David Darais: Data Privacy by Programming Language Design
May
15
2019
David Darais: Data Privacy by Programming Language Design

TECH TALK IMFD: Data Privacy by Programming Language Design

ABSTRACT: Data privacy is a growing concern for every individual, businesses, government, and organization. For example, some companies actively sell private information to third parties without customer consent, provoking increasing privacy concerns. At the same time, personal data can produce positive effects in society: companies perform many useful services for customers based on private data, government entities use personal information for the public good, and medical researchers use patient data to perform important research.

The state of the art in privacy protection for individuals is differential privacy, which enables the statistical analysis of data with a mathematical guarantee of privacy. Successful differential privacy approaches have been developed for aggregate statistics, database queries, and convex machine learning.

In order to achieve differential privacy, random noise is (typically) introduced during the manipulation of data, which results in less accurate results. In some cases, random noise is not enough, and more aggressive techniques must be used such as data clipping. An ongoing challenge in differentially-private algorithm design is to achieve a balance between privacy guarantees and accuracy of results.

We present Duet: a general-purpose programming language for enforcing differential privacy. Duet consists of two mutually embedded programming languages and uses a multi-tiered analysis to automatically provide state-of-the-art bounds on privacy leakage for any program written in the language. In case studies, we show the effectiveness of Duet through the implementation of differentially private convex machine learning algorithms, and an empirical analysis of the accuracy of trained machine learning models vs the non-private model. In future work we aim to achieve differentially private training of non-convex machine learning (e.g., neural networks) with high accuracy—an unsolved challenge in provably-private algorithm design.

BIO: David Darais is an Assistant Professor at the University of Vermont. David’s research focuses on tools for achieving reliable software in critical, security-sensitive, and privacy-sensitive systems. David received his BS, MS, and Ph.D. from the University of Utah, Harvard University and the University of Maryland. http://david.darais.com/

WHERE: Ramón Picarte Auditorium, Department of Computer Sciences, the University of Chile. Beauchef 851, North Building, Third Floor, Santiago.

DATE/TIME: May, Wednesday 15th, at 12.00.

FOR MORE INFORMATION: comunicaciones-imfd@imfd.cl

Prof. Rajeev Raman: In-Memory Data Mining via Succinct Data Structures
Apr
29
2019
Prof. Rajeev Raman: In-Memory Data Mining via Succinct Data Structures

TECH TALK: “In-Memory Data Mining via Succinct Data Structures”

ABSTRACT: “We often use “Big Data” techniques when most users only have “big data”. “Big data” can often be handled efficiently by applying standard algorithms developed, tried and tested but coupled with succinct data structures to reduce the memory usage of such algorithms, thus allowing the “big data” to be processed in memory. I will introduce some recent applications to mining “big data”.

BIO:

Rajeed Raman is Professor of Computer Science, at the Department of Informatics, in the University of Leicester, United Kingdom. His research interests are broadly in algorithms and complexity and mostly lie in the area of data structure design. In recent years he has worked in succinct data structures. SDS represent user-provided data using an amount of computer memory close to the information-theoretic minimum and support very rapid queries and updates on the data. SDS have been shown to have a very good theoretical and practical performance for a growing range of applications including text search and XML processing. In addition to studying algorithms from a mathematical viewpoint, Prof, Raman is actively involved in algorithm engineering, including the implementation, experimental testing, and fine-tuning of discrete algorithms; the development of software repositories and platforms which allow the use of, and experimentation with, efficient discrete algorithms; methodological issues including standards in the context of empirical research on algorithms and data structures; methodological issues regarding the process of converting user requirements into efficient algorithmic solutions and implementations. Prof. Raman is also interested in data mining and he is a member of an active working group on Knowledge Discovery and Machine Learning. His interests are mining of uncertain data and applications of succinct data structures to data mining.

WHERE: Auditorio Ramón Picarte, DCC U. de Chile, Beauchef 851, Edificio Poniente, Tercer Piso. Santiago.

DATE/TIME: Lunes 29 de abril de 2019, de 11.00 a 12.00 horas.

MORE INFORMATION: comunicaciones-imfd@imfd.cl

Evelyne Huber & John Stephens: Patterns of Inequality in Post-Industrial Societies and Implications for Latin America
Apr
29
2019
Evelyne Huber & John Stephens: Patterns of Inequality in Post-Industrial Societies and Implications for Latin America

OPEN SEMINAR IMFD: Patterns of Inequality in Post-Industrial Societies and Implications for Latin America.

WHERE: Auditorio de Letras, Facultad de Letras. Campus San Joaquín, Universidad Católica (Vicuña Mackenna 4860, Macul, Santiago).

DATE/TIME: April, Monday 29th, 18.00 (06 PM).

Evelyne Huber: Morehead Alumni Professor of Political Science at the University of North Carolina, Chapel Hill. She studied at the University of Zurich and received both her M.A. (1973) and Ph.D. (1977) from Yale University. She is the author of The Politics of Workers’ Participation: The Peruvian Approach in Comparative Perspective (1980); co-author of Democratic Socialism in Jamaica (with John D. Stephens, 1986); co-author of Capitalist Development and Democracy (with Dietrich Rueschemeyer and John D. Stephens, 1992); co-author of Development and Crisis of the Welfare State (with John D. Stephens, 2001); co-author of Democracy and the Left: Social Policy and Inequality in Latin America (with John. D. Stephens, 2012); co-winner of the Outstanding Book Award 1991-92 from the ASA Political Sociology Section, winner of the Best Book Award 2001 from the APSA Political Economy Section, and winner of the Outstanding Book Awards 2013 from the ASA Sociology of Development Section and the Political Economy of the World System Section. She has also contributed articles to, among others, World Politics, Latin American Research Review, Comparative Politics, Politics and Society, Comparative Political Studies, The Journal of Politics, Studies in Comparative International Development, Comparative Social Research, Political Power and Social Theory, American Journal of Sociology, American Sociological Review, and Economic Perspectives. She received an Honorary Doctorate in the Social Sciences from the University of Bern in 2010, a Guggenheim Fellowship in 2010, and the Distinguished Teaching Award for Post-Baccalaureate Instruction from the University of North Carolina in 2004. She is a former President (2012-13) of the Latin American Studies Association.

 

John D. Stephens: Gerhard E. Lenski, Jr., Distinguished Professor of Political Science and Sociology; Director, Center for European Studies, European Union Center of Excellence, and Trans-Atlantic Masters program; received his B.A. (1970) from Harvard University and his Ph.D. (1976) from Yale University. His main interests are comparative politics and political economy, with area foci on Europe, the Antipodes, Latin America, and the Caribbean. He teaches European politics and the political economy of advanced industrial societies. He is the author of The Transition from Capitalism to Socialism (1979) and co-author of Democratic Socialism in Jamaica (with Evelyne Huber, 1986), Capitalist Development and Democracy (with Evelyne Huber and Dietrich Rueschemeyer, 1992; Outstanding Book Award, Political Sociology Section, ASA), Development and Crisis of the Welfare State (with Evelyne Huber, 2001; Best Book Award, Political Economy Section, APSA), and Democracy and the Left: Social Policy and Inequality in Latin America (with Evelyne Huber, 2012, Outstanding Book Award, Sociology of Development Section, ASA; Best Book Award, Political Economy of the World System Section, ASA). He has also contributed articles to, among others, The American Political Science Review, American Journal of Sociology, American Sociological Review, The British Journal of Sociology, Comparative Politics, Comparative Political Studies, Journal of European Social Policy, and World Politics. He is currently working on a study of social investment policy in Europe and Latin America.

1st Data, Beer & Pizza IMFD, with presentations made by Jorge Pérez, Magdalena Saldaña and Marcelo Arenas
Apr
16
2019
1st Data, Beer & Pizza IMFD, with presentations made by Jorge Pérez, Magdalena Saldaña and Marcelo Arenas

How can computers understand today the natural language? What can unleash a troll in the social networks and why does incivility run through these platforms? Are computers almighty or do they have a limit on their capabilities?

Magdalena Saldaña, associate researcher of the Millennium Institute Foundational Research on Data (IMFD) and academic of the Faculty of Communications of the P. Universidad Católica; Jorge Pérez, associate researcher at IMFD and academic at the Department of Computer Science of the University of Chile, and Marcelo Arenas, director of IMFD and full professor of the Department of Computer Science of the P. Universidad Católica, will give three entertaining talks on these subjects, in a relaxed atmosphere in which the assistants will be able to converse with the researchers.

Data, Beer & Pizza is an IMFD entry-free event. The availability of seats will depend on attendance since the place has a capacity for just over 100 people. Attendees should only pay for their consumption.

WHEN: Tuesday, April 16 at 8:00 p.m.

WHERE: Azotea Mackenna (Mackenna Gallery, Vicuña Mackenna 38, Providencia, Santiago).

INFORMATION: comunicaciones-imfd@imfd.cl

Patricio del Sol: Data made in Chile and exported to 18 countries, the experience of Admetricks
Apr
12
2019
Patricio del Sol: Data made in Chile and exported to 18 countries, the experience of Admetricks

Admetricks is a Chilean company dedicated to marketing intelligence for internet advertising. His CTO and founder, Patricio del Sol, will describe in this talk how does a company work when it’s 100% based on data. It will explain the infrastructure used, how the clients achieve success by using data, how they operate and process information, the architectures and patterns of technological design, and -finally- the current problems and challenges.

SPEAKER: Patricio del Sol, Computer Science Engineer, Pontificia Universidad Católica. CTO and founder of Admetricks.

WHEN/WHERE: April, Friday 12th, from 12.00 to 13.00. Auditorio San Agustín, San Agustín Building, Campus San Joaquín, P. Universidad Católica (Vicuña Mackenna 4860, Metro San Joaquín, Santiago).

Register your attendance here: comunicaciones-imfd@imfd.cl

IMFD Talk at the University of Concepción: Interested in doing research on science of data?
Apr
12
2019
IMFD Talk at the University of Concepción: Interested in doing research on science of data?

The Millennium Institute Foundational Research on Data will offer a presentation of its research to all the students and academics of the Universidad de Concepción who may be interested in the area of science of data with a multidisciplinary approach.

The IMFD brings together professors and students from Computer Science, Journalism, Political Science, Statistics, Architecture and Design, among many other areas. All of them work with a multidisciplinary approach in the study of problems and phenomena related to data, covering the entire cycle: from the origin of data as a complex digital unit, up to its use and social impact.

