Category Archives: Highlights

U.Porto students – the grand winners of the Sogrape Impact Hack

Posted on

“Reinventing the wine sector, through innovation, for a positive impact!” was the motto of the first edition of the Sogrape Impact Hack, a Hackaton organised by Sogrape that took place at Casa Ferreirinha on September 21st.

There were 12 hours of competition in which the 110 participants were divided into teams of 4 to 6 and, through collective intelligence and expert mentoring, came up with “concrete and creative solutions with the aim of contributing to the construction of a more impactful and sustainable world, focused on attracting new consumers, creating new products and implementing sustainable practices”.

The three winning teams, formed by students from the University of Porto, stood out for the innovative solutions they presented and were awarded a cash prize, 20% of which went to non-profit, non-governmental organisations selected by the teams themselves.

The “Techrocks” team, made up of Francisca Osório, Filipe Ferreira, Mário Branco (L.EIC) and Silvia Rocha (Alumna M.EIC), won the first place with their idea for an application that allows personalised recommendations using AI algorithms.

The “S4U” team took 2nd place with “Iniciativa Zero”, which focussed on creating a brand of innovative, low-alcohol products aimed at younger consumers. The team included Inês Arinto, João Gomes, Pedro Pinheiro and Matilde Ribeiro, former and current students on the Management Master’s programme at the Faculty of Economics.

“For the WINe”, a team made up of Dinis Sousa (M.EIC), Gonçalo Barros, Maria Helena Matos, Miguel Tomás Rodrigues (L.EIC) and Rodrigo Sousa, came 3rd with an app that rewards customers with discounts, prizes and privileged access to new products by sharing information and purchase history.

With the success of this first edition stays the promise of a return next year.

Photo: Sogrape

Sofia Vieira Pinto at the 10th edition of the Heidelberg Laureate Forum

Posted on

The 10th Heidelberg Laureate Forum took place in Germany September 24-29, bringing together some of the brightest minds in mathematics and computer science. At this conference, 200 carefully selected young researchers in mathematics and computer science, spend a week interacting with the laureates of these disciplines: recipients of the Abel Prize, ACM A.M. Turing Award, ACM Prize in Computing, Fields Medal, IMU Abacus Medal and Nevanlinna Prize. Established in 2013, the HLF is annually organized by the Heidelberg Laureate Forum Foundation (HLFF).

With a blend of scientific and social program elements, the HLF platform is especially designed to initiate exchange among the participants. Laureates give lectures on subjects of their choosing which are primarily directed at the participating young scientists. Those lectures should be the starting point of intensive discussions between the laureates and the young researchers during the forum. This means that the Forum is not a classical scientific conference but a networking event meant to motivate and inspire the next generation of scientists. Providing a space for ideas to take shape and evolve is what defines the Forum’s underlying purpose.

Among these young researchers we find Sofia Vieira Pinto, a third-year L.EIC student, who was selected to take part in this fervent meeting. “It was undoubtedly one of the most memorable weeks of my career so far”, tells us Sofia upon her return to FEUP. “There’s no way to describe in mere words all the contacts, conversations and even friendships that took place over the course of these days. I had the opportunity to make contact with numerous award-winners and listen to the life stories of those who already have many pages to tell. It was a period of learning, development and personal fulfilment such as I’ve never experienced. It still feels like a dream come true”, concludes our promising student.

PhD Defense in Informatics Engineering: ”Argumentation mining from text using semantic approaches”

Posted on

Candidate:
Gil Filipe da Rocha

Date, Time and Place
October 2, 14:00, Room Professor Joaquim Sarmento (G129), DEC, FEUP

President of the Jury
Rui Filipe Lima Maranhão de Abreu, PhD, Full Professor, Faculdade de Engenharia, Universidade do Porto

