• EVIC 2021

    Since 2004
    Summer school
    in computational
    intelligence
    Free registration
    Transmission on youtube

What is EVIC?

The Latin American Summer School of Computational Intelligence (from its spanish acronym EVIC) is an annual event that has been held in Chile since 2004. Its objective is to promote Computational Intelligence, bringing together areas such as Intelligent Control, Robotics, Computer Vision, Computational Neuroscience, Science of Data, Pattern Recognition and Big Data among others. The event presents from the basic concepts to the latest research results in these areas. Its target audience is undergraduate and graduate students, university professors and professionals, from Chile and Latin American countries. EVIC is organized by the Chilean Chapter of the IEEE Computational Intelligence Society (IEEE-CIS) and the host University. EVIC is an event that was born at the Universidad de Chile, but to achieve a greater scope and diversity of its attendees, since 2013 it has become itinerant, taking place in different universities in the country, such as the University of La Frontera (EVIC 2013), the University of Concepción (EVIC 2015), the University of Valparaíso (EVIC 2017), the University of Santiago (EVIC2018) and the Austral University (EVIC 2019). This year it virtually returns to Temuco.

Computational Intelligence (CI) is an area of Engineering that is having a strong impact in all areas of knowledge, since it provides tools to automate the analysis of information and create systems capable of making autonomous decisions. These technologies will have a strong impact on society, strongly affecting the economy and people's daily lives, while offering opportunities to create new poles of development for the region. As in previous years, we have invited renowned speakers who will present their work and contributions in various applications areas of CI. The EVIC 2021 program includes Tutorials and Plenary Talks. In addition, due to the pandemic that affects us, this event will be online and free for its participants.

In the same week, another congress is realized by the Latin American Conference on Computational Intelligence (LA-cci). For more information click here

We then cordially invite you to be attentive to the registration deadlines so that you can participate in this interesting event. Very cordially.

EVIC 2021 Organizing Committee.

Place and date

EVIC 2021 will be realized by Universidad de La Frontera (UFRO) Chile, from December 4th to 5th 2021.

Program

EVIC2021 will be transmited on air on youtube in the EVIC-UFRO channel:click HERE

Program

During the days the following activities will be carried out:

  • 4 plenary sessions for all attendees. The plenary sessions will be held in the morning.
  • 4 tutoring sessions within 2 parallel tutorial tracks presented every day.

KEY SPEAKERS

Carol Huilin

Carol Huilin

Title: Smart digital innovation for human dignity

Resume: In times of pandemic, people's biometric data has become gold for digital platforms, and that is why digital innovations are being used to protect people's health. This master class will share some innovations created by Chileans for the self-care of people in times of crisis.

Barbara Poblete

Barbara Poblete

Title: Mining Social Networks to Learn about Rumors, Hate Speech, Bias and Polarization

Resume: Online social networks are a rich resource of unedited user-generated multimedia content. Buried within their day-to-day chatter, we can find breaking news, opinions and valuable insight into human behaviour, including the articulation of emerging social movements. Nevertheless, in recent years social platforms have become fertile ground for diverse information disorders and hate speech expressions. This situation poses an important challenge to the extraction of useful and trustworthy information from social media. In this talk I provide an overview of existing work in the area of social media information credibility, starting with our research in 2011 on rumor propagation during the massive earthquake in Chile in 2010. I discuss, as well, the complex problem of automatic hate speech detection in online social networks. In particular, how our review of the existing literature in the area shows important experimental errors and dataset biases that produce an overestimation of current state-of-the-art techniques. Specifically, these issues become evident at the moment of attempting to apply these models to more diverse scenarios or to transfer this knowledge to languages other than English. As a particular way of dealing with the need to extract reliable information from online social media, I talk about two applications, Twically and Galean. These applications harvest collective signals created from social media text to provide a broad view of natural disasters and real-world news, respectively.

Biography: https://imfd.cl/investigador/barbara-poblete/

Alice Smith

Alice Smith

Title: Engineering System Design Using Natural Systems

Resume:This talk will put forth several straightforward but successful implementations of analytical approaches inspired by natural systems to difficult engineering system design problems. These nature-based paradigms range in fidelity with their natural systems origins but all seek to leverage the structures and operations of nature doing what it does best – novelty detection, system optimization, adaptability to dynamic environments, robustness, and flexibility. More specifically, the well-known, but often misunderstood and misused, artificial intelligent paradigms of artificial neural networks, fuzzy logic, and evolutionary algorithms will be considered for use in engineering system design. Used judiciously and knowledgeably these approaches can offer significant advantages in diverse production and operational environments. A curated selection of applications from the speaker’s more than 25 years of experience in this field will be explained and objectively analyzed. The applications are (1) quality and process improvement of large-scale ceramic casting, (2) real-time placement of drones for ad hoc communications network connectivity, and (3) continuous monitoring of airport vehicles for predictive maintenance.