This talk is aimed at all those interested in, potentially, joining the IMFD team.

DATE: Friday, April 12th, 2019, from 12:00 to 13:00.

WHERE: Auditorium 105, Faculty of Engineering, University of Concepción. Edmundo Larenas 219, Concepción.

PROGRAMME:

Welcome Coffee

Presentation:

-Pablo Barceló, Alternate Director IMFD, full professor, Dept. of Computer Sciences, U. de Chile.

-Andrea Rodríguez, Vice-Rector for Research and Development UdeC, associate researcher IMFD.

-Sergio Toro, DemoData UdeC director, Professor of Political Science and Public Administration UdeC, researcher IMFD.

-Diego Seco, Dept. director . of Computer Engineering and Cs. of Computation UdeC, researcher IMFD.

-Questions

Bárbara Poblete in the Nerd Nites. “Do not believe everything you read: social networks under scrutiny”.
Apr
09
2019
Bárbara Poblete in the Nerd Nites. “Do not believe everything you read: social networks under scrutiny”.

Bárbara Poblete, an academic from the Department of Computer Science at the Universidad de Chile and an associate researcher at the Millennium Institute Foundational Research on Data is part of the opening session of the 2019 version of the successful Nerd Nites. In her talk “Do not believe everything you read: social networks under scrutiny”, Bárbara will address from a computational point of view the challenges of taking advantage of the benefits of social networks for the society, and -at the same time- the urgent need to create tools that enables the user to become into a critical consumer of this information.

WHEN: Tuesday, April 9th, 20.45 pm.

WHERE: Teatro IF (Avda. Italia esquina Francisco Bilbao, Providencia, Santiago).

More information:

http://bit.ly/NN9abril

www.facebook.com/NerdNitesSCL/

Radu Grosu, TU Wien. Towards Explainable RNNs: Modeling, Learning and Verification
Apr
04
2019
Radu Grosu, TU Wien. Towards Explainable RNNs: Modeling, Learning and Verification

ABSTRACT: We introduce a new type of recurrent neural networks which we call WormNets, as they were inspired by a biophysical model for neurons and synapses in the C. Elegans worm. WormNets are interpretable, smaller in size, and more robust to noise attacks when compared to classic RNNs. They can also take advantage of the rich trove of neural policies developed by nature through billions of years of evolution. We show how to model with WormNets and learn their parameters, or even learn the WormNets from scratch, without considerable penalty, by using state-of-the-art RNN learning techniques. We also discuss how to verify WormNets.

BIO: Radu Grosu is a full Professor and the Head of the Cyber-Physical Systems Group within the Institute of Computer-Engineering of the Vienna University of Technology. Grosu is also a Research Professor at the Department of Computer Science, of the State University of New York at Stony Brook, USA.

The research interests of Radu Grosu include modeling, analysis and control of cyber-physical systems and of biological systems. The applications focus of Radu Grosu includes smart-mobility, Industry 4.0, smart-buildings, smart-agriculture, smart-health-care, smart-cities, IoT, cardiac and neural networks, and genetic regulatory networks.

Radu Grosu is the recipient of the National Science Foundation Career Award, the State University of New York Research Foundation Promising Inventor Award, the Association for Computing Machinery Service Award, and is an elected member of the International Federation for Information Processing, Working Group 2.2.

Before receiving his appointment at the Vienna University of Technology, Radu Grosu was an Associate Professor in the Department of Computer Science, of the State University of New York at Stony Brook, where he co-directed the Concurrent-Systems Laboratory and co-founded the Systems-Biology Laboratory.

Radu Grosu earned his doctorate (Dr.rer.nat.) in Computer Science from the Faculty of Informatics of the Technical University München, Germany. He was subsequently a Research Associate in the Department of Computer and Information Science, of the University of Pennsylvania, an Assistant, and an Associate Professor in the Department of Computer Science, of the State University of New York at Stony Brook, USA.

WHERE: Ramón Picarte Auditorium, Department of Computer Sciences, the Universidad de Chile. Beauchef 851, North Building, 3rd Floor, Santiago.

WHEN: Thursday, April 4th, from 3PM to 4PM.

Information: comunicaciones-imfd@imfd.cl

Travis Gagie, UDP: Indexing Genomic Databases
Mar
15
2019
Travis Gagie, UDP: Indexing Genomic Databases
ABSTRACT: Indexing Genomic Databases: Since the first human genomes were sequenced and assembled de novo nearly twenty years ago, hundreds of thousands of others have been sequenced and assembled by variation calling.  That much data is a valuable resource but also a problem for algorithms and data structures designed to handle megabytes or gigabytes but not terabytes or petabytes.  In this talk we take variation calling itself as an example and consider why it should become easier as genomic databases grow, why it has not so far, and why it will soon.  Specifically, we describe a new and scalable version of the FM-index data structure underlying modern DNA aligners.
BIO: Travis Gagie is currently an associate professor at the Universidad Diego Portales and a researcher at the Chilean Center for Biotechnology and Bioengineering, specializing in compressed data structures for bioinformatics.  He received a BSc in Cognitive Science from Queen’s University, an MSc in Computer Science from the University of Toronto and a Dr. Rer. Nat. in Bioinformatics from Bielefeld University, Germany. Between his masters and doctorate, he studied for a year at the National Research Center in Pisa and worked for two years at the University of Eastern Piedmont.  After graduating he worked as a post-doctoral researcher at the University of Chile, Aalto University and the University of Helsinki.  He has published over a hundred conference and journal papers, served on the committees of about two dozen conferences and workshops and recently co-chaired the 2018 International Symposium on String Processing and Information Retrieval (SPIRE).
WHEN: Friday, March 15th, 2019.
TIME: 12.00 – 13.00
WHERE: Philippe Flajolet Auditorium, Department of Computer Sciences, Universidad de Chile (Beauchef 851, West Building, 3rd Floor)
Bárbara Poblete en Congreso Futuro 2019: “Hiperconectados, pero vulnerables”
Jan
18
2019
Bárbara Poblete en Congreso Futuro 2019: “Hiperconectados, pero vulnerables”

Panel “Solos e hiperconectados, la paradoja de nuestros tiempos”, Congreso Futuro 2019:

En una época en que parte importante de las comunicaciones entre las personas ocurre en plataformas digitales, el principal riesgo para la sociedad es considerar que son una fuente absoluta de información. Lo que se publica en redes sociales no siempre es editado o verificado, lo que lleva a los usuarios -sin saberlo- a creer y propagar información que no es correcta o que refleja una visión parcial de la realidad. Por tanto, al estar hiperconectados -señala la profesora Bárbara Poblete, académica de la U. de Chile e investigadora del Instituto Milenio Fundamentos de los Datos- lo que ocurre es que somos más vulnerables al efecto amplificador de las redes sociales: la velocidad de diseminación de datos falsos o sesgados y la burbuja que se genera al relacionarnos digitalmente con personas que comparten valores similares.

“Las personas interactúan en redes sociales con quienes piensan como ellas, generando una burbuja que termina por hacer que su mirada de mundo sea más estrecha, polarizada, hasta radicalizada”, señala la panelista de Congreso Futuro 2019. La investigadora, especialista en estos temas, señala que para ella el verdadero desafío a futuro consiste en desarrollar y fortalecer el pensamiento crítico que permita a las personas cuestionar el contenido al que son expuestos y buscar fuentes adicionales de información.

FECHA: Viernes 18 de enero de 2019.

HORA: 15.00 horas.

LUGAR: Catedral 1158, Santiago

MÁS INFORMACIÓN: https://www.congresofuturo.cl/proximos-eventos/cf-santiago-viernes-18-de-enero-2019

Marcelo Arenas: “Are computers allmighty?”
Jan
17
2019
Marcelo Arenas: “Are computers allmighty?”

En el evento resumen de las Noches Nerd 2018, los organizadores prepararon una selección con las mejores exposiciones, presentadas por expertos de las más diversas áreas. Marcelo Arenas, profesor titular del Departamento de Ciencia de la Computación de la Universidad Católica y Director del Instituto Milenio Fundamentos de los Datos será uno de los seis conferencistas.

FECHA: Jueves 17 de Enero
HORA: Desde las 19:30.
LUGAR: Teatro IF, Avenida Francisco Bilbao 465, esquina Av. Italia, Providencia, Santiago
Valor: $15.000 general / $7.500 estudiantes
Más información: www.lacasadegoethe.com

Juan F. Sequeda: Integrating Semantic Web in the Real World, A Journey Between Two Cities
Jan
11
2019
Juan F. Sequeda: Integrating Semantic Web in the Real World, A Journey Between Two Cities

Abstract: An early vision in Computer Science has been to create intelligent systems capable of reasoning on large amounts of data. Today, this vision can be delivered by integrating Relational Databases with the Semantic Web using the W3C standards: a graph data model (RDF), ontology language (OWL), mapping language (R2RML) and query language (SPARQL). The research community has successfully been showing how intelligent systems can be created with Semantic Web technologies, dubbed now as Knowledge Graphs. However, where is the mainstream industry adoption? What are the barriers to adoption? Are these engineering and social barriers or are they open scientific problems that need to be addressed? This talk will chronicle our journey of deploying Semantic Web technologies with real world users to address Business Intelligence and Data Integration needs, describe technical and social obstacles that are present in large organizations, and scientific and engineering challenges that require attention.

Bio: Juan Sequeda’s goal as a scientist is to create knowledge from inscrutable data reliably. Juan’s research interests are on the intersection of Logic and Data for (ontology-based) data integration and semantic/graph data management. Sequeda has received the NSF Graduate Research Fellowship, 2nd Place in the 2013 Semantic Web Challenge for his work on ConstituteProject.org, Best Student Research Paper at the 2014 International Semantic Web Conference and the 2015 Best Transfer and Innovation Project by the Institute for Applied Informatics. He is on the Editorial Board of the Journal of Web Semantics, member of multiple program committees (ISWC, ESWC, WWW, AAAI, IJCAI). He was the General Chair of AMW2018, PC chair of ISWC 2017 In-Use track, co-creator of COLD workshop (7 years co-located at ISWC). As an entrepreneur, Juan is a product manager, does business development and strategy, and works with customers to understand their problems. He has served as a bridge between academia and industry as the current chair of the Property Graph Schema Working Group, member of the Graph Query Languages task force of the Linked Data Benchmark Council (LDBC) and past invited expert member and standards editor at the World Wide Web Consortium (W3C).

