Anúncios
Pitch Workshop - Tuesday 6 and 13 July 9h00
5 julho 2021, 07:51 • Ana Almeida Matos
We will have two sessions of our Pitch Workshop:
Tuesday 6 July 9h00
Title: Search-Oriented Conversational Assistant
Presenter: Gonçalo Raposo
Abstract:
State-of-the-art dialogue models often produce factually inaccurate responses. Transformers may be fine-tuned for tasks such as response generation, and are able to produce fluent and well-written results, due to the very large amount of text they are exposed to during pre-training. However, generated responses tend to suffer from factual incorrectness and knowledge hallucination. These problems often arise because the models only consider the given conversation, and thus any knowledge present in the generated response comes implicitly from the model parameters. This work aims to introduce a retrieval step that will search for passages related to the given utterance and explicitly use them to generate a response. The PEGASUS model, i.e. a state-of-the-art Transformer for text summarization, is fine-tuned to address answer generation as a task of summarizing the retrieved passages, conditioned on the current conversation. A few conversational datasets are considered for experiments, as well as a community support dataset, in order to evaluate the system in a customer support scenario. The obtained results show that the system is able to make use of the retrieved knowledge to generate consistent and factually accurate responses. Moreover, by relying on a retrieval stage, the system also provides more interpretable responses.
Title: Towards Dataset Comparability: An Approach based on User Behavior
Presenter: João Góis
Abstract:
A current concern in today's society is to mitigate the risk of global climate change. Although there have been several initiatives to achieve more sustainable worldwide energy distribution, improper energy use remains an issue. In this work, a new methodology is proposed for detecting and analyzing energy consumption in buildings. For illustration, the methods are applied to some appliances of the REFIT dataset. The proposed methods enable a straightforward and rigorous distinction of different consumption patterns and, consequently, the definition of user profiles for each building throughout the seasons of the year.
Title: Learning prognostic biomarkers from three-dimensional biomedical data of psychiatric disorders
Presenter: Leonardo Duarte Rodrigues Alexandre
Abstract:
The number of patients diagnosed with a mental disorder (depression, attention deficit and hyperactivity disorder, anxiety, bipolar disorder, and/or schizophrenia) is considerably rising. Due to the isolation of the population during the pandemic, this number is especially high. Thus, the need for prognostic markers is essential to place appropriate diagnostics and treat patients with the appropriate therapies in accordance with their unique neurobiological profile. This treatment can be critical to prevent morbidity and, in some cases mortality. Despite being essential to diagnose these patients correctly, this task is hampered due to many symptoms overlapping between mental disorders. Thus, my PhD’s aim is to develop machine learning approaches to identify prognostic biomarkers in psychiatric disorders and support therapeutic choices from cohorts with available neuroimaging, cognitive and molecular data. This type of data is typically consolidated using a three-dimensional tensor representation with the dimensions being patients-variables-time. Techniques such as triclustering and temporal pattern mining, which remain largely unexplored within mental disorder research papers, will be used to explore this three-dimensional space to discover meaningful patterns. We will then proceed to: 1) understand the extent to which three-dimensional patterns discovered from neurobiological data assist the understanding of complex neurophysiological relationships, such as boundaries and overlaps between mental disorders, to better understand disease progression, and response to stimuli after drug admission, 2) assess the impact that different methods have in finding, classifying, and exhaustively searching the three-dimensional space for meaningful biomarkers, as well as provide statistical tests to guarantee the statistical significance of the discovered biomarkers, 3) extend the proposed machine learning approaches towards predictive tasks as to place new diagnostics, prognostics, and therapy recommendations for new patients, using their neurobiological profile against the found patterns.