Biography: https://www.eng.auburn.edu/~aesmith/

Piero Bonissoni

Piero Bonissoni

Title: Analytics for Industrial AI: Leveraging Model Ensembles

Resume: In the past, analytic model creation was an artisanal process, as models were handcrafted by experienced, knowledgeable model-builders. More recently, the use of meta-heuristics, such as evolutionary algorithms, has provided us with limited levels of automation in model building and model maintenance. Data-driven models are becoming a commodity, as we have access to a large number of data-driven models by a combination of crowdsourcing, cloud-based evolutionary algorithms, public-domain libraries, outsourcing, in-house development, and legacy models. In this context, the critical issue will be model ensemble selection and fusion, rather than model generation. First, we will illustrate the use of model ensembles within the context of assests Prognostics and Health Maintenance (PHM) of assets such as aircraft engines, medical imaging devices, and locomotives. We will cover a few case studies of anomaly detection, diagnosis, prediction, and optimization. Then, we will describe the evolution of analytic models in the era of cloud computing, and propose the use of customized model ensembles on demand, inspired by Lazy Learning. This approach is agnostic with respect to the origin of the models, making it scalable and suitable for a variety of applications. We will present results on the fusion of data-driven models for FlyQuest, a GE-sponsored Kaggle competition, in which we crowdsourced the generation of models predicting the estimated runway and gateway arrival (ERA, EGA) over a month of US flights. We will also describe the fusion of hybrid ensembles, combining physics-based and data-driven models to leverage domain knowledge to improve the performance of analytics. Finally, we will highlight some research trends, challenges and opportunities for Machine Learning techniques in this dynamic context of big data and cloud computing.

Biography: https://www.datasciencecentral.com/profile/PieroPBonissone

TUTORIALS SPEAKERS

Felipe Tobar

Felipe Tobar

Title: Multioutput Gaussian Processes

Resume: This talk will focus on the multioutput extension of Gaussian processes (GP), referred to as Multioutput Gaussian processes (MOGP). Akin to their scalar-valued counterpart, MOGPs are Bayesian nonparametric generative models for time series, which, in addition to modelling temporal dependencies among data, also account for across-channel relationships. In this regard, the main challenge in MOGPs is the construction of covariance functions that are as capable as possible to identify relationships among different time series while fulfilling the structural properties (e.g., positive definiteness) of the full multioutput covariance. We will start with a motivation for MOGPs and they can be constructed by mixing independent GPs, then, we will revise standard approaches to covariance design and their implications. Lastly, we will present dedicated software for MOGPs with examples and real-world applications.

Biography: http://www.dim.uchile.cl/~ftobar/

Marie Gonzalez Carlos Molina

Marie González / Carlos Molina

Title: “Logmeter: un sistema de medición automática industrial. Generación de modelos predictivos para el volumen sólido de madera y algoritmos de clasificación que incorporan una alta componente de visión en tiempo real”

Resume: Logmeter es un sistema automático que usa tecnología Laser Lidar para estimar el volumen y las variables biométricas de la madera apilada siendo transportada en camiones. Este sistema se basa en generación de nubes de puntos 3D y en algoritmos matemáticos avanzados para calcular el volumen. Luego, usando metodologías de Machine Learning se construyen modelos predictivos para estimar o predecir el volumen sólido de cada camión. Los modelos predictivos logran relacionar las variables biométricas estimadas automáticamente por el Logmeter con un volumen sólido utilizado como referencia. En el trabajo que se presentará aquí, el volumen sólido de referencia está estimado usando una metodología basada en las fórmulas de Arquímedes. Los resultados obtenidos hasta ahora muestran un sistema capaz de predecir el volumen sólido en forma automática con un intervalo de confianza cercano al +-10% (intervalo al 95%) a nivel de banco y +-7% a nivel de camión. Estos resultados muestran la precisión y exactitud del sistema Logmeter y reafirma su posición como estado del arte en medición automática de volumen sólido de madera. La evolución del sistema Logmeter considera la utilización de visión artificial en tiempo real con el objetivo de realizar un conteo de la carga durante el paso del camión para condiciones lumínicas y ambientales de una faena forestal. Para esto, se utilizó la red convolucional YOLO, capaz de segmentar múltiples objetos en imágenes. Adicionalmente, se trabajó en generar un sistema de iluminación y control de cámaras que permitiera el correcto funcionamiento durante las 24 horas del día. Luego de realizar un entrenamiento con múltiples condiciones lumínicas, ambientales y de carga del camión, se incorporó un sistema de selección de conteo para cada banco. La implementación actual logra contar los troncos totales con un error de 0.35% y de 2.38% en la evaluación por banco. En suma, en este tutorial se detallan dos aproximaciones de utilización de técnicas de Machine Learning que permiten estimar variables de faenas forestales en condiciones reales .