Research Interest: Knowledge Graphs, Semantic Web, Databases, (Ontology-Based) Data Integration, Semantic and Graph Data Management.

Date: 11/01/2019, 12.00 to 13.00.

Where: Auditorium Javier Pinto, Department of Computer Science, Universidad Católica. (Edificio San Agustín, 4to piso, Vicuña Mackenna 4860, Santiago).

More information: fundamentos@imfd.cl

Martin Hilbert: Ciencia Social Computacional: entendiendo y prediciendo el comportamiento humano al nivel de una ciencia
Jan
08
2019
Martin Hilbert: Ciencia Social Computacional: entendiendo y prediciendo el comportamiento humano al nivel de una ciencia

Abstract: Hasta hace unos años, en ciencias sociales, cuando un estudio lograba explicar de 10% a 20% de la varianza de un fenómeno, sus resultados eran publicados en los más prestigiosos journals e influían en el desarrollo de políticas. Hoy sabemos que muchos de ellos fracasaron. En paralelo, una proporción cada vez mayor de interacciones humanas comenzó a llevarse a cabo en internet, produciendo una huella digital que crece de manera masiva y cuyo estudio puede generar conocimientos sin precedentes sobre la sociedad, su funcionamiento y sus intrincadas redes, incluso aquellas que habían permanecido ocultas hasta ahora. Y es que la tecnología no solo ha revolucionado a la sociedad, sino también la forma en que podemos entenderla. La inteligencia artificial permite detectar patrones ocultos con herramientas analíticas, como el aprendizaje automático y el procesamiento de lenguaje natural. Las simulaciones computacionales nos ayudan a explorar y explicar las más variadas situaciones hipotéticas. Gracias a estos avances, en los últimos años hemos empezado a predecir el comportamiento humano y social con 80% y 90% de precisión. Los estudios sociales se están convirtiendo en una ciencia. ¿Cuáles pueden ser las consecuencias?.

Bio: Martin Hilbert es profesor en la Universidad de California, Davis (EE.UU.) Su investigación multidisciplinaria aborda el rol de la información y el conocimiento en el desarrollo de sistemas sociales complejos. Doctor en Ciencias Económicas y Sociales (2006) y en Comunicación (2012). Creador y coordinador del Programa Sociedad de la Información de CEPAL. En sus 15 años de trabajo como Oficial de Asuntos Económicos de la ONU, fue asesor técnico para el desarrollo digital de más de 20 países. Su trabajo ha sido publicado en los journals más reconocidos, como Science, Psychological Bulletin, Trends in Ecology and Evolution y World Development. Aparece regularmente en medios como The Wall Street Journal, Washington Post, The Economist, NPR, BBC y Die Welt, entre otros. Más información en www.martinhilbert.net.

Fecha y hora: Martes 8 de enero de 2019, 15.00 horas.

Lugar: Auditorio Ramón Picarte, Dpto. de Ciencias de la Computación, Universidad de Chile (Beauchef 851, Edificio Norte, Tercer Piso, Santiago).

Consultas y registro: fundamentos@imfd.cl

George Vega-Yon: Workshop on data analysis using R. Session 2
Jan
07
2019
George Vega-Yon: Workshop on data analysis using R. Session 2

On Monday, January 7th, 2019, from 10 am to 1 pm, George Vega-Yon will host the second session of the Workshop on Data Analysis in R. Vega-Yon is a Ph.D. student in Biostatistics, at the University of Southern California (USC) in the USA; Master of Science, Caltech (USA); Master in Economics and Public Policy, Adolfo Ibáñez University, Chile.

His research focuses on Computational Statistics applied to the modeling of biological and socio-technical systems, such as the evolution of genetic functions and RR.SS.

The workshop is open to all interested parties, but it is recommended to have intermediate knowledge of R to get the most out of the content. It will be held in Classroom C201, Building C (Civil Construction) of the San Joaquín Campus, Pontificia Universidad Católica (Vicuna Mackenna 4860, Macul, Metro San Joaquín).

Contents of Tutorial 2:

Although R was not designed with High-Performance Computing (HPC), thanks to the fruitful community of R users there are several ways in which it can be used to solve problems using HPC.

This workshop will provide a general overview of what can be done in the area with R, giving emphasis to parallel computing applications, as well as presenting some of the available tools.

The session will be developed as a mixture of presentation and work blocks. Attendees are expected to write R programs to answer questions that will be presented throughout the workshop.

Registration: comunicaciones-imfd@imfd.cl

George Vega Yon: Data Analysis in R, Part One.
Dec
20
2018
George Vega Yon: Data Analysis in R, Part One.

Data Analysis in R, Part One, by George Vega Yon. This session will be divided in three sessions and will include theory and application of social networks statistical models in R and spatial econometrics:

1st Block, Motivation:
(a) When data isn’t IID.
(b) Models families: Graphos vs. behavior.

2nd Block, Network models:
(a) ERGMs and variations: MERGMs, TERGMs, SERGMs, Relational Event Models, ERGMitos, etc.
(b) Latent networks.
(c) SOAM and others.
(d) Network models estimate: Estimation diagnostics, convergence.
(e) Non parametrics models: matching and permutation.

3rd Block, Spatial Econometrics:
(a) Spatial autoregresive models: description, assumptions, etc.
(b) Family of models: Heteroscedasticity, autocorrelated mistakes, IV sol and estimate methods, SARAR, SAR Probit/Logit/MLogit

Bio: George G. Vega Yon, PhD student in Biostatistics in the University of South California, USA. Master in Sciences, California Institute of Technology (Caltech, USA), Master in Economics and Public Policies, Adolfo Ibáñez University, Chile. His research interests are computational statistics applied to the modelling of biological and socio-technical systems, as evolution from genetical functions and social networks.

Date: Thursday, December 20th, from 10.00 to 13.00.

Place: Room C201, Building Edificio, Campus San Joaquín, Pontificia Universidad Católica (Vicuña Mackenna 4860, Macul. Metro San Joaquín, Santiago).

José Emilio Labra: Validating RDF data: ShEx and SHACL compared
Dec
17
2018
José Emilio Labra: Validating RDF data: ShEx and SHACL compared

Although the benefits of RDF for data representation and integration are indisputable, it has not been embraced by everyday programmers and software architects who care about safely creating and accessing well-structured data. Semantic web projects still lack some common tools and methodologies that are available in more conventional settings to describe and validate their data.
Two technologies have recently been proposed for RDF validation: Shape Expressions (ShEx) and Shapes Constraint Language (SHACL). ShEx was designed as an intuitive and human-friendly high level language for RDF validation in 2014 and is being actively developed by the W3C ShEx community group. SHACL, proposed by the Data Shapes Working Group, was accepted as a W3C Recommendation in July 2017.
In the talk, we will provide short introductions to both ShEx and SHACL using examples, provide a comparison between both and discuss some ideas for future work and applications related with RDF validation.

Bio: PhD. Jose Emilio Labra Gayo, Associate Professor at the University of Oviedo, Spain. He is the main researcher of the WESO research group and was a member of the RDF Data Shapes working group. He is co-author of the “Validating RDF data” book (http://book.validatingrdf.com), implemented the Shaclex library which supports both ShEx and SHACL, and maintains the online RDF validation service RDFShape (http://rdfshape.weso.es).

Date/Venue: December 17, 2018 at 2pm. Philipe Flajolet Auditorium (Beauchef 851, West Building, 3rd. Floor, Santiago).

Marcelo Arenas y Juan Pablo Luna: Datos, desde sus fundamentos a su impacto social
Dec
11
2018
Marcelo Arenas y Juan Pablo Luna: Datos, desde sus fundamentos a su impacto social

Abstract: La gestión de los gobiernos o la comprensión de complejos fenómenos sociales son sólo algunos ejemplos de áreas en las que los datos se están convirtiendo en un recurso clave. Sin embargo, existen desafíos que deben ser abordados con urgencia si se espera hacer un uso eficiente, de calidad y confiable de esa información. El Instituto Milenio Fundamentos de los Datos (IMFD) abordará en esta charla las investigaciones que está llevando a cabo para entender y atender estos problemas.

Integración eficiente de información
¿Es posible integrar de manera eficiente datos en formatos distintos, alojados en diversos lugares, y originados por variadas fuentes? Y una vez integrados, ¿es posible depurarlos a través de una curatoría científicamente aplicada, para poder extraer de ellos conocimiento? El IMFD estudia sistemas que permitan resolver este desafío, con foco especial en la información pública.

Datos, ciencia y sociedad
Los instrumentos utilizados hoy para estudiar tendencias de la sociedad han mostrado no ser capaces de representarlas a cabalidad. Encuestas y proyecciones presentan sesgos dados por la cantidad acotada de datos que incluyen, el origen de los mismos y las formas de obtenerlos. En el IMFD estamos estudiando cómo ampliar el espectro de recolección de información, conectando diversas fuentes y, a través de una integración eficiente de información, contar con una mejor imagen de los fenómenos sociales.

Trazabilidad y transparencia
En un momento en que la confianza de los ciudadanos en las distintas instituciones del Estado ha ido cayendo, uno de los grandes objetivos es preservar la integridad y, por ende, la transparencia de los datos públicos. En el IMFD buscamos integrar a la gestión de datos y procesos públicos tecnologías para la acreditabilidad, trazabilidad y seguimiento de la información, como blockchain y algunos mecanismos de incentivos sobre esta estructura, de manera tal de garantizar la confiabilidad y transparencia de estos datos y procesos.

Expositores:

Marcelo Arenas es profesor titular del Departamento de Ciencia de la Computación de la Pontificia Universidad Católica y director del Instituto Milenio Fundamentos de los Datos. Es Doctor en Ciencia de la Computación de la Universidad de Toronto (Canadá). Se especializa en sistemas de manejo y uso de datos, y la comprensión y búsqueda de mejores algoritmos computacionales. Su investigación en el campo de web semántica ha tenido impacto importante a nivel mundial.

Juan Pablo Luna es profesor titular del Instituto de Ciencia Política de la Pontificia Universidad Católica e investigador asociado del Instituto Milenio Fundamentos de los Datos. Es Doctor en Ciencia Política de la Universidad de Carolina del Norte en Chapel Hill, Estados Unidos. Es autor del libro “Segmented representation: political party strategies in unequal democracies” (Oxford University Press 2014) y de “En vez del optimismo: Crisis de representación política en el Chile actual” (Catalonia, CIPER 2017).

