views on what constitutes suicidal thoughts. So how to obtain reliable gold standard labels is fundamentally challenging and, we hypothesize, depends largely on what is asked of the annotators and what slice of the data they label. We conducted multiple rounds of data labeling and collected annotations from crowdsourcing workers and domain experts. We aggregated the resulting labels in various ways to train a series of supervised models. Our preliminary evaluations show that using unanimously agreed labels from multiple annotators is helpful to achieve robust machine models.

cs.LG↗

arXiv2015

Job-related discourse on social media

Working adults spend nearly one third of their daily time at their jobs. In this paper, we study job-related social media discourse from a community of users. We use both crowdsourcing and local expertise to train a classifier to detect job-related messages on Twitter. Additionally, we analyze the linguistic differences in a job-related corpus of tweets between individual users vs. commercial accounts. The volumes of job-related tweets from individual users indicate that people use Twitter with distinct monthly, daily, and hourly patterns. We further show that the moods associated with jobs, positive and negative, have unique diurnal rhythms.

cs.SI↗

arXiv2014

Tuning the Diversity of Open-Ended Responses from the Crowd

Crowdsourcing can solve problems that current fully automated systems cannot. Its effectiveness depends on the reliability, accuracy, and speed of the crowd workers that drive it. These objectives are frequently at odds with one another. For instance, how much time should workers be given to discover and propose new solutions versus deliberate over those currently proposed? How do we determine if discovering a new answer is appropriate at all? And how do we manage workers who lack the expertise or attention needed to provide useful input to a given task? We present a mechanism that uses distinct payoffs for three possible worker actions---propose,vote, or abstain---to provide workers with the necessary incentives to guarantee an effective (or even optimal) balance between searching for new answers, assessing those currently available, and, when they have insufficient expertise or insight for the task at hand, abstaining. We provide a novel game theoretic analysis for this mechanism and test it experimentally on an image---labeling problem and show that it allows a system to reliably control the balance betweendiscovering new answers and converging to existing ones.

cs.HC↗

arXiv2013

Respondent-Driven Sampling in Online Social Networks

Respondent-driven sampling (RDS) is a commonly used method for acquiring data on hidden communities, i.e., those that lack unbiased sampling frames or face social stigmas that make their mem- bers unwilling to identify themselves. Obtaining accurate statistical data about such communities is important because, for instance, they often have different health burdens from the greater population, and without good statistics it is hard and expensive to effectively reach them for pre- vention or treatment interventions. Online social networks (OSN) have the potential to transform RDS for the better. We present a new RDS recruitment protocol for (OSNs) and show via simulation that it out- performs the standard RDS protocol in terms of sampling accuracy and approaches the accuracy of Markov chain Monte Carlo random walks.

cs.SI↗

arXiv2008

Dichotomy Results for Fixed Point Counting in Boolean Dynamical Systems

We present dichotomy theorems regarding the computational complexity of counting fixed points in boolean (discrete) dynamical systems, i.e., finite discrete dynamical systems over the domain {0,1}. For a class F of boolean functions and a class G of graphs, an (F,G)-system is a boolean dynamical system with local transitions functions lying in F and graphs in G. We show that, if local transition functions are given by lookup tables, then the following complexity classification holds: Let F be a class of boolean functions closed under superposition and let G be a graph class closed under taking minors. If F contains all min-functions, all max-functions, or all self-dual and monotone functions, and G contains all planar graphs, then it is #P-complete to compute the number of fixed points in an (F,G)-system; otherwise it is computable in polynomial time. We also prove a dichotomy theorem for the case that local transition functions are given by formulas (over logical bases). This theorem has a significantly more complicated structure than the theorem for lookup tables. A corresponding theorem for boolean circuits coincides with the theorem for formulas.

cs.CC↗

arXiv2007

Guarantees for the Success Frequency of an Algorithm for Finding Dodgson-Election Winners