Members
Hugo Ricardo Gonçalo Oliveira, PhD, Associate Professor, Department of Computer Science, Faculdade de Ciências e Tecnologia, Universidade de Coimbra;
Bruno Emanuel da Graça Martins, PhD, Associate Professor, Department of Electrical and Computer Engineering, Instituto Superior Técnico, Universidade de Lisboa;
Eugénio da Costa Oliveira, PhD, Emeritus Professor, Department of Informatics Engineering, Faculdade de Engenharia, Universidade do Porto;
Sérgio Sobral Nunes, PhD, Associate Professor, Department of Informatics Engineering, Faculdade de Engenharia, Universidade do Porto;
Henrique Daniel de Avelar Lopes Cardoso, PhD, Department of Informatics Engineering, Faculdade de Engenharia, Universidade do Porto (Supervisor).

 Abstract
“The aim of argumentation mining is to automatically detect, identify and extract arguments from natural language text. The end goal is to provide a structured representation of the arguments (argument diagrams) that can be automated and analyzed in many different ways. Argumentation is a rhetorical act that has been studied for centuries and has been influenced by different research fields such as philosophy, linguistics, computer science, and artificial intelligence. In general, arguments are justifiable positions where pieces of evidence (premises) are offered in support of a claim (conclusion). Some characteristics of natural language text and, more specifically, of argumentation exposition make argumentation mining a complex task. Indeed, the ambiguity of natural language text, the assumption of commonsense reasoning and implicit knowledge, different writing styles, and the inherent complexity of argument diagrams are some of the challenges that argumentation mining systems have to overcome. Addressing these challenges, especially across different languages and text genres, demands robust argumentation mining systems. In this thesis, we conduct research toward the development of a robust computational system that can be employed to detect, identify, and extract argumentative content across different languages and text genres. Our vision is to deploy such a system to address argumentation mining in less-resourced languages (such as Portuguese) and in text genres that feature high-variability of argument exposition profiles (such as opinion articles). To this end, we combine techniques from computational linguistics and machine learning with knowledge from argument structure and rhetoric theories to automatically identify argumentative reasoning in natural language texts. To study argumentation mining in a less-resourced language and a challenging text genre, we conduct an annotation study to create a corpus annotated with arguments from opinion articles written in Portuguese. To address the challenging argumentation mining task, we propose a relation-based approach and context-aware models motivated by argumentation theory foundations and tailored to overcome some of the challenges of argument exposition. To tackle this task in a less-resourced language, we investigate how cross-language learning techniques can be employed to explore annotated resources from different languages and improve the performance of machine learning models in a target language. Finally, to improve the robustness of argumentation mining systems across different genres, we leverage recent advancements in language modeling capabilities and propose a cross-genre approach for argumentation mining.”

INForum 2023 – September 7/8 @FEUP

Posted on

The 2023 edition of INForum, now in its 14th edition, will be held at the Faculty of Engineering of the University of Porto on September 7 and 8, with the local organization by Jácome Cunha, João Paulo Fernandes, João Pascoal Faria and Rui Maranhão, Professors at DEI, and João Saraiva from the University of Minho.

Bringing together the national community, INForum is a privileged place for the dissemination, discussion and recognition of scientific work and innovation and technological advances in Computer Science. INForum thus offers a specialised stage to promote, on the one hand, the exchange of knowledge and experience between academia and industry and, on the other hand, the debut of young researchers looking for dissemination, constructive criticism and encouragement of their work. INForum is therefore a national event for sharing and strengthening community spirit.

Computer Science is a consolidated area of Research and Development in Portugal, supported by a network of internationally recognised research centres and the offer of undergraduate and postgraduate courses by practically all Portuguese higher education institutions. It is also an area in which several Portuguese companies present R&D results of international relevance.

INForum is organised in thematic sessions on topics proposed by the community and selected by the organisation. The topics have their own Programme Committees (PC), which liaise with the Chairs of the Programme Committee in the processes of organising the sessions (call for submissions, review and selection of papers, publication of proceedings, etc.) in order to take advantage of a single support infrastructure and guarantee the coherence of the event.