Información: 11/12/2019, 11.00 horas. Sala Celso Furtado, CEPAL (Daj Hammarskjold s/n, Vitacura).

Espacio Público: Investigación orientada a la incidencia en políticas públicas: La experiencia del ThinkTank Espacio Público
Dec
07
2018
Espacio Público: Investigación orientada a la incidencia en políticas públicas: La experiencia del ThinkTank Espacio Público

Abstract: A partir de su experiencia en la elaboración e implementación de la agenda anticorrupción en la Comisión Engel, Daniel García (Director Ejecutivo) y Miguel Jorquera (Investigador), expondrán sobre estrategias para que los trabajos sobre investigación académica logren influir en la opinión pública, con el objetivo de convertirse en políticas que contribuyan a mejorarla calidad de vida de las personas.

Fecha y lugar: Viernes 7 de diciembre de 2018, 12.00 horas. Sala Javier Pinto, Edificio San Agustín, 4to Piso, Campus San Joaquín, Universidad Católica (Vicuña Mackenna 4860, Macul, Santiago).

Muhammad Imran: AI and Social Media for Disaster Response and Associated Challenges
Dec
06
2018
Muhammad Imran: AI and Social Media for Disaster Response and Associated Challenges

Fecha: Jueves 6 de diciembre a las 10:30am.
Lugar: P303 – Auditorio Philippe Flajolet, Beauchef 851

AI and Social Media for Disaster Response and Associated Challenges

Abstract: Sudden-onset emergencies such as natural or man-made disasters bring uncertainties in which time-critical information needs emerge from formal response organizations, affected communities and other concerned population. The growing adaption of Information and Communication Technologies (ICT) and Social Networks such as Twitter, Facebook has created numerous opportunities to disseminate and consume critical information during an on-going situation. However, time-critical analysis of high-velocity social media streams containing high-volume data involves solving multiple challenges including realtime parsing of brief and informal messages, handling information overload issues, and classifying, summarizing, and prioritizing different types of information. In this talk, I will present our work on solving some of these challenges.

Bio: Muhammad Imran is a Research Scientist at the Qatar Computing Research Institute where he leads the Crisis Computing team. His interdisciplinary research focuses on natural language processing, text mining, human-computer interaction, and applied machine learning. Imran has published over 70 research papers in top-tier international conferences and journals including ACL, SIGIR, ICWSM, WWW, and ASONAM. Two of his papers received the Best Paper Award.

Dominik Tomaszuk: Cheminformatics meets the Linked Data and SKOS
Nov
23
2018
Dominik Tomaszuk: Cheminformatics meets the Linked Data and SKOS

Viernes 23/11/2018
Sala Philippe Flajolet, DCC UChile
12.00 horas.

Cheminformatics meets the Linked Data and SKOS

Abstract: Cheminformatics is evolving from being an area of study associated mainly with drug discovery into a discipline that holds the access, management, store, and sharing of chemical data. In this field, there is a lack of truly interoperable databases that would allow for information exchange between databases. We have decided to address this matter, therefore we propose a new approach to data storage and retrieval, based on the Simple Knowledge Organization System (SKOS) model that provides a way to organize and access knowledge. In this presentation, we present Chemical Vocabulary for Molecular Entities (CVME), which is a metaformat for describing molecules, and supports any existing chemical formats. CVME supports semantic interoperability, a feature that is lacking in other formats. Moreover, we introduce ChemSKOS database that supports our metaformat, and meets Linked Data and FAIR data principles.

Bio: Dr Dominik Tomaszuk is a researcher at the University of Bialystok, Faculty of Mathematics and Informatics (Institute of Informatics), Poland. Dominik holds an M.Sc. (2008) in Computer Science, from the Bialystok University of Technology, Poland. He also holds a Ph.D. (2014) in Computer Science from the Warsaw University of Technology, Poland. His current research focuses on Semantic Web, RDF, Property Graphs, NoSQL databases and cheminformatics.

XIV SEMINARIO INTERNACIONAL INTELIGENCIA ARTIFICIAL
Nov
22
2018
XIV SEMINARIO INTERNACIONAL INTELIGENCIA ARTIFICIAL

Fundación Copec-UC elige a la Inteligencia Artificial como tema de su Seminario Internacional 2018

Será la versión 14º de un evento que convoca a la academia, a la industria y a las instituciones gubernamentales en torno a un tema de vanguardia. Este año, la Inteligencia Artificial contemplada como la cuarta revolución industrial, se abordará desde la experiencia y el análisis de expertos internacionales de renombre mundial y también se conocerán casos chilenos de destacados investigadores nacionales.

Este año se han invitado a expositores internacionales quienes entregarán una visión ampliada de los desafíos y oportunidades de la IA en el mundo. El primero es el profesor Jitendra Malik, uno de los científicos informáticos más distinguidos del mundo que ha realizado importantes contribuciones en los campos de la visión artificial, el modelado computacional de la visión biológica, los gráficos por computadora y el aprendizaje automático. Actualmente, se desempeña como Profesor Arthur J. Chick en la División de Ciencias de la Computación, Departamento de Ingeniería Eléctrica y Ciencias de la Computación, UC Berkeley. Con más de 150 trabajos de investigación en el campo de la visión artificial, es uno de los científicos más citados en el mundo.

El segundo invitado es Subbarao Kambhampati, profesor de Ciencias de la Computación e Ingeniería en la Universidad Estatal de Arizona y Presidente de la Asociación para el Avance de la Inteligencia Artificial. Su investigación se centra en la planificación automatizada y la toma de decisiones, especialmente en el contexto de los sistemas de AI con conciencia humana. Él es un experto galardonado que ha realizado valiosos estudios sobre las percepciones públicas y los impactos sociales de la IA.

Además, expondrán sus casos de éxito chilenos los académicos Álvaro Soto y Bárbara Poblete, destacados profesores de la Universidad Católica y la Universidad de Chile respectivamente, e investigadores asociados del Instituto Milenio Fubdamentos de los Datos, quienes contarán lo que en nuestro país se está desarrollando en estas temáticas y qué nos queda por delante.

También se presentarán dos casos de éxito de empresas. La primera, de Huawei, a cargo del emprendedor Gabriel Gurovich y la segunda, de Microsoft expuesta por Wilson Pais, Director Nacional de Tecnología en Microsoft Chile.

El Ministro de Hacienda, Felipe Larraín participará analizando los desafíos de la IA para el Estado y el Presidente de la Sofofa, Bernardo Larraín, expondrá sobre las oportunidades y las amenazas de la IA para las empresas.

La jornada, que pretender ser muy inspiradora, con foco en los desafíos futuros de Chile y el mundo, se realizará el jueves 22 de noviembre entre las 8.30 y las 13 horas en el Salón Fresno del Centro de Extensión UC y los participantes deben inscribirse previamente en la página web de la Fundación Copec-UC, www.fcuc.cl

Más información en https://fcuc.cl/seminarios/inteligencia-artificial-seminario-internacional-2018/

JUEVES 22 DE NOVIEMBRE DE 08:30 A 13:00
SALÓN FRESNO – CENTRO DE EXTENSIÓN UC
ALAMEDA 390, SANTIAGO

Equipo mixto anota más goles: Conversatorio sobre interdisciplina
Nov
16
2018
Equipo mixto anota más goles: Conversatorio sobre interdisciplina

Viernes 16 de noviembre, 11.00 horas
Sala 202, Centro de Innovación
Campus San Joaquín, Pontificia Universidad Católica

Equipo mixto anota más goles

Conversatorio sobre el impacto de la interdisciplina en la investigación científica. Participan: Virginia Garretón, fundadora de la Corporación Capital Biodiversidad, Directora Ejecutiva de la Iniciativa Científica Milenio (2015-2018); y Daniela Thumala, académica del Departamento de Psicología de la Universidad de Chile, investigadora asociada del Centro de Investigación Gero Chile.

Moderan: Carla Alberti, académica del Instituto de Ciencia Política de la Universidad Católica e investigadora del IMFD; Pablo Barceló, profesor titular de la Universidad de Chile y director alterno del Instituto Milenio Fundamentos de los Datos.

Rossano Schifanella: “Combined Effect of Content Quality and Social Ties on User Engagement”
Nov
15
2018
Rossano Schifanella: “Combined Effect of Content Quality and Social Ties on User Engagement”

Abstract: The dynamics of attention in social media tend to obey power laws. Attention concentrates on a relatively small number of popular items neglecting the vast majority of content produced by the crowd. Although popularity can be an indication of the perceived value of an item within its community, previous research has highlighted the gap between success and intrinsic quality. As a result, high quality content that receives low attention remains invisible and relegated to the long tail of the popularity distribution. Moreover, the production and consumption of content is influenced by the underlying social network connecting users by means of friendship or follower-followee relations. This talk will present a large scale study on the complex intertwinement between quality, popularity and social ties in an online photo sharing platform, proposing a methodology to democratize exposure and foster long term users engagement.

Bio: Rossano Schifanella is an Assistant Professor in Computer Science at the University of Turin, Italy, where he is a member of the Applied Research on Computational Complex Systems group. He is a visiting scientist at Nokia Bell Labs and a former visiting scientist at Yahoo Labs and at the Center for Complex Networks and Systems Research at the Indiana University where he was applying computational methods to model the behavior of (groups of) individuals and their interactions on social media platforms. His research embraces the creative energy of a range of disciplines across data mining, network analysis, urban science, computational social science, and data visualization.

Date: November 15th, 2018, 10.00 h

Venue: Auditorium Philipe Flajolet, Dept. Computer Sciences, the University of Chile.
Escuela de Fundamentos de los Datos (JCC 2018)
Nov
09
2018
Escuela de Fundamentos de los Datos (JCC 2018)

Viernes 9 de Noviembre 2018
Antonio Varas 880, Providencia, UNAB

Sesión I – 9:00 – 10:30 hrs. – “Grafos y lenguajes de consulta”

9:00 – 9:30 hrs: Marcelo Arenas (Pontificia Universidad Católica de Chile), “G-CORE: Definiendo un lenguaje de consulta estándar para grafos”.