Title: Persistent Memory for Data-intensive applications
Presenter: Ilia Kuzmin
Abstract:
Real-world applications have complex constraints on the hardware they run on. Many of them require intensive computations to process large amounts of data and to achieve decent performance enough resources (like CPU and RAM) should be supplied. Yet any resources go for a cost, and finding optimal configuration could be a challenge by itself. Furthermore, having millions of source code lines exist, it is practically impossible to adjust each of them to use cutting-edge technology features, thus abstraction layers (like Operations Systems) should provide an opportunity to employ full hardware capacity transparently.
My current work is focused primarily on non-uniform memory access technologies. In particular, it focused on incorporating large amounts of cheap, energy-efficient, yet slow random access memory to the data-intensive applications on the system level, to enable performance boost without changing particular application implementation.
Tuesday 13 July 9h00
Title: Cooperation Dilemmas on Imperfect Information in Hybrid Populations
Presenter: Henrique Fonseca
Abstract:
The mechanism of Indirect Reciprocity (IR) provides an elegant solution to the cooperation dilemma by arguing that reputations and social norms are core elements of human social decision making. This has been studied in the fields of ecology, psychology or economy, both mathematically or computationally. However, little has been researched regarding the dynamics forced by imperfect information, i.e. when agents have diversified opinions on the same matters. Moreover, those that tackle this problem often assume that individuals have binary non-null opinions on all agents in a population, something that tends not to scale well with population sizes. Here I tackle the problem of IR with imperfect information by changing the commonly used computational models to include a third reputation besides Good and Bad: the Unknown reputation. This leads to new unexplored dynamics in Evolutionary Game-Theory based simulations capable of tackling questions such as: How to regard strangers for cooperation to arise?; What are the roles of gossip, empathy or social conformity in providing consensus to chaos driven populations?; and What are the impacts on information dissemination of having different cognitive capabilities?
Title: Enforcing GDPR Compliance Through In-Depth Tracking of Personal Information on Websites
Presenter: Mafalda Ferreira
Abstract: Modern web applications provide many useful services to end-users which require them to blindly share their data. In 2018, the European Union issued the GDPR, a comprehensive legislation that defines a system of laws aimed at promoting the deployment of extensive security mechanisms for the protection of users’ data and prevention of privacy breaches. Unfortunately, most modern systems tend to be optimized for performance, cost, and reliability, leaving security as a secondary goal. As a result, not only the web users remain prone to numerous risks, including the exposure of sensitive data, but the organizations themselves may incur high fees in the case of non-compliance with the GDPR. My current work studies the implications that GDPR holds in web applications and clarifies the requirements organizations need to follow when managing their information systems. Particularly, I am developing RuleKeeper, a web application framework, tailored to provide data security and privacy protections according to GDPR-compliant policies.
Presenter: João Gonçalves
Abstract:
Persistent Memory (PM) is a new storage technology that promises access latency close to DRAM while guaranteeing the persistence of data across program restarts.Unfortunately, this technology offers limited support for atomic writes.Unlike DRAM, whose state is completely reset after an application or machine crash, PM state is kept, and dirty (partial) reads might still be observed upon recovery, resulting in inconsistent post-failure executions.This behaviour motivated the emergence of PM crash-consistency testing in the literature.Nevertheless, the state-of-the-art is limited by long execution times, dependence on manual annotations and the lack of support for multithreaded applications.The latter is relevant given that a system that is both crash-consistent in single-threaded execution and data-race free might still exhibit so-called persistency races in specific thread interleavings. My current work focuses on this particular topic, aiming to develop a testing tool that detects persistency races in PM applications in a scalable manner.
Presenter: José Bernardo Rocha
Abstract:
Procedural content generation is a popular topic in the games industry and research field, as it allows for faster development of content at reduced cost. Additionally, it can support artists and game developers in the creation of more diverse and realistic content, and promotes replayability in games as it potentially generates infinite and novel content for players to explore. However, in terms of generating cooperative content, and, specifically, content that requires strong collaboration between both players, there is not much work and results. In this paper, we propose a solution to procedurally generate cooperative levels for the cooperative game Geometry Friends. The generator makes use of genetic algorithms for the core part of generating the layout of the levels and characters' positions. The results show that it is possible to create levels with the approach providing as input a set of restrictions on the areas of the level that should be reached by each character.
