Michael Martin

Technische Universität München, Computer Science, Faculty Member

Followers

Following

Public Views

Interests

Uploads

Papers by Michael Martin

The eBible Corpus: Data and Model Benchmarks for Bible Translation for Low-Resource Languages

arXiv (Cornell University), Apr 19, 2023

Efficiently and accurately translating a corpus into a low-resource language remains a challenge,... more Efficiently and accurately translating a corpus into a low-resource language remains a challenge, regardless of the strategies employed, whether manual, automated, or a combination of the two. Many Christian organizations are dedicated to the task of translating the Holy Bible into languages that lack a modern translation. Bible translation (BT) work is currently underway for over 3000 extremely low resource languages. We introduce the eBible corpus: a dataset containing 1009 translations of portions of the Bible with data in 833 different languages across 75 language families. In addition to a BT benchmarking dataset, we introduce model performance benchmarks built on the No Language Left Behind (NLLB) neural machine translation (NMT) models. Finally, we describe several problems specific to the domain of BT and consider how the established data and model benchmarks might be used for future translation efforts. For a BT task trained with NLLB, Austronesian and Trans-New Guinea language families achieve 35.1 and 31.6 BLEU scores respectively, which spurs future innovations for NMT for low-resource languages in Papua New Guinea.

Download

A Multi-Paradigm Approach to Grounded Theory

Information Systems Foundations: Theory, Representation and Reality

While grounded theory methodology was intended to be used as a general research method accepting ... more While grounded theory methodology was intended to be used as a general research method accepting any type of data, most grounded theory studies in the literature refer to research based on qualitative data. This paper aims to address this perceived neglect by describing our experiences and our approach while using grounded theory across qualitative and quantitative research paradigms. The case-based discussion presented in this paper, explains how the combination of these paradigms in exploratory studies can benefit research outcomes. The discussion follows the authors' approach to a competitive research grant opportunity that resulted in a comprehensive study into the use and management of information and communication technology in 231 Australian organisations. We propose that the use of classic grounded theory as a general research method enables researchers to capitalise on greater opportunities to participate in substantial team-based exploratory research endeavours.

Download

An Elephant Never Forgets: Effective Analogies for Teaching Statistical Modeling

Topics from Australian Conferences on Teaching Statistics, 2014

Analogies are useful and potent tools for introducing new topics in statistics to students. Marti... more Analogies are useful and potent tools for introducing new topics in statistics to students. Martin (J Stat Educ 11(2); Proceedings of the 6th Australian Conference on Teaching Statistics) considered the case for teaching with analogies in introductory statistics courses, and also gave many examples of particular analogies that had been successfully used to make difficult statistical concepts more accessible to students. In this chapter, we explore more deeply analogies for statistical concepts from more advanced topics such as regression modeling and high-dimensional data.

Music of 18 Performances: Evaluating Apps and Agents with Free Improvisation

We present a study where a small group of experienced iPad musicians evaluated a system of three ... more We present a study where a small group of experienced iPad musicians evaluated a system of three musical touchscreenappsandtwoserver-basedagentsover18controlled improvisations. Theperformers’perspectiveswererecorded through surveys, interviews, and interaction data. Our agent classifies the touch gestures of the performers and identifies new sections in the improvisations while a control agent returns similar messages sourced from a statistical model. The three touch-screen apps respond according to design paradigms of reward, support, and disruption. In this study of an ongoing musical practice, significant effects were observed due to the apps’ interfaces and how they respond to agent interactions. The “reward” app received the highest ratings. The results were used to iterate the app designs for later performances.

Download

OntoWiki – An authoring, publication and visualization interface for the Data Web

Semantic Web, 2015

OntoWiki is a front-end application for the Semantic Data Web, which was originally developed to ... more OntoWiki is a front-end application for the Semantic Data Web, which was originally developed to support distributed knowledge engineering scenarios. Due to its extensibility it also serves as a development framework for knowledge intensive applications. On the surface, OntoWiki is a generic user interface for arbitrary RDF knowledge graphs. It supports the navigation through RDF knowledge bases using SPARQL-generated lists, tables and trees (e.g. class trees and taxonomies). All resources are automatically represented as hyperlinks and backlinks are created whenever feasible, thus enabling users to easily traverse entire knowledge graphs. Since all collections of resources displayed in OntoWiki are generated by SPARQL queries, they can be further refined by applying additional filters. In order to explore large datasets, a comprehensive statistical data management and visualization method was integrated. We render an architectural overview and explain the navigation, exploration and visualization features provided for RDF based information and derived statistics. To showcase the versatility of OntoWiki and its various deployments in the Linked Data Web, we describe some large-scale use cases in the domains of enterprise data integration, governmental data publication and digital humanities. This article is the first comprehensive presentation of OntoWiki summarizing the advancements after the initial publication in 2006 .