9:30 – 10:00 hrs: Juan Reutter (Pontificia Universidad Católica de Chile), “Grafos y datos en la web”

10:00 – 10:30 hrs: Diego Arroyuelo (Universidad Técnica Federico Santa María), “Estructuras de datos compactas y comprimidas para grafos”

Sesión II – 11:00 – 12:00 hrs. – “Inteligencia artificial y datos”

11:00 – 11:30 hrs.: Marcelo Mendoza (Universidad Técnica Federico Santa María), “Fake news, bots y cyborgs”

11:30 – 12:00 hrs. Denis Parra (Pontificia Universidad Católica de Chile), “Explicabilidad en sistemas de recomendación”

Sesión III – 15:00 – 16:30 hrs. – “Big data y redes sociales”

15:00 – 15:30 hrs.: Bárbara Poblete (Universidad de Chile), “Sensores sociales: ¿Para qué pueden servir y cómo utilizarlos?”

15:30 – 16:00 hrs.: Hans Löbel (Pontificia Universidad Católica de Chile), “Explaining subjective perceptions of public spaces as a function of the built environment”

16:00 – 16:30 hrs.: Isabelle Beaudry (Pontificia Universidad Católica de Chile), “Inferencia estadística para redes sociales”

Francisco Vial: “Adaptively Learning the Parallelepiped: A Key-Recovery Attack Against the First Fully Homomorphic Cryptosystem”
Nov
08
2018
Francisco Vial: “Adaptively Learning the Parallelepiped: A Key-Recovery Attack Against the First Fully Homomorphic Cryptosystem”

Abstract: In a groundbreaking series of articles, Craig Gentry proposed in 2009 the first fully homomorphic encryption scheme. In the first variation of the scheme, secret keys are bases of polynomial ideal lattices, which provide algebraic structures that can be exploited by an attacker. In this talk, we introduce the Adaptively Learning the Parallelepiped problem (ad-LP), and show its relation to the problem of extracting a secret key of Gentry’s scheme, given bounded access to a decryption oracle. We then describe a geometric algorithm that learns an approximation of the parallelepiped using only O(n log(n)^3) decryptions (where n is the dimension of the lattice), and a practical depth-first search version. We use an implementation to demonstrate the attack using a standard CPU with C++, GMP, and Sage, which extracted secret keys in dimension 334 (safe against lattice reduction techniques) with 87,000 decryption queries in about 15 minutes. We also discuss some countermeasures and extensions of the attack against other cryptographic schemes.

Date: November 8th, 2018. 10.00 h.

Venue: Sala de Consejo, Depto. de Ciencia de la Computación, P. Universidad Católica.

Marcelo Mendoza, Nicolás Torres: “Creating High Level Content Descriptors for Recommender Systems Datasets” (JCC 2018)
Nov
07
2018
Marcelo Mendoza, Nicolás Torres: “Creating High Level Content Descriptors for Recommender Systems Datasets” (JCC 2018)

ABSTRACT. Information Retrieval and Recommender Systems have been frequently evaluated using indexes based on variants and extensions of precision-like measures. Likewise, approaches for diversity evaluation have been proposed. However, these measures are usually defined in terms of a set of high level content descriptors known as \textit{information nuggets} that are hard to obtain. We propose a method to create these nuggets using social tags, providing datasets with annotations to evaluate content diversity in recommender systems. Since recommending items to a target user is analogous to searching documents from a query, this method might be extended to Information Retrieval.

Date: November 7th, 2018. 15.00 h.

Venue: Universidad Andrés Bello, Antonio Varas 880, Providencia, Santiago.

Marcelo Mendoza, Pablo Ormeño, Carlos Valle: “Ad-hoc information retrieval based on boosted latent Dirichlet allocated topics” (JCC 2018)
Nov
06
2018
Marcelo Mendoza, Pablo Ormeño, Carlos Valle: “Ad-hoc information retrieval based on boosted latent Dirichlet allocated topics” (JCC 2018)

ABSTRACT: Latent Dirichlet Allocation (LDA) is a fundamental method in the text mining field. We propose strategies for topic and model selection based on LDA that exploits the semantic coherence of the topics inferred, boosting the quality of the models found. Then we study how our boosted topic models perform in ad-hoc information retrieval tasks. Experimental results in four datasets show that our proposal improves the quality of the topics found favoring document retrieval tasks. Our method outperforms traditional LDA-based methods showing that model selection based on semantic coherence is useful for document modeling and information retrieval tasks.

Date: November 6th, 2018. 15.40 h.

Venue: Universidad Andrés Bello, Antonio Varas 880, Providencia, Santiago.

Sheila McIlraith: Artificial Intelligence: Past, present, and future (JCC 2018)
Nov
05
2018
Sheila McIlraith: Artificial Intelligence: Past, present, and future (JCC 2018)

Invited by IMFD, Sheila McIlraith will be present at the 2018 Chilean Computer Congress.

Sheila McIlraith is a Professor in the Department of Computer Science, University of Toronto. Prior to joining U of T in 2004, McIlraith spent six years as a Research Scientist at Stanford University, and one year at Xerox PARC. McIlraith is the author of over 100 scholarly publications in the area of knowledge representation and automated reasoning, and in particular automated plan generation and sequential decision making. She is currently serving as Past-President of KR Inc., the international scientific foundation concerned with fostering research and communication on knowledge representation and reasoning. McIlraith is a fellow of the Association for the Advancement of Artificial Intelligence (AAAI), associate editor of the Journal of Artificial Intelligence Research (JAIR), serves on the editorial board of Artificial Intelligence Magazine, and is a past associate editor of the journal Artificial Intelligence (AIJ). She was recently program co-chair of the 32nd AAAI Conference on Artificial Intelligence (AAAI-18), and is past program co-chair of the 13th International Conference on Principles of Knowledge Representation and Reasoning (KR2012), and the International Semantic Web Conference (ISWC2004). McIlraith’s early work on Semantic Web Services has had notable impact. In 2011 she and her co-authors were honoured with the SWSA 10-year Award, recognizing the highest impact paper from the International Semantic Web Conference, 10 years prior. Her research has also made practical contributions to the development of next-generation NASA space systems and to emerging Web standards.

Date: November 5th, 2018. 19.00-20.30 h.

Venue: Universidad Andrés Bello. Antonio Varas 880, Providencia, Santiago

Bárbara Poblete modera panel “Cómo reclutan los empleadores”, Chile WiC 2018.
Oct
26
2018
Bárbara Poblete modera panel “Cómo reclutan los empleadores”, Chile WiC 2018.

Promover el ingreso de mujeres a carreras ligadas a la informática, así como visibilizar su quehacer e investigaciones en el área, son algunos de los objetivos del VII Chile WIC “Encuentro Chileno de Mujeres en Computación”, actividad organizada por la Universidad Técnica Federico Santa María -a través de su Departamento de Informática- junto con las universidades de Chile, Pontificia Universidad Católica y Los Andes. El programa de Mujeres en Computación contempla los paneles: “Iniciativas de género para escolares”, “Cómo reclutan los empleadores: tips y consejos para tener éxito”, “Iniciativas de género: Mujeres en TI”; charlas profesionales que abordaron variadas temáticas como la prevención de femicidios, sistemas de big data para el análisis hídrico y datos abiertos en América Latina; la competencia estudiantil que incluye la presentación de posters de proyectos, entre otras actividades.

Loreto Bravo: “Shopping mall attraction and social mixing”
Oct
19
2018
Loreto Bravo: “Shopping mall attraction and social mixing”

Abstract: In Latin America, shopping malls seem to offer and open, safe and democratic version of the public space. However, it is often difficult to quantitatively measure whether they indeed foster, hinder or are neutral with respect to social inclusion. In this talk we present that, by using mobile phone network records, we can provide a socio-economic characterization of mall visitors and also show that some sectors do indeed modify their mobility patterns to go to malls, to go to malls that are farther away, thus fostering social inclusion.

Bio: Dr. Loreto Bravo is the director of the Data Science Institute, an alliance between the Faculty of Engineering of the Universidad del Desarrollo and Telefónica R&D Chile. Loreto Bravo is also professor in the Master Program in Data Science and works in the Digital Transformation Center of the same university. Engineering Degree from the Universidad Católica in Chile, PhD in Computer Science, University of Carleton (Canada). She has been researcher at the Database Group of the University of Edinburgh (Scotland) and professor at the Universidad de Concepción, Chile. Her current goal is promoting applied research through formal partnerships with the industry.

Date: October 19th, 2018. 12.00 h.

Venue: Auditorio San Agustín, Campus San Joaquín, Universidad Católica.

Andreas Wiese: “Approximation algorithms for the Geometric Knapsack problem”
Oct
12
2018
Andreas Wiese: “Approximation algorithms for the Geometric Knapsack problem”

Abstract: Many optimization problems are NP-hard and therefore we do not expect to find algorithms for them that are efficient, i.e., run in polynomial time, and find the optimal solution for any given instance. Therefore, we are interested in approximation algorithms which are algorithms that run in polynomial time and provably find solutions that differ from the optimum by at most some bounded factor, called the approximation ratio.

In this talk I will present approximation algorithms for the 2-dimensional knapsack problem. Given are a square knapsack and a set of items that are axis-parallel rectangles. Each item has a profit associated with it. The goal is to pack a subset of the given items non-overlappingly into the knapsack in order to maximize the total profit of the packed items. This problem generalizes the well-studied (one-dimensional) knapsack problem. I will present an algorithm with an approximation ratio of 1.89+eps and varios other results for the problem. Key to all results is to show that there are good solutions that have a relatively simple structure.

Bio: Andreas Wiese is an assistant professor at the Industrial Engineering department of the Universidad de Chile. He finished his PhD in 2011 at TU Berlin and was a postdoc at TU Berlin, La Sapienza in Rome and at the MPI for Informatics in Saarbruecken. In his research he focuses on combinatorial optimization and approximation algorithms, i.e., on algorithms that are efficient and compute solutions that are provably close to the optimum.

Fecha: 12 de octubre de 2018, 12.30 a 13.30 horas.

Lugar: Auditorio Phillippe Flajolet, Depto. Ciencias de la Computación, U. de Chile.

(Español) Juan Pablo Luna: “Universos Paralelos: La Segmentación de la Ciudadanía Civil, Política y Social en América Latina y Chile”
Sep
28
2018
(Español) Juan Pablo Luna: “Universos Paralelos: La Segmentación de la Ciudadanía Civil, Política y Social en América Latina y Chile”

Abstract: We assume that in a modern democracy people would have access to basic rights linked to civil, political and social citizenship. Without such access, democratic agency becomes problematic. Based on data from the LAPOP survey and the work with more than 100 thousand observations for the 2012-2016 period, Juan Pablo Luna and Rodrigo Medel analyzed the perceptions of Latin American citizens regarding their access to citizenship rights in the region. Based on this work, they carried out a comparative analysis of the situation of Chile in the regional context. The most relevant empirical finding accounts for a marked functional, territorial and socioeconomic segmentation of the perceptions of Chileans regarding their degree of perceived access o democratic citizenship.