Cross-Seminars - 2 July 16h30 - 4x short on climate change, bioprocesses and cells
1 julho 2021, 10:29 • Ana Almeida Matos
SEMINAR 3 – 2 July / 16.30h
This session will be composed of 4 short seminars:
Tittle: What (really) drives corporations to issue debt that has to be applied in environmental-friendly projects?CANCELLED
Presenter: Rodrigo Graça, PhD student
Abstract:
Climate change is one of the greatest challenges of our contemporary society. The modifications of Earth’s climate carries potentially high economic and social costs. These consequences can be significantly mitigated, but only with a fast and extensive transition of the current economic system towards a greener and less carbon-intensive model, as urged by the Intergovernmental Panel on Climate Change (IPCC) of the United Nations in their most recent report. This ambitious endeavour is solely feasible with the involvement of both public and private sectors. Positive Impact Fixed Income products, such as the Green Bonds are valuable tools to mobilize and direct private funds into projects that actively contribute to a more sustainable economy, and hence complementing and enhancing the actions conducted by Governments towards this direction. These products have been reporting a fast growth since their emergence in 2007, suggesting the priorities’ redefinition by investors and consumers towards higher accountability for environmental, social and governance (ESG) aspects. The PhD research proposal is still going to be developed, although it is expected to aim around the following questions. First, how effective are these financial products concerning their impact on the environmental and social factors that they were designed to achieve? Second, what is added value for the issuers and the investors that include these products in their portfolios?- Title: Use of Dynamic Optimization Algorithms for the development of bioprocesses
Presenter: João Antunes
Abstract:
The dynamic optimization or open-loop optimal control of a bioprocess is a procedure by which the optimal control variables are ascertained through the application of an optimization algorithm to a given model of the bioprocess in question, usually described in form of differential and algebraic equations (DAEs) and subsequent constraints on state and control variables. This optimization procedure is an important tool that can give a better understanding of a process as its complex, and usually non-linear, nature can obscure the procedure by which the processes optimal production might be reached. I study the effect of different optimization methods applied in the development of hybrid semiparametric modelling to develop a batch-to-batch control of a given biological process. - Title: A novel symbolic framework for hybrid semiparametric modelling of bioprocesses
Presenter: José Pinto
Abstract:
In this project a hybrid modelling toolbox to integrate machine learning and mechanistic biological models was developed in MATLAB/OCTAVE. Symbolic calculus is adopted to “fuse” physical mathematical equations with machine learning equations. Furthermore, this tool obeys the Systems Biology Markup Language (SBML). The compliance with the SBML standard enables it to easily adapt existing mechanistic SBML models to hybrid semiparametric versions. In addition, the storage of hybrid SBML models in public databases becomes possible. - Title: Dynamic models describing cell growth, central carbon metabolism and virus production in animal cells?
Presenter: João Ramos, SBE research group, NOVA University Lisbon
Abstract:
Currently large number of datasets are being generated and more complex, leading to a need for model-based data integration and analysis for process optimization. However, currently few models exist that describes both cell growth and intracellular metabolism of animal cell lines using dynamic mechanistic models.
We have developed dynamic mathematical models of cell growth, virus production and metabolism and show their applications for different cell lines. In this approach our models use ordinary differential equations to simulate changes in viable cell concentration, and volume, virus titer, concentration of extracellular substrates, and intracellular concentrations of key metabolites from the central carbon metabolism.
We show that such models allow accurate estimation of the release of metabolic by-products such as lactate and ammonium directly from the intracellular reactions and model simulations hints at the existence of distinct cellular physiological states. Furthermore, we use this approach to study metabolism during virus replication and showed that there is no significant changes comparing metabolism of infected and non-infected cells. Finally, we demonstrate the usage of this modeling approach for media optimization.