This edition will also feature keynotes from Cristina Videira Lopes (Chancellor’s Professor at the University of California, Irvine) and Pedro Saleiro (Senior Director of AI Research at Feedzai).

* O Fim da Programação (como a conhecemos)

“Cristina (Crista) Lopes é Chancellor’s Professor na School of Information and Computer Sciences at University of California, Irvine, com interesses de investigação em Linguagens de Programação, Engenharia de Software e Ambientes Virtuais Distribuídos. É IEEE Fellow e ACM Distinguished Scientist. Ela é a vencedora do Prêmio Pizzigati de 2016 para Software de Interesse Público pelo seu trabalho na plataforma de mundo virtual OpenSimulator. O seu livro ‘Exercises in Programming Style’ recebeu críticas excelentes, incluindo ter sido escolhido como ‘Livro Notável’ pelas revisões do ACM Best of Computing.”

** Innovating from within: AI Research at Feedzai

“Pedro Saleiro is Senior Director of Research at Feedzai where he heads the AI research group. Before joining Feedzai in 2019, Pedro did a postdoc in Fair Machine Learning at the University of Chicago and he was a research data scientist at the Center for Data Science and Public Policy working with Rayid Ghani. During his time at UChicago, he co-developed Aequitas, the first open-source library to audit bias and fairness of decision-making systems. Pedro holds a PhD in Machine Learning from University of Porto.”

The conference programme can be found here.

PhD Defense in Digital Media: ”Towards Human-in-the-Loop Computational Rhythm Analysis in Challenging Musical Conditions”

Posted on

Candidate:
António Humberto e Sá Pinto

Date, Time and Place:
September 8, 14:30, Sala de Atos FEUP

President of the Jury:
António Fernando Vasconcelos Cunha Castro Coelho, PhD, Associate Professor with Habilitation, Faculdade de Engenharia da Universidade do Porto;

Members:
Magdalena Fuentes, PhD, Assistant Professor, Music and Audio Research Lab (MARL) and Integrated Design & Media (IDM), New York University (NYU);
Jason Hockman, PhD, Associate Professor, School of Computing and Digital Technology (DMT), Birmingham City University (UK);
Matthew Edward Price Davies, PhD, Senior Scientist,  SiriusXM/Pandora (USA) – (Supervisor);
Rui Pedro da Silva Nóbrega, PhD, Assistant Professor, Departamento de Informática, Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa;
Aníbal João de Sousa Ferreira, Associate Professor, Departamento de Engenharia Eletrotécnica e de Computadores, Faculdade de Engenharia da Universidade do Porto.

The thesis was co-supervised by Prof Rui Luís Nogueira Penha, Coordinating Professor at ESMAE, and Prof Gilberto Bernardes de Almeida, Assistant Professor at FEUP.

Abstract:

“Music Information Retrieval (MIR) is an interdisciplinary field focused on the extraction, analysis, and processing of information from various musical representations.
Grounded on the automatic analysis of musical facets such as rhythm, melody, harmony, and timbre, MIR enables applications in areas like music recommendation, automated music transcription, and intelligent music composition tools. Rhythm, an integral element of music, provides a foundation for decoding music’s complex relational structures and layered depth. Computational rhythm analysis is thus central to MIR research. It encompasses a wide range of tasks, such as the pivotal beat tracking, which unlocks the use of musical time across many MIR systems. However, conventional beat-tracking methods have struggled when dealing with complex musical features, such as expressive timing or intricate rhythmic patterns. While specialized approaches demonstrate some degree of adaptation, they do not generalise to diverse scenarios. Deep learning methods, while promising in addressing these issues, depend heavily on the availability of substantial annotated data. In scenarios requiring adaptation to user subjectivity, or where acquiring annotated data is challenging, the efficacy of beat-tracking methods lowers, thus leaving a gap in the applicability of computational rhythm analysis methods. This thesis investigates how user-provided information can enhance computational rhythm analysis in challenging musical conditions. It initiates the exploration of human-in-the-loop strategies with the aim of fostering adaptability of current MIR techniques. By focusing on beat tracking, due to its fundamental role in rhythm analysis, our goal is to develop streamlined solutions for cases where even the most advanced methods fall short. This is achieved by utilising both high-level and low-level user inputs —- namely, the user’s judgement regarding the expressiveness of the musical piece and annotations of a brief excerpt —- to adapt the state of the art to abstract particularly demanding signals. In an exploratory study, we validate the shared perception of rhythmic complexity among users as a proxy for musical expressiveness, and consequently as a key performance enhancer for beat tracking. Building upon this, we examine how highlevel user information can reparameterise a leading-edge beat-tracker, augmenting its performance to highly expressive music. We then propose a transfer learning method that finetunes the current state of the art, hereafter referred to as the baseline, to a concise user-annotated region. This method exhibits versatility across varied musical styles and offers potential solutions to the inherent limitations of previous approaches. Incorporating both user-guided contextualisation and transfer learning into a human-in-the-loop workflow, we undertake a comprehensive evaluation of our adaptive techniques. This includes examining the key customisation options available to users and their effect on performance enhancement. Our approach outperforms the current state of the art, particularly in the challenging musical content of the SMC dataset, with an improvement over the baseline F-measure of almost 10 percentage points (corresponding to over 16%). However, these quantitative improvements require further interpretation due to the inherent differences between our file-specific, human-in-the-loop technique and traditional dataset-wide methods, which operate without prior exposure to specific file characteristics. With the aim of advancing towards a user-centric evaluation framework for beat tracking, we introduce two novel metrics: the E-Measure and Annotation Efficiency. These metrics account for the user perspective regarding the annotation and finetuning process. The E-Measure is a variant of the F-measure focused on the annotation correction workflow and includes a shifting operation over a larger tolerance window. The Ae is defined as the relative (to the baseline) decrease in correction operations enabled by the fine-tuning process, normalised by the number of user annotations. Specifically, we probe the theoretical upper bound of beat tracking accuracy improvement over the SMC dataset. Our results show that the correct beat estimates provided by our approach surpass those of the state of the art by more than 20%. When considering the full length of the files, we can further frame this improvement in terms of gain per unit of user effort, quantifying the annotation efficiency of our approach. This is reflected in the substantial reduction of required corrections, with nearly 2/3 fewer corrections per user annotation compared to the baseline. In the final phase, we evaluate our human-in-the-loop strategy’s adaptability across a range of musical genres and instances presenting significant challenges. Our exploration extends to various rhythm tasks, including beat tracking, onset detection, and (indirectly) metre analysis. We apply this user-driven strategy to three unique genres with complex rhythm structures, such as polyrhythms, polymetres, and polytempi. Our approach exhibits swift adaptability, enabling efficient utilisation of the state-of-the-art method while bypassing the need for extensive retraining. This results in a balanced integration of data-driven and user-centric methods into a practical and streamlined solution.”

Keywords: Music Information Retrieval; User-centric; Transfer Learning; Beat Tracking.

PhD Defense in Informatics Engineering: ”Scaling-up organization of document sets to facilitate their analysis”

Posted on

Candidate:
Rui Portocarrero Macedo de Morais Sarmento

Date, Time and Place:
July 24, 14:00, Sala de Atos DEGI (L202A), FEUP

President of the Jury:
Carlos Manuel Milheiro de Oliveira Pinto Soares, PhD, Associate Professor, Departamento de Engenharia Informática, Faculdade de Engenharia da Universidade do Porto.

Members:
José Fernando Ferreira Mendes, PhD, Full Professor, Departamento de Física, Universidade de Aveiro;
Bruno Emanuel da Graça Martins, PhD, Associate Professor, Departamento de Engenharia Electrotécnica e de Computadores, Instituto Superior Técnico da Universidade de Lisboa;
Pavel Bernard Brazdil, PhD, Emeritus Professor, Faculdade de Economia, Universidade do Porto (Co-Supervisor);
Henrique Daniel de Avelar Lopes Cardoso, PhD, Associate Professor, Departamento de Engenharia Informática, Faculdade de Engenharia da Universidade do Porto;
Sérgio Sobral Nunes, PhD, Associate Professor, Departamento de Engenharia Informática, Faculdade de Engenharia da Universidade do Porto.