In this talk, Juan Pablo Luna will present the results of this research and will discuss its implications for future research on the quality of democracy in Chile and the region.

Bio: Professor of the Institute of Political Science of the Pontifical Catholic University of Chile and Associate Researcher of the Millennium Institute for Foundational Research on Data. Doctor in Political Science from the University of North Carolina at Chapel Hill, United States. He is the author of “Segmented representation: political party strategies in unequal democracies” (Oxford University Press, 2014) and “Instead of optimism. Crisis of political representation in the current Chile “(Catalonia, CIPER, 2017). In 2014 he co-edited the book “The Resilience of the Latin American Right” (Johns Hopkins University Press). He is editor of Latin American Politics and Society (academic journal edited by Cambridge University Press). He has been a visiting professor at the universities of Columbia (2018), Brown (2016), Harvard (2013), Sciences-Po (Paris) and Princeton (2008). He is a regular columnist at CIPER-Chile.

Date: September 28th, 12.00 h.

Venue: Hall N5, Campus San Joaquín, Universidad Católica.

Themis Palpanas: “Data Series Management: Fulfilling the Need for Big Sequence Analytics”
Sep
14
2018
Themis Palpanas: “Data Series Management: Fulfilling the Need for Big Sequence Analytics”

Abstract: There is an increasingly pressing need, by several applications in diverse domains, for developing techniques able to index and mine very large collections of sequences, or data series. Examples of such applications come from social media analytics and internet service providers, as well as from a multitude of scientific domains. It is not unusual for these applications to involve numbers of data series in the order of hundreds of millions to billions, which are often times not analyzed in their full detail due to their sheer size. However, no existing data management solution (such as relational databases, column stores, array databases, and time series management systems) can offer native support for sequences and the corresponding operators necessary for complex analytics.

In this talk, we argue for the need to study the theory and foundations for sequence management of big data sequences, and to build corresponding systems that will enable scalable management and analysis of very large sequence collections. We describe recent efforts in designing techniques for indexing and mining truly massive collections of data series that will enable scientists to easily analyze their data. We discuss novel techniques that adaptively create data series indexes, allowing users to correctly answer queries before the indexing task is finished. Finally, we present our vision for the future in big sequence management research, including the promising directions in terms of storage, distributed processing, and query benchmarks.

Bio: Themis Palpanas is Senior Member of the Institut Universitaire de France (IUF), a distinction that recognizes excellence across all academic disciplines, and professor of computer science at the Paris Descartes University (France), where he is director of diNo, the data management group. He received the BS degree from the National Technical University of Athens, Greece, and the MSc and PhD degrees from the University of Toronto, Canada. He has previously held positions at the University of Trento, and at IBM T.J. Watson Research Center, and visited Microsoft Research, and the IBM Almaden Research Center.
His interests include problems related to data science (big data analytics and machine learning applications). He is the author of nine US patents, three of which have been implemented in world-leading commercial data management products. He is the recipient of three Best Paper awards, and the IBM Shared University Research (SUR) Award.
He is curently serving on the VLDB Endowment Board of Trustees, as an Editor in Chief for the BDR Journal, Associate Editor for VLDB 2019, Associate Editor in the TKDE, and IDA journals, as well as on the Editorial Advisory Board of the IS journal, and the Editorial Board of the TLDKS Journal. He has served as General Chair for VLDB 2013, Associate Editor for VLDB 2017, and Workshop Chair for EDBT 2016, ADBIS 2013, and ADBIS 2014, General Chair for the PDA@IOT International Workshop (in conjunction with VLDB 2014), and General Chair for the Event Processing Symposium 2009.

Date: September 14th, 2018, 12.00 h.

Venue: Auditorio San Agustín, Dept. of Computer Science, Universidad Católica.

Prof. César Hidalgo (MIT): “How do nations learn? The principles for the collective learning”
Sep
07
2018
Prof. César Hidalgo (MIT): “How do nations learn? The principles for the collective learning”

Abstract: In this talk, Hidalgo will present three basic principles that govern the creation and dissemination of knowledge: the principle of experience, the principle of similarity (relatedness), and the principle of the intensity of knowledge. Based on these ideas, Hidalgo will discuss strategies and channels for disseminating knowledge. Finally, he will present tools for the integration, distribution and visualization of data on a large scale (such as datausa.io, datachile.io, dataafrica.io and atlas.media.mit.edu), all of them designed to help the executive and strategic tasks in large organizations.

Bio: César A. Hidalgo leads the Collective Learning group at The MIT Media Lab and is an Associate Professor of Media Arts and Sciences at MIT. Hidalgo’s work focuses on understanding how teams, organizations, cities, and nations learn. At the Collective Learning group, Hidalgo studies knowledge flows and also creates software tools to facilitate learning in organizations. Hidalgo’s academic publications have been cited more than 12,000 times and his online systems have received more than 100 million pageviews and numerous awards. Hidalgo’s latest book, Why Information Grows (Basic Books, 2015), has been translated to over ten languages. Hidalgo is also the co-author of The Atlas of Economic Complexity (MIT Press, 2014), and a co-founder of Datawheel LLC, a company that has professionalized the creation of large data visualization engines. Hidalgo’s contributions were recognized in 2018 with the Lagrange Prize.

Date: September 7th, 2018. 15.30 h.

Venue: Andrónico Luksic Auditorium, Campus San Joaquín, Universidad Católica.

Luis-Daniel Ibáñez: Proyectos QROWD y Observatorio Europeo de Blockchain
Aug
31
2018
Luis-Daniel Ibáñez: Proyectos QROWD y Observatorio Europeo de Blockchain

Abstract: El seminario versará sobre las actividades científicas y de innovación de dos proyectos actualmente en ejecución en la Universidad de Southampton (i) QROWD es una acción innovación patrocinada por el programa Horizon2020 enfocado en la integración de humanos en la cadena de valor de Big Data, en particular para el caso de transporte urbano inteligente. Examinaremos dos herramientas desarrolladas en el marco del proyecto: Un explorador virtual de ambientes urbanos para localizar infraestructura de movilidad mediante trabajadores remotos, y la aplicación de encuestas de movilidad usando teléfonos inteligentes. (ii) El Observatorio de Blockchain de la UE, iniciativa de la Comisión Europea para producir una fuente exhaustiva de conocimiento sobre Blockchains y proponer recomendaciones de políticas, regulaciones y estrategias al respecto. En este marco, examinaremos avances en dos de los temas que la Universidad de Southampton desarrolla para el observatorio: El impacto de la tecnología Blockchain sobre la nueva normativa de protección de datos (y viceversa), y el uso de Blockchains para descentralizar el proceso de desarrollo y comunicación de resultados científicos.

Bio: Luis-Daniel Ibáñez es Doctor por la Universidad de Nantes – Francia, donde trabajó en criterios de consistencia débil para intercambio de grandes números de bases de datos enlazados, en el contexto de la Web de datos abiertos enlazados. desde 2015 se desempeña como Research Fellow en el equipo de Ciencias del Web y la Internet de la Universidad de Southampton, Reino Unido. De 2015 a 2017 se desempeñó como supervisor del proceso de selección de ODINE (Open Data Incubator for Europe), que recibió más de mil propuestas de ideas de negocio alrededor de datos abiertos. Desde 2017 se desempeña como líder técnico de la acción de innovación H2020 QROWD.

Lugar: Sala Felipe Flajolet, Depto. Ciencias de la Computación, Universidad de Chile (Beauchef 851, Piso 3, Edificio Poniente, Santiago)

Fecha: 31 de agosto de 2018

Hora: 12.00 a 13.00 horas

Federico Olmedo: “Differential Privacy: How to make Privacy and Data Mining Compatible”
Aug
31
2018
Federico Olmedo: “Differential Privacy: How to make Privacy and Data Mining Compatible”

Abstract: The data society we live in offers unprecedented opportunities to exploit available data. The goal of _data mining_ is precisely to extract useful information from available data and assist in e.g., decision-making or event-prediction. In doing so, available data usually contain sensitive information about individuals, and releasing even _aggregate_ information about a set of individuals may seriously compromise their individual privacy. In this context, _differential privacy_ has been established as the de facto framework for mining sensitive data in a privacy-aware manner.

In this talk I will give a short (and gentle) introduction to differential privacy. I will approach it from both a theoretical and a practical perspective and conclude discussing some recent developments.

Bio: Federico Olmedo is a full-time professor in the Computer Science Department at the University of Chile. Before joining the University of Chile, he spent three years as a postdoctoral researcher in the Modeling and Verification Group at the RTWH Aachen University, Germany, and in 2014 he earned his PhD degree in Computer Science from the Technical University of Madrid, Spain. His research interests are programming languages, in particular, probabilistic program verification and language-based security.

Date: August 31st, 2018

Time: 15.00-16.00

Venue: Ada Lovelace Auditorium, Computer Sciences Departament, University of Chile

Felipe Tobar: A gentle introduction to Gaussian processes with applications
Aug
24
2018
Felipe Tobar: A gentle introduction to Gaussian processes with applications

Abstract: The Gaussian process (GP) is a probabilistic model for functions. Unique advantages of the GP are its generality, the fact training and prediction can be performed analytically, and its ability to represent uncertainty. In this talk, we first present the concept of a generative model to introduce the probabilistic perspective to Machine Learning, then, via an intuitive extension of basic generative models we introduce the GP model. We will also show how to train (or adjust) the GP model in the light of observed data, and how it can be used in real-world applications of denoising, prediction, reconstruction and deconvolution of general time series. The talk concludes proposing a novel generative model to address an inherent drawback of GP models: its inability to model non-Gaussian data.

Bio: Felipe Tobar holds an MSc in Electrical Engineering from Universidad de Chile in 2010, and a PhD in Signal Processing from Imperial College London in 2014. He was then with the Machine Learning Group at the University of Cambridge as an Associate Researcher for one year. From August 2015 Felipe has been with Universidad de Chile, as a Research Fellow at the Center for Mathematical Modeling. Felipe‘s research interests lie in the interface between Machine Learning and Statistical Signal Processing and include approximate inference, Gaussian processes, kernel methods, and spectral estimation.