Link to entire session: https://videoconf-colibri.zoom.us/j/89214599997
Cross-Seminars - 28 June 16h - Efficient Pricing for Algorithmic Trading systems
25 junho 2021, 11:02 • Ana Almeida Matos
SEMINAR 2 – 28 June / 16h - 16.45h
Title: Efficient Pricing for Algorithmic Trading systems
Presenter: Alexander Toropov, lead architect at Itiviti a Broadbridge Business
Join by SIP: 93484389197@zoomcrc.com
Join by Dial Up: Meeting ID: 934 8438 9197
Passcode: 319970
Find a local number: https://itiviti.zoom.us/u/agxPd1DKl
Abstract:
Fintech is a progressive area that combines cutting edge technologies and methods from several different realms: from software and hardware development for distributed latency-critical systems to quantitative modelling using advanced mathematical approaches.
This talk is going to introduce to the audience several basic challenges in the derivative pricing task and is going to follow the next structure.Introduction to derivative instruments: Let us start with a brief introduction to derivative contracts considering an example of future and option. What properties do the contracts have and how those contracts are used in the real world.
Pricing of derivative contracts: Valuation of derivative contracts is a complex question. Starting with basic methods and coming to the most complex and advanced ones which are used in Tbricks trading platform.
Application of pricing: Pricing of derivatives contracts is a very powerful tool where advanced trading strategies can be used to make a better decision on what to buy and sell and when. Let us discuss a few of them like Market Making or Volatility trading.Speaker details on LinkedIn.
Cross-Seminars - 21 June 16h - Conceptual and technical structure of the exhibition: A THIRD REASON
21 junho 2021, 13:47 • Ana Almeida Matos
We will be announcing here the program of the Cross-Seminars series, organized by the students. The keynote for the seminars is that they share research work, on different topics, that either cross disciplinary boundaries or use techniques from one area to solve problems in another area.
The first one will take place today. You are very welcome to attend. Stay tuned for the following ones!
SEMINAR 1 – 21 June / 16h00 - 16h45
Title: Conceptual and technical structure of the exhibition: A THIRD REASON
Speakers: Alexandre Estrela, Borja Caro
Abstract:
Alexandre Estrela’s work is an investigation on the essence of images that expands spatially and temporally through different supports. In his videos and installations Estrela examines the subject’s psychological reactions to images in their interaction with matter. Each piece has several layers to which we are initiated step by step. The works are not just there to be watched, but rather to be unfolded. Each piece convokes synesthetic experiences, visual and sound illusions, aural and chromatic sensations that function as perceptive traps, leading the subject towards conceptual levels. With this strategy Estrela is constantly problematising the elements that constitute the act of perceiving, splitting vision into further sensible dimensions towards the unseen and the unheard. His work has been widely exhibited nationally and internationally: https://travesiacuatro.com/eng/artista/alexandre-estrela-2/
Borja Caro focuses in the composition of experimental electronic music. During his time at the University of the Arts London he presented his work at the Digital Design Drop in the Music Hackspace at the Victoria and Albert Museum. Back in Madrid, among several projects, he worked in the organization and promotion of “unconventional” musical events” and also released two albums. Many of his live presentations also took place internationally. Now in Lisbon, he has been collaborating with the artist Alexandre Estrela, for whom he develops interactive audiovisual systems and sound design in the context of the artist’s own work.
Course requirements
16 março 2021, 12:49 • Ana Almeida Matos
Dear all,
Following our first meeting, the specification of the requirements for completing this course is published in Section Método de Avaliação/Evaluation Methods.
Further details will from now on be handled via our Slack workspace. Please join! https://join.slack.com/t/interdiscipli-htj4811/shared_invite/zt-orkawxfq-UUdfwWZ0OXB_MHyMqjXeZg
Ana