The thesis was supervised by João Manuel Portela da Gama, PhD, Full Professor at Faculdade de Economia da Universidade do Porto.

Abstract:

“The summarization and organization of document production of an organization in an intuitive and scalable way for massive amounts of data is of great importance in supporting decision-making.

This thesis intends to develop a theoretical and practical study to solve these challenges. The contents of this thesis were born after developing a static software prototype to analyze and provide decision support from text documents and a network of authors of scientific documentation. Several advantages were proved from the use of this mentioned prototype. Nonetheless, there were some concerns regarding the prototype’s ability to cope with higher dimensional networks and also a massive amount of documents. The development case study considers the affinity between authors on a large scale and constantly evolving. The first challenge is to scale the representation methods of documents of the authors. The second challenge is to capture the temporal development of the organization. Considering this context, we developed and implemented streaming techniques for the characterization of each author and other sub-units of the organization. Thus, by integrating into affinity groups identified by keywords and relevance measures that characterize them. We have finished this work by testing several developed algorithms to minor the disadvantages of the original prototype and gathering a panoply of solutions for problems related to text streaming techniques, considering a large-scale approach for the corresponding analysis. Information Retrieval techniques were used, and the analysis of social networks and streaming data was necessary. We solved several associated issues with efficient text streams analysis, using several techniques from pure streams analysis techniques to evolving complex networks techniques. These techniques that served as a base to innovation and contribution with more than ten new algorithms proved to improve the prototype and solve the issues that initially drove us to improve and contribute to several related areas of text analysis and streams.”

keywords: Streaming; Text Mining; Social network Analysis; Social network Visualization.

We are pleased to announce that ASAP 2023 will be an entirely in-person event hosted at the Faculty of Engineering of the University of Porto!

Posted on

The 34th IEEE International Conference on Application-specific Systems, Architectures, and Processors (ASAP 2023) is organized by the Faculty of Engineering of the University of Porto in Porto, Portugal, July 19 – July 21, 2023.

The history of the ASAP conference traces back to the International Workshop on Systolic Arrays, organized in 1986 in Oxford, UK. It later developed into the International Conference on Application-Specific Array Processors. With its current title, it was organized for the first time in Chicago, USA, in 1996. Since then, it has alternated between Europe and North America.

PhD Defense Digital Media: ”Connect-the-Dots: Artificial Intelligence and Automation in Investigative Journalism”

Posted on

Requested by:
Joana Rodrigues da Silva

Date, time and place
July 19, 14h30, room L119 DEMEC (FEUP)

President of the Jury:
António Fernando Vasconcelos Cunha Castro Coelho, PhD, Associate Professor with Habilitation, Departamento de Engenharia Informática, Faculdade de Engenharia da Universidade do Porto.

Members:
Teresa Isabel Lopes Romão, PhD, Associate Professor, Departamento de Informática, Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa;
Luís António Santos, PhD, Assistant Professor, Departamento de Ciências de Comunicação, Instituto de Ciências Sociais da Universidade do Minho;
Miguel Ângelo Rodrigues Midões, PhD, Invited Adjunct Professor, Departamento de Comunicação e Arte, Escola Superior de Educação do Instituto Politécnico de Viseu;
Helena Laura Dias de Lima, PhD, Associate Professor, Departamento de Ciências da Comunicação e da Informação, Faculdade de Letras da Universidade do Porto (Supervisor);
Alexandre Miguel Barbosa Valle de Carvalho, PhD, Assistant Professor, Departamento de Engenharia Informática, Faculdade de Engenharia da Universidade do Porto.