Date: 24 de agosto de 2018, 12.00 a 13.00 horas

Venue: Sala Felipe Flajolet, Departamento de Ciencias de la Computación, U. de Chile

I Workshop Instituto Fundamentos de los Datos – 9 y 10 de agosto de 2018
Aug
09
2018
I Workshop Instituto Fundamentos de los Datos – 9 y 10 de agosto de 2018

Abstract: Workshop que reúne a todos los investigadores y colaboradores del Instituto Milenio Fundamentos de los Datos. Se presentarán los Proyectos Emblemáticos, ejes transversales que buscan maximizar el impacto social de las áreas de investigación de nuestro centro. Además, se presentarán los principios transversales del trabajo del instituto, su sentido y objetivo, y los desafíos de la unidad de Innovación y Transferencia.

Date: August 9th and 10th, 2018.

Venue: Hotel Piedra Verde, Machalí, Chile.

Richard Mayr (University of Edinburgh): “Markov Decision Processes and Stochastic Games on Infinite Arenas”
Aug
06
2018
Richard Mayr (University of Edinburgh): “Markov Decision Processes and Stochastic Games on Infinite Arenas”

Abstract: Consider Markov Decision Processes (MDPs) and 2-player turn-based stochastic games on countably infinite game graphs, with objectives expressible by parity conditions (including special cases like reachability, safety, Buchi and co-Buchi objectives). We give an overview of results about the determinacy (for games) and about the memory requirements of epsilon-optimal and optimal strategies, respectively (for MDPs and games). Moreover, we highlight the differences between infinite game graphs and finite game graphs, and the reasons why (epsilon-)optimal strategies need memory for certain objectives.
is a very popular proof assistant, used in a variety of academic and industrial projects, for formalizing both computer science results as well as mathematical ones.

About the speaker: Richard Mayr received a Msc in computer science from TU-Munich, Germany, (1994) and a PhD in computer science from TU-Munich (1998). He received scholarships from the DAAD and the DFG in support of his research at the University of Edinburgh, UK, (1999) and the University of Paris 7, France, (2000), and completed his Habilitation for Informatics at the University of Freiburg, Germany, in 2002. He was assistant professor at the University of Freiburg (2001-2004) and at North Carolina State University, USA, (2004-2007). In 2008 he was appointed to the post of Lecturer at the School of Informatics (LFCS) at the University of Edinburgh, UK. His research interests include automated verification, automata and temporal logic, model-checking and semantic equivalence checking, formal verification of real-time and probabilistic systems, infinite-state Markov chains and stochastic games.

Date: Monday, August 6 2018, from 13:00-14:00.
Venue: Sala Javier Pinto, Edificio San Agustin, 4to piso, PUC, Campus San Joaquín
Gianpiero Canessa (U. Adolfo Ibáñez): “An algorithm for binary chance-constrained problems using IIS”
Jul
20
2018
Gianpiero Canessa (U. Adolfo Ibáñez): “An algorithm for binary chance-constrained problems using IIS”

Abstract: We propose an algorithm based on infeasible irreducible subsystems (IIS) to solve general binary chance-constrained problems. By leveraging on the problem structure we are able to generate good quality upper bounds to the optimal value early in the algorithm, and the discrete domain is used to guide us efficiently in the search of solutions. We apply our methodology to individual and joint binary chance-constrained problems, demonstrating the ability of our approach to solve those problems. Extensive numerical experiments show that, in some cases, the number of nodes explored by our algorithm is drastically reduced when compared to a commercial solver. Keywords: Chance-constrained programming; Infeasible irreducible subsystems; Integer programming.

About the speaker: Ingeniero Civil Industrial con Master en Management Science y actualmente cursando el ultimo agno del programa doctoral DIIIO de la UAI. Trabaje un agno en consultoria y luego se me present la oportunidad de trabajar en un Proyecto de investigacion financiado por INRIA bajo la tutela de la Dr. Javiera Barrera. Desde entonces he dedicado mi vida a la investigacion y docencia, al encontrar una passion que no pude descubrir en la industria. Mi Proyecto future es tratar de ser un Puente entre la industria y la academia para Chile, pues creo que es verdadero camino al desarrollo.

Date: Friday, July 20 2018, from 12:00-13:00.
Venue: Auditorio San Agustín, PUC, Campus San Joaquín

María Jesús Lobo: Transiciones interactivas para aplicaciones cartográficas
Jul
18
2018
María Jesús Lobo: Transiciones interactivas para aplicaciones cartográficas

Abstract: Hoy tenemos acceso a una gran cantidad de datos, provenientes de diferentes fuentes y que presentan diferentes características. Por ejemplo, contamos con múltiples representaciones geográficas: mapas, imágenes satélite, bases de datos vectoriales, entre otras. Los usuarios de sistemas de información geográficos necesitan combinar y comparar estas representaciones heterogéneas de una misma región. Por ejemplo, para mantener la base de datos de OpenStreetMap al día, los mapas existentes se comparan con imágenes satelitales recientes. Sin embargo, la mayoría de las técnicas existentes, como superponer dos representaciones con distintas transparencias o presentarlas de forma yuxtapuesta, no toman en cuenta la información contenida en los mapas. Estas técnicas no son eficientes en tareas relacionadas a áreas como la planificación urbana y la cartografía en situaciones de crisis.

Presentará tres proyectos que estudian este problema. Primero, un estudio de usuarios controlado para entender qué técnicas de comparación de mapas son las más eficientes. Luego, MapMosaic, una técnica dinámica de composición de mapas que se basa en la información geográfica. Discusiones con usuarios expertos sugieren que esta técnica permite una mejor exploración de la información espacial que los sistemas existentes. El tercer proyecto es Baia, un modelo para definir animaciones avanzadas entres dos imágenes satélite y que permite la creación de animaciones que son percibidas como más realistas que las transiciones existentes.

Bio: María Jesús Lobo (PhD INRIA), Postdoctoral Researcher ENAC, Toulousse, Francia en el área de Visualización de Datos interactiva. Obtuvo el grado de Magister en Ciencias de la Ingeniería en la Pontificia Universidad Católica de Chile y luego realizó su doctorado en la Université Paris Sud bajo la supervisión de Emmanuel Pietriga y Caroline Appert, en dónde se especializó en las áreas de interacción humano computador y visualización de datos. Ha publicado en importantes conferencias como ACM CHI y en journals como IEEE TVCG .Se interesa especialmente en nuevas técnicas para visualizar e interactuar con datos espaciales heterogéneos, como mapas topográficos e imágenes satelitales.

Fecha y Hora: Miércoles 18 de Julio, 13:00 hrs.
Lugar: Sala Javier Pinto, DCC UC, Campus San Joaquín, Chile
Alejandro Corvalán: “MACEDA dataset on the Mapuche conflict”
Jun
29
2018
Alejandro Corvalán: “MACEDA dataset on the Mapuche conflict”

Abstract: Self-determination (SD) disputes are some of the most common conflicts in the world. Around a third of the civil wars in the last decades relate to demands for increased autonomy or independence (Sambanis and Milanovic, 2014). While the literature has mainly focused on the most violent conflicts, such as civil wars, recent studies have acknowledged the diversity of strategies adopted by groups within SD movements, including a wide range of less violent and non-violent strategies (Cunningham et al., 2017; Cunningham, 2013). Recent cross-country data have shown how these strategies differ not only between SD movements but also between groups that take part in each SD movement (Sambanis et al., 2018; Cunningham et al., 2017). Of the more than three hundred stateless nations and groups pressing for greater self-determination, around 12% are indigenous groups in The Americas (Sambanis et al., 2018). Their claims are mostly an increase in autonomy within the state rather than independence.

In this talk we focus on a particular case: the conflict between the Mapuche indigenous group and the Chilean state. We introduce MACEDA, the first systematic record of the events related to this conflict. The conflict has its roots in a military confrontation in the second half of the nineteenth century, which ended in the incorporation of the Mapuche territory into the effective control of the Chilean state, with most land privatized and the indigenous population confined to reductions. Disputes over land and other resources as well as claims for greater autonomy have existed since then. We focus on the post-dictatorship period, collecting data between 1990 and 2016.

MACEDA reports almost 2,000 events manually coded from local media. With these data we show a wide-range of strategies adopted by different actors within the Mapuche SD movement. The strategies refer to non-violent actions beyond institutional politics, such as land invasions, mass demonstrations and hunger strikes, as well as actions adopting a varying degree of violence including arsons, riots and bombs. We also report the responses by the Chilean state, which include police raids, ejections, detentions and the use of the anti-terrorism law. Beyond shedding light on the strategies taken by the actors within this SD conflict, we provide information about the evolution of these strategies. The conflict has become more violent over time, and the number of strategies and actors involved has increased and diversified. MACEDA provides data that is useful for the understanding of the determinants, mechanisms and consequences of the Mapuche SD conflict. Moreover, the data set contributes to improving the understanding of indigenous SD conflicts, and less-violent SD conflicts in general, for which only scarce empirical literature exits.

About the speaker: Alejandro Corvalán obtained his BSc. and MSc. in Physics, and MSc. in Economics at Universidad de Chile, and Ph.D. in Economics at New York University. He is an Associate Professor at Universidad Diego Portales and Research Associate at the Millennium Institute for Market Imperfections and Public Policy. His research field is Political Economy.

Date: Friday, June 29 2018, from 12:00-13:00.

Venue: Sala N Bralic , Facultad de Matemáticas, P. Universidad Católica, Campus San Joaquín.

Éric Tanter, U. of Chile: “A (Way Too) Short Introduction to Coq”
Jun
22
2018
Éric Tanter, U. of Chile: “A (Way Too) Short Introduction to Coq”

Abstract: Coq is a very popular proof assistant, used in a variety of academic and industrial projects, for formalizing both computer science results as well as mathematical ones. Coq is based on a theoretically very clean model, the Calculus of Inductive Constructions, which allows to express all of mathematics in a constructive manner. Following the Curry-Howard correspondence, Coq is also a functional programming language. A logical proposition is a type, and a proof of the proposition is a well-typed program of that type.

In this 1-hour quick tutorial, we will give a very brief introduction to the basics of Coq and how it can be used to reason about languages. Starting from an elementary language of arithmetic expressions, we will build a compiler to a simple stack machine and prove it correct. We will also implement a simple program rewriting optimization and prove it correct. This toy example should give you a good idea of what Coq is about, and the joy (and pain!) of mechanized semantics.