In the year 1876 the mathematician Charles Dodgson, who wrote fiction under the now more famous name of Lewis Carroll, devised a beautiful voting system that has long fascinated political scientists. However, determining the winner of a Dodgson election is known to be complete for the Θ_2^p level of the polynomial hierarchy. This implies that unless P=NP no polynomial-time solution to this problem exists, and unless the polynomial hierarchy collapses to NP the problem is not even in NP. Nonetheless, we prove that when the number of voters is much greater than the number of candidates--although the number of voters may still be polynomial in the number of candidates--a simple greedy algorithm very frequently finds the Dodgson winners in such a way that it ``knows'' that it has found them, and furthermore the algorithm never incorrectly declares a nonwinner to be a winner.

cs.DS↗

arXiv2006

Plane Decompositions as Tools for Approximation

Tree decompositions were developed by Robertson and Seymour. Since then algorithms have been developed to solve intractable problems efficiently for graphs of bounded treewidth. In this paper we extend tree decompositions to allow cycles to exist in the decomposition graph; we call these new decompositions plane decompositions because we require that the decomposition graph be planar. First, we give some background material about tree decompositions and an overview of algorithms both for decompositions and for approximations of planar graphs. Then, we give our plane decomposition definition and an algorithm that uses this decomposition to approximate the size of the maximum independent set of the underlying graph in polynomial time.

cs.DS↗

arXiv2005

Cluster Computing and the Power of Edge Recognition

We study the robustness--the invariance under definition changes--of the cluster class CL#P [HHKW05]. This class contains each #P function that is computed by a balanced Turing machine whose accepting paths always form a cluster with respect to some length-respecting total order with efficient adjacency checks. The definition of CL#P is heavily influenced by the defining paper's focus on (global) orders. In contrast, we define a cluster class, CLU#P, to capture what seems to us a more natural model of cluster computing. We prove that the naturalness is costless: CL#P = CLU#P. Then we exploit the more natural, flexible features of CLU#P to prove new robustness results for CL#P and to expand what is known about the closure properties of CL#P. The complexity of recognizing edges--of an ordered collection of computation paths or of a cluster of accepting computation paths--is central to this study. Most particularly, our proofs exploit the power of unique discovery of edges--the ability of nondeterministic functions to, in certain settings, discover on exactly one (in some cases, on at most one) computation path a critical piece of information regarding edges of orderings or clusters.

cs.CC↗

arXiv2005

The Complexity of Computing the Size of an Interval

Given a p-order A over a universe of strings (i.e., a transitive, reflexive, antisymmetric relation such that if (x, y) is an element of A then |x| is polynomially bounded by |y|), an interval size function of A returns, for each string x in the universe, the number of strings in the interval between strings b(x) and t(x) (with respect to A), where b(x) and t(x) are functions that are polynomial-time computable in the length of x. By choosing sets of interval size functions based on feasibility requirements for their underlying p-orders, we obtain new characterizations of complexity classes. We prove that the set of all interval size functions whose underlying p-orders are polynomial-time decidable is exactly #P. We show that the interval size functions for orders with polynomial-time adjacency checks are closely related to the class FPSPACE(poly). Indeed, FPSPACE(poly) is exactly the class of all nonnegative functions that are an interval size function minus a polynomial-time computable function. We study two important functions in relation to interval size functions. The function #DIV maps each natural number n to the number of nontrivial divisors of n. We show that #DIV is an interval size function of a polynomial-time decidable partial p-order with polynomial-time adjacency checks. The function #MONSAT maps each monotone boolean formula F to the number of satisfying assignments of F. We show that #MONSAT is an interval size function of a polynomial-time decidable total p-order with polynomial-time adjacency checks. Finally, we explore the related notion of cluster computation.

cs.CC↗

arXiv2000

Low Ambiguity in Strong, Total, Associative, One-Way Functions

Rabi and Sherman present a cryptographic paradigm based on associative, one-way functions that are strong (i.e., hard to invert even if one of their arguments is given) and total. Hemaspaandra and Rothe proved that such powerful one-way functions exist exactly if (standard) one-way functions exist, thus showing that the associative one-way function approach is as plausible as previous approaches. In the present paper, we study the degree of ambiguity of one-way functions. Rabiand Sherman showed that no associative one-way function (over a universe having at least two elements) can be unambiguous (i.e., one-to-one). Nonetheless, we prove that if standard, unambiguous, one-way functions exist, then there exist strong, total, associative, one-way functions that are $\mathcal{O}(n)$-to-one. This puts a reasonable upper bound on the ambiguity.

cs.CC↗