Creating and sharing knowledge for telecommunications

Workshop on Interactions Between Nonextensive Entropies, Machine Learning, Language, and Physics


on 05-07-2022

... Project DeepSPIN is organising the Workshop on Interactions Between Nonextensive Entropies, Machine Learning, Language, and Physics on July 5, 2022, which will be held on IST's Amphitheaterre Abreu Faro, in the Interdisciplinary Complex.

Nonextensive statistical mechanics [1] is a generalization of the standard Boltzmann-Gibbs theories of statistical mechanics, inspired by the seminal work of Constantino Tsallis. This generalized theory has had a very strong impact in many disciplines and a wide range of applications, including statistical mechanics, thermodynamics, information geometry, statistics, machine learning, and natural language processing.

In this interdisciplinary workshop, will be present Constantino Tsallis, as well as speakers from several disciplines, ranging from mathematics and physics to machine learning and language, who will discuss their use of nonextensive entropies and Tsallis statistics on various applications, namely, Mário Figueiredo, Ben Peters, André Martins, José Mourão, and Frederico Fiuza.


Program:
11:00 - 12:00 | "Why is it easier to understand what is energy than what is entropy?" by Constantino Tsallis
14:00 - 14:30 | "Tsallis entropies and kernel methods" by Mário Figueiredo
14:30 - 15:00 | "Tsallis entropies and entmax for language generation" by Ben Peters
15:00 - 15:45 | "From Sparse Modeling to Sparse Communication" by André Martins
(Coffee break)
16:00 - 16:3 | “Training deep neural networks: replace gradient descent by the Feynman path integral and possible extension to the nonextensive formalism” by José Mourão
16:30 - 17:00 | “Accelerating the understanding of nonlinear dynamical systems using machine learning” by Frederico Fiuza


The workshop will be also transmitted by Zoom. Here's the link: https://videoconf-colibri.zoom.us/j/91599759679


[1] Tsallis, Constantino, "Introduction to nonextensive statistical mechanics: approaching a complex world." Springer 1.1 (2009): 2-1


To know more about this project, follow the link: More Information..

In Memoriam session “An overview of the many contributions of José Manuel Bioucas-Dias to remote sensing”


on 29-06-2022

... On June 29, 2022, at 12:30 pm, it will take place the In Memoriam session “An overview of the many contributions of José Manuel Bioucas-Dias to remote sensing”, presented by Mário Figueiredo (DEEC, IST-IT Lisbon)

This session takes place within the scope of the “Spatial Data Science Webinar series… think spatially about your data science problems” promoted by CERENA (Centro de Recursos Naturais e Ambiente) in partnership with INESC-ID.

José Bioucas-Dias (1960-2020)
Graduated, PhD and aggregated in Electrical and Computer Engineering from Instituto Superior Técnico in 1985, 1995 and 2007, respectively. He has been a professor at the Department of Electrical and Computer Engineering (DEEC) since 1985. He was our member, belonging to the Image Analysis and Patterns group (IT Lisbon), where he developed research in image processing, optimization, and remote sensing. In 2019, he was among the most influential researchers in the world, for the second year running, as he was part of the top 1% of scientists whose work is most cited by other colleagues in their field, in this case in the Geosciences category. According to Clarivate Analytics, the list features the “most influential scientific minds in the world”, representing one of the most prestigious sources for assessing the impact of scientific publications, researchers and institutions.


Photo credits: IST

Further information: https://tecnico.ulisboa.pt/pt/eventos/ciclo-de-webinars-metodos-e-aplicacoes-da-ciencia-de-dados-espaciais/

The seminar takes place in person (in the room VO.15 - Civil Building), but can also be followed by zoom transmission at:
More Information..