About the speaker: Éric Tanter is a Full Professor of the Computer Science Department of the University of Chile. He received his PhD from both the University of Nantes and the University of Chile in 2004. His research interests cover programming languages and software engineering, ranging from the theoretical underpinnings of programming languages to the empirical study of the practice of programming. He has published many articles in, and is regularly involved in the program committees and editorial boards of, major conferences and journals in these areas. Recently, he has been mostly involved in the foundations and practice of gradual typing and verification.

Date: Friday, June 22 2018, from 12:00-13:00.

Venue: Auditorio Philippe Flajolet, Universidad de Chile, Beauchef 851, Santiago.

Francisco Förster: “The Universe in a stream”
Jun
01
2018
Francisco Förster: “The Universe in a stream”

Abstract: With a new generation of large etendue (the product of field of view and mirror area) survey telescopes there is a growing need for astronomical alert processing systems. Astronomical alerts correspond to detected changes in the sky with an astrophysical origin. Astronomical alert processing systems involve the real–time processing of data for alert generation, real–time annotation and classification of alerts (up to 10 million events per night) and real–time reaction to interesting alerts using available astronomical resources. We are building a new alert classification and reaction system called ALeRCE: Automatic Learning for the Rapid Classification of Events. ALeRCE is an initiative led by an interdisciplinary and interinstitutional group of scientists from U. Católica (DCC), U. Chile (CMM, DIE), U. Concepción (DCC), the Millennium Institute for Astrophysics – MAS and U. Nacional Andres Bello – UNAB (DCF) in Chile, in collaboration with international researchers from Caltech (CD3), Harvard U. (IACS–SEAS) and U. of Washington (Dirac). In this talk I will discuss some data science challenges in astronomy and in particular for the problem of alert classification, including the ingestion, annotation, database access, processing power, machine learning classification and visualization of these alerts.

About the speaker: Francisco Förster Burón is an astronomer and Research Scientist at the Center for Mathematical Modelling (CMM) at Universidad de Chile. He is also a young researcher of the Millennium Institute of Astrophysics (MAS) and a Fondecyt Iniciación Fellow. He did his Ph.D. thesis on Type Ia Supernovae progenitors at the University of Oxford (2009) with Philipp Podsiadlowski. His current work focuses in constraining the progenitors of supernovae via a real-time analysis of large volumes of data taken with the DECam instrument and using the supercomputer at the National Laboratory for High Performance Computing.

Date: Friday June 1 2018, from 12:00-13:00.

Venue: Auditorio Philippe Flajolet, Universidad de Chile, Beauchef 851, Santiago.

Prof. Jeremy Barbay, U. of Chile: MultiVariate Analysis of Dynamic Programming
May
11
2018
Prof. Jeremy Barbay, U. of Chile: MultiVariate Analysis of Dynamic Programming

Abstract: Many practical problems can be reduced recursively to smaller or simpler instances, down to the base cases. Most often, a straightforward implementation of such reduction fails to yield a solution running in reasonable time. In many cases it is Dynamic Programming which yields solutions of industrial value, based on an adequate tuning of the reduction and the memoization of past computations, in areas as diverse as Stringology, Bio Informatics and many other areas, with solutions running in time polynomial in the size of the input, and space linear in this input size (e.g. Longest Common Sub-Sequence (LCSS) in time within $O(mn)$ and space within $O(n+m)$). Yet, such worst-case results analysis of algorithms has often been criticized as overly pessimistic. As a remedy, some researchers have turned towards multivariate analysis where the execution cost of algorithms is measured as a function of not just the input size but of other parameters that capture, in various ways, the difficulty of the input instance. This approach has been the subject of extensive work on arrays of elements on problems such as sorting, computing the intersection or the union of sorted arrays; and on data structures supporting operators on permutations, Multisets and on Strings. While the technique of dynamic programming is used in many polynomial algorithms introduced in the context of parameterized complexity of NP-hard problems, the dynamic programming is rarely analyzed parametrically itself. The only exception, in the case of the computation of the Insert Swap Edit Distance, yield a much better understanding of the problem, and even improvements on the worst case complexity among instances of fixed sizes.

We aim to systematically consider the application of multivariate analysis to other problems for which dynamic programming has yield good results, such as those used for the computation of the Frechet Distance between two sequence of points, with application to computational geometry, and to the indexed search in multimedia databases; of the distance between skeletons and meshes, with application to computational geometry, and to the indexed search in multimedia databases; of the String Edit Distance between two strings, for various subsets of operators from the commonly considered set {Delete, Insert, Replace, Swap}, with applications to Bioinformatics and text search; of the Tree Edit Distance between two structured texts, with applications to phylogenetics, to the detection of plagiarism and to the evaluation of commits in systems of version control; of the Block Edit Distance between two strings, with applications to the analysis of genomes as well as to the detection of plagiarism.

About the speaker: Jeremy received a Bachelor of Science degree in Mathematics in 1997 in Rouen, a Master degree in 1998 and a Philosophy Doctorate in 2002, both in Computer Science at the Laboratoire de Recherche en Informatique of the University of Orsay, under the supervision of Claire Mathieu. He was a posdoctoral fellow at the department of Computer Science of the University of British Columbia until 2004 and an assistant professor at the Cheriton School of Computer Science of the University of Waterloo until 2008. He is now an assistant professor at the department of Computer Science of the University of Chile in Santiago, Chile.

His main research is about the analysis of algorithms and data-structures on finer classes of instances than those merely defined by their size, which yields the concepts of adaptive (analysis of) algorithms, instance optimality, output sensitive and parameterized complexity, compressed data structures and indexes, and of formal measures of compressibility. His work has contributed to clarify the relations between those topics and has introduced a few useful concepts, such as the direct relation between permutation compression and adaptive sorting (TCS2013); the first Compressed Index achieving space within o(n Hk) + O(n) instead of merely within o(n lg s) (ALGO2014); Succinct Indexes (TALG2011); and (input order oblivious) Instance Optimality in Computational Geometry (FOCS2009). Albeit a theoretician by formation, Jeremy did implement and experimentally evaluate some of his theoretical results, either in collaboration with students or on his own (JEA2009,WEA2006). He is interested in other topics of research, in particular in pedagogy.

Date: Friday, May 11th 2018.

Time: 12:00-13:00.

Where: Sala Álvaro Campos, DCC Universidad Católica, Campus San Joaquín.

Víctor Dalmau: Approximation of MIN CSPs
Apr
27
2018
Víctor Dalmau: Approximation of MIN CSPs

Abstract: An instance of the constraint satisfaction problem (CSP) is given by a family of constraints on overlapping sets of variables, and the goal is to assign values from a fixed domain to the variables so that all constraints are satisfied. In the optimization version, the goal is to maximize the number of satisfied constraints (MAX CSP) or, alternatively, to minimize the number of unsatisfied constraints (MIN CSP). This problem is usually parameterized by the set, Gamma, of relations allowed in the constraints, usually called constraint language. It turns out that MAX CSP/MIN CSP is computationally hard for most constraint languages, which motivates the study of approximation algorithms. In this talk we will focus on the approximation of MIN CSPs. We shall start addressing the following question: which constraint languages give rise to a MIN CSPs that is constant-factor approximable? We shall also study some other weaker approximation notions such polynomial loss and robust approximation.

Bio: Associate Professor at the Department of Information Technologies, Universitat Pompeu Fabra, Barcelona, Spain. He obtained a degree and a Ph.D on Computer Science at Universitat Politécnica de Catalunya, Spain. His main research is on Constraint satisfaction, which is the problem of deciding whether there exists an assignment of values to variables satisfying some given restrictions. The framework is general enough to express many common problems in areas such as logistics, computer vision, scheduling, and artificial intelligence, to name only a few. His work focuses on the theoretical aspects of the problem, involving techniques and concepts coming from areas as diverse as combinatorics, logic, database theory and universal algebra. He is also interested broadly in complexity theory, logic in computer science, and computational learning theory. He has published 23 journal papers and 29 conference (peer-reviewed) papers. He received the “Ramón y Cajal” Fellowship and the ICDT 2012 best paper award.

Date: Friday 27th April 2018, 10.00.-

Place: Auditorio San Agustín, Campus San Joaquín, Pontificia Universidad Católica, Santiago. Chile.

Dr. Erick Elejalde, Universidad de Concepción What the media do in the shadows: A computational investigation of the Propaganda Model
Apr
13
2018
Dr. Erick Elejalde, Universidad de Concepción What the media do in the shadows: A computational investigation of the Propaganda Model

Abstract: The Propaganda Model (PM) discussed in Manufacturing Consent is a theory in political economy that states that the mass media are channels through which governments and major power groups pass down certain ideologies and mold a general consent according to their own interests. According to the authors, every piece of news has gone through a set of filters that ultimately yield the source event as newsworthy. Current developments in communications, the digital availability of large-scale of news online streaming from every corner of the world, together with our increasing capability to process all this information in a lot of different ways, give us the perfect environment to test social theories using quantitative methods. In our work we take advantage of all these data to test, empirically, the theory laid out in the PM. Previous works have used machine learning and natural language processing techniques, but focused only on showing bias to a political party by a sample of the major news outlets. Here we make a first attempt in the formalization of the model and the filters, and we help to provide an explanation of how the media works taking a computational approach.

Results illustrate a measurable media bias, showing a marked favoritism of Chilean media for the ruling political parties. This favoritism becomes clearer as we empirically observe a shift in the position of the mass media when there is a change in government. Furthermore, results support the PM by showing that high levels of concentration characterize the Chilean media landscape in terms of ownership and topical coverage. Our methods reveal which groups of outlets and ownership exert the greatest influence on news coverage and can be generalized to any nation’s news system. Our studies on the geographic news coverage also give indications of the presence of the second filter (advertising). Experiments on predicting the communities with the biggest share of readership show this to be highly correlated with those regions with the greatest population, better socioeconomic status, and a distinct political preference. As far as we know, this is the first time that there has been an attempt to empirically test this (or any other) political economy theory using data science.

Bio: Erick Elejalde has a Ph.D. in Computer Science from the University of Concepción under the supervision of Leo Ferres, Ph.D. and Johan Bollen Ph.D. Erick has research experience on concurrent data structures, multicore algorithms, big data analysis and social media.

Date: Viernes 13 de Abril
Time: 10:00 – 11:00 hrs.
Where: Auditorio Ramón Picarte DCC (3er. piso, Edificio Norte, Beauchef 851, Santiago)