Download

Answering Your Patients' Questions About Cross-Linking

“It's Like… You Know”: The Use of Analogies and Heuristics in Teaching Introductory Statistical Methods

Journal of Statistics Education, 2003

Download

Improving the Performance of Semantic Web Applications with SPARQL Query Caching

Lecture Notes in Computer Science, 2010

The performance of triple stores is one of the major obstacles for the deployment of semantic tec... more The performance of triple stores is one of the major obstacles for the deployment of semantic technologies in many usage scenarios. In particular, Semantic Web applications, which use triple stores as persistence backends, trade performance for the advantage of flexibility with regard to information structuring. In order to get closer to the performance of relational database-backed Web applications, we developed an approach for improving the performance of triple stores by caching query results and even complete application objects. The selective invalidation of cache objects, following updates of the underlying knowledge bases, is based on analysing the graph patterns of cached SPARQL queries in order to obtain information about what kind of updates will change the query result. We evaluated our approach by extending the BSBM triple store benchmark with an update dimension as well as in typical Semantic Web application scenarios.

Download

Polynomial formal verification of of multipliers

Proceedings. 15th IEEE VLSI Test Symposium (Cat. No.97TB100125)

Until recently, verifying multipliers with formal methods was not feasible, even for small input ... more Until recently, verifying multipliers with formal methods was not feasible, even for small input word sizes. About two years ago, a new data structure, called Multiplicative Binary Moment Diagram (*BMD), was introduced for representing arithmetic functions over Boolean variables. Based on this data structure, methods were proposed by which verification of multipliers with input word sizes of up to 256 bits became now feasible. Only experimental data has been provided for these verification methods until now. In this paper we give a formal proof that logic verification using *BMDs is polynomially bounded in both space and time, when applied to the class of Wallace-tree like multipliers.

Download

Graphical Displays

Wiley StatsRef: Statistics Reference Online, 2014

Tracking Chemical Changes in a Live Cell: Biomedical Applications of SR-FTIR Spectromicroscopy

Spectroscopy, 2003

Synchrotron radiation-based Fourier transform infrared (SR-FTIR) spectromicroscopy is a newly eme... more Synchrotron radiation-based Fourier transform infrared (SR-FTIR) spectromicroscopy is a newly emerging bioanalytical and imaging tool. This unique technique provides mid‒infrared (IR) spectra, hence chemical information, with high signal-to-noise at spatial resolutions as fine as 3 to 10 microns. Thus it enables researchers to locate, identify, and track specific chemical events within an individual living mammalian cell. Mid-IR photons are too low in energy (0.05−0.5 eV) to either break bonds or to cause ionization. In this review, we show that the synchrotron IR beam has no detectable effects on the short- and long-term viability, reproductive integrity, cell-cycle progression, and mitochondrial metabolism in living human cells, and produces only minimal sample heating (<0.5°C). We will then present several examples demonstrating the application potentials of SR-FTIR spectromicroscopy in biomedical research. These will include monitoring living cells progressing through the cel...

Download

An Adaptive, Automatic Multiple-Case Deletion Technique for Detecting Influence in Regression

Technometrics, 2014

ABSTRACT Critical to any regression analysis is the identification of observations that exert a s... more ABSTRACT Critical to any regression analysis is the identification of observations that exert a strong influence on the fitted regression model. Traditional regression influence statistics such as Cook&#39;s distance and DFFITS, each based on deleting single observations, can fail in the presence of multiple influential observations if these influential observations “mask” one another, or if other effects such as “swamping” occur. Masking refers to the situation where an observation reveals itself as influential only after one or more other observations are deleted. Swamping occurs when points that are not actually outliers/influential are declared to be so because of the effects on the model of other unusual observations. One computationally expensive solution to these problems is the use of influence statistics that delete multiple rather than single observations. In this paper we build on previous work to produce a computationally feasible algorithm for detecting an unknown number of influential observations in the presence of masking. An important difference between our proposed algorithm and existing methods is that we focus on the data that remains after observations are deleted, rather than on the deleted observations themselves. Further, our approach uses a novel confirmatory step designed to provide a secondary assessment of identified observations.