Marie González Biography: Ingeniera eléctrica con Magíster en ciencias de la Ingeniería de la Universidad Católica de Chile, con especialización en procesamiento de señales. He desarrollado investigación y desarrollo con foco en la aplicación de técnicas de visión computacional, inteligencia de máquina y reconocimiento de patrones en áreas diversas, que incluyen sistemas con aplicaciones industriales y artísticas. Actualmente me desempeño como ingeniera de desarrollo en Woodtech S.A., donde mi trabajo se ha enfocado en el desarrollo de sistemas de medición para procesos de la industria forestal, mediante el uso de técnicas de visión artificial en tiempo real.

Carlos Molina Biography: Doctor en Ingeniería eléctrica de la Universidad de Chile. Durante mi desarrollo académico mi focalización estuvo en reconocimiento de patrones, aprendizaje de máquinas, procesamiento de señales y tecnologías de la comunicación e información. Mi trabajo específico estuvo centrado en las tecnologías del habla, en particular en el campo del reconocimiento de voz robusto y también en el aprendizaje de segundo idioma usando procesamiento de voz. En mi estadía como investigador de la Universidad de Chile, mi experiencia se vio significativamente incrementada tanto en el ámbito académico como en el área de las ciencias aplicadas, trabajando en investigaciones multidisciplinarias y liderando equipos de desarrollos donde fue posible alcanzar interesantes desarrollos en el campo de las telecomunicaciones y el reconocimiento de patrones, consiguiendo además, importantes financiamientos estatales y privados. En el presente, trabajo como líder del equipo de desarrollo en la empresa Woodtech en donde la mayor parte del tiempo estamos dedicados a desarrollar software con una alta componente tecnológica en procesamiento de imágenes, nubes de puntos 3D, etc., particularmente para la industria forestal, tanto nacional como internacional.

Leticia Seijas

Leticia Seijas

Title: “Las redes neuronales convolucionales en la sociedad de la Inteligencia Artificial”

Resume:En la actualidad, las redes neuronales convolucionales o CNN, enmarcadas en el paradigma de Deep Learning, toman cada vez más relevancia en el camino hacia la sociedad de la Inteligencia Artificial (IA), donde es necesario resolver problemas que involucran el tratamiento de grandes volúmenes de datos y su análisis. Esto obliga a desarrollar modelos más complejos que permitan manejar conceptos de alto nivel de abstracción. En este contexto, las CNN son una herramienta poderosa que permite desarrollar diversas aplicaciones que impactan en nuestras sociedades, desde aquellas que monitorean aspectos medioambientales, por ejemplo a través de la clasificación de imágenes satelitales, hasta otras vinculadas a la salud, como el diagnóstico de enfermedades respiratorias y COVID-19 en Radiología, entre muchas otras. Este tutorial tiene el objetivo de presentar las CNN y su relevancia en la actualidad. Se describirá el contexto que las convierte en una herramienta fundamental, el paradigma al que pertenecen dentro de la IA, sus principales características y estructura, aspectos a ser considerados en el entrenamiento, arquitecturas representativas, aspectos de implementación incluyendo transfer learning, ejemplos prácticos y aplicaciones en general.

Biography: Leticia Seijas has received her Ph.D. in Computer Science at Universidad de Buenos Aires (UBA), Argentina (2011). She is Professor in Computer Engineering and Informatics at Universidad Nacional de Mar del Plata (UNMDP), Argentina, since 2017. Researcher at Institute of Scientific and Technological Research in Electronics ICYTE-CONICET/UNMDP. Main research areas: Pattern Recognition, Deep Learning, Signal Processing, DataMining. Teaching and research at UBA (2003-2017). She has more than two decades of professional work in Information Technology and Computer Science. Senior Member of the IEEE, member of CIS and GRSS Chapters. She has been President of the IEEE Computational Intelligent Society-Argentina Section (2018).

Fernando Buarque

Fernando Buarque

Title: Fish School Search (FSS) for Beginners

Resume: Fish School Search (FSS), proposed by Bastos Filho and Lima Neto in 2008 is an optimization algorithm inspired on the collective behavior of fish schools. In the session we are going to learn the mechanisms of feeding and coordination movement that create the search operators, the core idea that makes the fishes “swim” toward the positive gradient in order to “eat” and “gain weight”. We will also learn how the collective behavior in which heavier fishes influence the search process as a whole, as well as what makes the barycenter of the fish school move toward better places in the search space over the iterations. Finally, the FSS main principles will be explained to the audience, namey: Simple computations in all individuals (i.e. fish) Various means of storing information (i.e. weights of fish and school barycenter), Local computations (i.e. swimming is composed of distinct components), Low communications between neighboring individuals (i.e. fish are to think local but also be socially aware), Minimum centralized control (mainly for self-controlling of the school radius), Some distinct diversity mechanisms (this to avoid undesirable flocking behavior), Scalability (in terms of complexity of the optimization/search tasks) and Autonomy (i.e. ability to self-control functioning). The websites of FSS are https://fbln.me/fss/ and https://en.wikipedia.org/wiki/Fish_School_Search

