PhD in Informatics Seminar

Dynamic Ensemble of Content-based Image Retrieval Systems using Online Learning with Expert Advice

Transmissão através de Videoconferência

Por Soraia Alarcão.

Content-Based Image Retrieval (CBIR) systems are useful to store and efficiently retrieve similar images from large collections. The main challenge in CBIR systems is to choose features that are sufficiently discriminative to represent the images under comparison while keeping them compact to ensure that the system is timely and computationally efficient. Since human perception of image similarity is subjective, semantic and task-dependent, unveiling the perfect combination of discriminative features is highly domain-specific and dependent on the type of image.
The most common approach is to combine multiple descriptors using an early fusion approach, where all descriptors are assumed to have the same importance, but they may not yield the same results for different types of images. An alternative is to use weights to early-fuse the descriptors using genetic algorithms or weighted functions. With both approaches, the process of designing a new CBIR for new datasets or domains involves a huge experimentation overhead, leading to multiple fine-tuned CBIR systems.
To overcome this experimentation effort, we propose a metaCBIR solution, a novel application of online learning with expert advice to create an ensemble of CBIR systems that allows us to dynamically converge to the best combination of systems, by taking advantage of user’s feedback. The resulting ensemble will be less dataset and domain dependent, while being able to take advantage of the experiments already performed to create the individual CBIR systems. Our solution is designed to be model-agnostic, modular, and scalable. Each CBIR in the ensemble returns its set of most similar images, that will be further combined in a dynamic fashion using weights to produce the final set of images similar to the given query image. The weight of each CBIR is updated based on the quality of its results, assessed by one or more human evaluators.
We conducted experiments on 13 benchmark datasets from the Biomedical, Real, and Sketch domains. metaCBIR was able to select the best CBIR sets across domains quickly (usually, less than 25 queries need to receive human feedback).

More information:

Transmissão em direto via Zoom.

Logótipo da Unite!

Summer School no âmbito da Rede UNITE! (University Network for Innovation, Technology and Engineering), de que faz parte a ULisboa.

Cartaz do evento

Uma iniciativa com o objetivo de aproximar a ciência dos jovens de todas as idades, com desafios, palestras e muita diversão, num formato inteiramente online!

Primeira edição das Manhãs de Matemática na Aberta, dedicada à Geometria, com a participação de Carlos Florentino (Ciências ULisboa). 

Hi-Phi Seminar, por Lorenzo Baravalle (FCUL, CFCUL) e Maurizio Esposito (FCUL, CIUHCT).

Imagem ilustrativa da sessão

Sessão de homenagem no âmbito da 13.ª Reunião Anual da APG - Associação Portuguesa de Geólogos.

Colóquio de Matemática, por Teresa Monteiro Fernandes (Departamento de Matemática | Ciências ULisboa).

O curso visa apresentar uma série de perspetivas sobre o desenvolvimento conceitual da estatística e sobre o uso de inferências e modelos estatísticos nas ciências empíricas.

Mapa antigo

Workshop internacional associado ao projeto Medea-Chart.

Imagem ilustrativa do curso, acompanhada de várias informações úteis

Um curso dirigido ao público em geral e a estudantes de arte ou ciência (a partir dos 16 anos), com interesse em desenho de observação, ilustração e biologia, com ou sem experiência em desenho.

The objective of this course is to introduce participants to the details of communication and writing scientific publications.

Imagem ilustrativa do programa, acompanhada do respetivo título

Um programa de aceleração de ideias e negócios, promovido pela redeSaúde, com candidaturas até 30 de setembro de 2021.

The objective of this course is to introduce participants to the details of communicating science to non-specialized audiences, including, but not exclusive to, public and private stakeholders, students and teachers, and media professionals.

This course aims to evidence the importance of natural history collections for the study of biodiversity, to show new tools and approaches to extract and disseminate biodiversity data from natural history collections and to increase awareness of young researchers for the scientific and cultural value of Natural History Museums.

The course aims at providing an introduction to stable isotopes ratios as tools to understand global, ecosystem and community level bio/geo cycles; light isotopes: H, N, C, O and S isotopes, geo cycles and ecological significance; sampling and analytical methodologies in stable isotope analysis.

EvoS aims at turning evolutionary theory into a common language to areas that pertain to the natural world, including human affairs.

The course SoilEco aims at introducing attendants to an updated state of the art of diversity of the soil biota and the functional roles played by soil organisms in key ecological processes.

In this course, we promote a multidisciplinary approach presenting the most recent findings on the topic and challenging the traditional way of considering symbiotic associations as exceptions and not as the rule.

The goal of this course is to provide to the participants with current and practical knowledge on urban ecology, including ecological and social aspects.

This course aims to explore ways of communicating science to non-specialized audiences, such as policy makers, industry, general public (including students and teachers), through their engagement and participation in citizen science activities.

This course introduces the field of island biogeography, a discipline that has long influenced other research areas such as macroecology, community ecology, evolution and conservation biology.

Scientists, as the main actors in the production of scientific knowledge, have the responsibility of having an active voice in communicating this knowledge – and that can be achieved for example by a conscious use of the communication tools at their disposal and a better articulation with journalists, among other means.

This course aims at providing students with basic knowledge of R programming, allowing them to manipulate and visualize data with R.

This course aims at providing students with statistical knowledge and tools to manipulate, analyze and visualize biological data with R. It also includes an introduction to modeling, simulations and Bayesian statistics.

Under the general framework of Global Change Ecology, the goal of this course is to provide the participants with the most recent and practical knowledge on the use of Functional Diversity.

The course provides essential skills and knowledge that enable the participants to develop climate change adaptation strategies.