Automatic extraction of object-oriented component interfaces

Proceedings of the international symposium on Software testing and analysis - ISSTA '02, 2002

Component-based software design is a popular and effective approach to designing large systems. W... more Component-based software design is a popular and effective approach to designing large systems. While components typically have well-defined interfaces, sequencing information-which calls must come in which order-is often not formally specified. This paper proposes using multiple finite state machine (FSM) submodels to model the interface of a class. A submodel includes a subset of methods that, for example, implement a Java interface, or access some particular field. Each state-modifying method is represented as a state in the FSM, and transitions of the FSMs represent allowable pairs of consecutive methods. In addition, state-preserving methods are constrained to execute only under certain states. We have designed and implemented a system that includes static analyses to deduce illegal call sequences in a program, dynamic instrumentation techniques to extract models from execution runs, and a dynamic model checker that ensures that the code conforms to the model. Extracted models can serve as documentation; they can serve as constraints to be enforced by a static checker; they can be studied directly by developers to determine if the program is exhibiting unexpected behavior; or they can be used to determine the completeness of a test suite. Our system has been run on several large code bases, including the joeq virtual machine, the basic Java libraries, and the Java 2 Enterprise Edition library code. Our experience suggests that this approach yields useful information.

Download

Feedforward effects on predictions in a dynamic battle scenario

PsycEXTRA Dataset

Commanders face many challenges in their efforts to control the battlefield. Friction (i.e., sour... more Commanders face many challenges in their efforts to control the battlefield. Friction (i.e., sources of delay) in the commander's control system, coupled with the dynamics of the battlefield, requires commanders to act before threatening battlefield events occur. Effective control of the battlefield thus requires accurate predictions. This paper describes the results of a preliminary study concerned with the effect of FeedForward (FF) on the accuracy of predictions in dynamic battle situations. FF, given in the form of expert advice prior to simulated battle, did not reliably improve predictions. Exploratory analyses, however, indicate that FF guided attention to a subset of the task variables important for accurate prediction. Furthermore, FF produced quicker and more decisive victories than practice alone. In conjunction with the positive performance trend for the FF group, these findings indicate that FF facilitates strategy development and may lead to higher levels of Dynamic Decision Making (DDM) performance over time.

Download

A classification tree analysis of broadband adoption in Australian households

Proceedings of the 6th international conference on Electronic commerce - ICEC '04, 2004

Abstract Broadband communication technologies offer households many opportunities, including grea... more

Bootstrap and Jackknife, Overview

Wiley StatsRef: Statistics Reference Online, 2014

The Soil Moisture Active Passive (SMAP) Mission

Proceedings of the IEEE, 2010

The Soil Moisture Active Passive (SMAP) mission is one of the first Earth observation satellites ... more The Soil Moisture Active Passive (SMAP) mission is one of the first Earth observation satellites being developed by NASA in response to the National Research Council's Decadal Survey. SMAP will make global measurements of the soil moisture present at the Earth's land surface and will distinguish frozen from thawed land surfaces. Direct observations of soil moisture and freeze/thaw state from space will allow significantly improved estimates of water, energy, and carbon transfers between the land and the atmosphere. The accuracy of numerical models of the atmosphere used in weather prediction and climate projections are critically dependent on the correct characterization of these transfers. Soil moisture measurements are also directly applicable to flood assessment and drought monitoring. SMAP observations can help monitor these natural hazards, resulting in potentially great economic and social benefits. SMAP observations of soil moisture and freeze/thaw timing will also reduce a major uncertainty in quantifying the global carbon balance by helping to resolve an apparent missing carbon sink on land over the boreal latitudes. The SMAP mission concept will utilize L-band radar and radiometer instruments sharing a rotating 6-m mesh reflector antenna to provide highresolution and high-accuracy global maps of soil moisture and freeze/thaw state every two to three days. In addition, the SMAP project will use these observations with advanced modeling and data assimilation to provide deeper root-zone soil moisture and net ecosystem exchange of carbon. SMAP is scheduled for launch in the 2014-2015 time frame.

Download

The transformational dimension in the realization of business value from information technology

The Journal of Strategic Information Systems, 2006

Econometric studies have highlighted factors that appear to explain the differential effects of i... more

Communicating Effectively with Words, Numbers and Pictures: Drawing on Experience

ABSTRACT. A key requirement of all actuarial practice areasis the ability of the actuary to analyze

Download

Robust BCa–JaB method as a diagnostic tool for linear regression models

Journal of Applied Statistics, 2014

Download

Michael Martin

Uploads

Papers by Michael Martin

Log In