Biography: https://fbln.me

OUTREACH

José Alfredo Costa

José Alfredo Costa

Title: Considerations on ethics in AI applications

Resume: We are facing the digital transformation. The use of advanced systems using Artificial Intelligence (AI) has grown exponentially. The talk will address some considerations on ethical dilemmas in the use of AI and their implications, including personal data usage in social networks, autonomous cars, industrial automation, among others. Great challenges for AI systems include explainability and transparency in decision making by machine learning algorithms, e.g., in deep learning models. Another issue is identifying biases (in the data) and their mitigation, if not the elimination of such problems, which may lead to system unusability, e.g., by government and institutions.

Biography: http://www.dee.ufrn.br/~alfredo/

SPONSORS

Pablo Vera Cadenas

ARAUCO

Pablo Vera Cadenas - Líder Data Scientist

Biography: Magister en Ciencias de la Ingeniería e Ingeniero Civil Electricista de la Universidad de Chile, Diplomado en Big Data y Data Science de la Pontificia Universidad Católica de Valparaíso con trece años de experiencia en Investigación y Desarrollo en temas relacionados con Inteligencia Artificial y Big Data. Durante los últimos años me he enfocado en la investigación aplicada en la innovación para la industria minera y forestal. Científico de Datos Líder en Centro de Transformación Digital, Arauco.

Pablo Zegers

ANASTASIA

Pablo Zegers - C.O. Anastasia

Biography: Pablo Zegers was born in Santiago, Chile, in 1968. He received his Bachelor in Industrial Engineering degree and his Professional Industrial Engineer certification from the Pontificia Universidad Católica, Chile, in 1992, his Master of Science degree from The University of Arizona, USA, in 1998, and his Doctor of Philosophy degree, also from The University of Arizona, in 2002. He started to work as a professor at the College of Engineering and Applied Sciences of the Universidad de los Andes, Chile, in 2002 and left the university as an Associate Professor in 2017. He was the Academic Director of that College from 2006 to 2010, and its Interim Dean for a brief period at the end of 2010. During his stay at the university, he led, or was part of, the teams that designed the strategic plan of the college, created Electrical Engineering and Computer Science, renovated the academic curricula of all the departments of the college, hired tens of professors, and reorganized the administrative structure of the college. As a teacher, his main achievement was advising more than 90 engineering students through their final engineering thesis. He has many publications whose topics range from the use of the mathematical theory of information theory in artificial intelligence to the design of algorithms that found thousands of binary stars in the Magellanic Clouds that orbit our galaxy. His interests are information theory, artificial intelligence, machine learning, and neural networks. He is a co-founder of Anastasia, focused on developing artificial intelligence tools for commercial businesses, and Sortbox, dedicated to building machines based on artificial intelligence for the aggrotech sector. He was recognized as the 2020 Outstanding Electrical Engineer by the Chilean electrical companies and the IEEE. He was a Member of the Board of the Chilean IEEE Section, and he is a Senior Member of the IEEE.

Registration form

FREE REGISTER

Fill the form with your data and then push SEND button. After this, you will receive an email with the zoom link to the speaks.

Nombre

Last name

Second last name

Country

City

Address

Email

Repeat the Email

Activity

Career

University or Institute where study/studied

Native People

IEEE Membership Number (optional)

***Any question consult: evic@ufrontera.cl

Team organizer

FERNANDO HUENUPAN

General Chair

Universidad de la Frontera

MILLARAY CURILEM

Organizing member

Universidad de la Frontera

CESAR SAN MARTIN

Organizing member

Universidad de la Frontera

NELSON AROS

Organizing member

Universidad de la Frontera



José Delpiano

Organizing member

Universidad de Los Andes

Sebastían Maldonado

Organizing member

Universidad de Chile

Pablo Zegers

Organizing member

Sortbox/Anastasia/Webdox

VICTOR POBLETE R.

Organizing member

Universidad Austral

DORIS SÁEZ

Organizing member

Universidad de Chile

CLAUDIO HELD

Organizing member

GONZALO ACUÑA

Organizing member

Sponsors

Logo UFRO
Logo IEEE
Logo Alaya
Logo Anastasia
Logo Anastasia

Copyright © 2021 - EVIC2021 | All Rights Reserved