Abstract:

“After the COVID-19 epidemic and the consequent humanitarian crises that devastated the planet, there is a need to claim the role of investigative journalism as a watchdog and permeation system of social justice and democracy through public exposure. We are witnessing a sharp decrease in investment in this journalism specialty, either because of its impertinence in dealing with public management issues or because of the time spent on this type of investigation, which fundamentally takes longer than current journalism to produce results. In this sense, we perceive the influence of automation and artificial intelligence in information production processes to highlight all human tasks with the possibility of being carried out in less time by technological systems. Considering this possibility, there was an interest in studying, in-depth, how robotics and the application of artificial intelligence through platforms that support the usual procedure of journalism can help and even improve the global state of investigative journalism practice, nowadays. The Connect-the-Dots platform and the DODO assistant emerge as a hypothetical digital solution for some of the problems that investigative journalism currently faces. It could be a way to practically implement investigative journalism in its scope of innovation by integrating tools and open-source practices in a Design-Based-Research approach to knowledge archaeology.”

Keywords: Artificial Intelligence, Automation, Investigative Journalism, Design-Based-Research, Digital Media.

PhD Defense in Digital Media: ”Modelo para utilização da prosódia e da interacção no acesso às expressões matemáticas através da fala sintetizada para pessoas com deficiência visual”

Posted on

Candidate:
Adriana Silva Souza

Date, time and place:
July 10, 10:30,  Sala de Atos FEUP

President of the Jury:
António Fernando Vasconcelos Cunha Castro Coelho, PhD, Associate Professor with Habilitation, Faculdade de Engenharia da Universidade do Porto.

Members:
Vitor Manuel Pereira Duarte dos Santos, PhD, Assistant Professor, NOVA Information Management School, Universidade Nova de Lisboa;
João Manuel Pereira Barroso, PhD, Associate Professor with Habilitation, Vice-Reitor para a Inovação, Transferência de Tecnologia e Universidade Digital, Universidade de Trás-os-Montes e Alto Douro;
João Paulo Ramos Teixeira, PhD, Coordinator Professor, Departamento de Eletrotecnia, Escola Superior de Tecnologia e Gestão do Instituto Politécnico de Bragança;
Maria Selene Henriques da Graça Vicente, PhD, Assistant Professor, Departamento de Psicologia, Faculdade de Psicologia e de Ciências da Educação da Universidade do Porto;
Maria do Rosário Marques Fernandes Teixeira de Pinho, PhD, Associate Professor, Departamento de Engenharia Eletrotécnica e de Computadores, Faculdade de Engenharia da Universidade do Porto;
Diamantino Rui da Silva Freitas, PhD, Associate Professor, Departamento de Engenharia Eletrotécnica e de Computadores, Faculdade de Engenharia da Universidade do Porto (supervisor).

Abstract:

“The synthesized speech of mathematical contents still presents several challenges. For Mathematics to be understood by people with visual impairment, it needs to be verbalized in detail, which generates long outputs and causes cognitive overload; in addition, Mathematics has quite peculiar rules. Therefore, most of the time, prosodic limits such as pauses, and intonation are not adequately synthesized. This investigation proposes a model that uses prosody and interaction to access mathematical expressions to minimize the problems mentioned. We relied on the Design-Based Research methodology to develop the model and divided the study into four stages. In the first stage, a systematic literature review was carried out. We conducted an initial exploration investigation with interviews with students with visual impairments and braille teachers and analyzed the mathematics spoken by speech synthesizers. In the second stage of the investigation, a corpus of mathematical expressions spoken by math professors was created to support prosody research. Intonation and pauses were the investigated prosodic components. Although the studies have yet to go into intonation in-depth, we did some tests of prosodic modulation of the fundamental frequency, highlighting stretches of mathematical expressions according to the level in the MathML tree. Concerning pauses, we identified their main patterns in mathematical expressions. We also performed an eye-tracking experiment with sighted people to understand the cognitive processes surrounding mathematical expressions’ observation, analysis, and processing. In the third stage, a linear regression model which calculates the pauses for mathematical expressions dynamically was created and evaluated with visually impaired students. The results showed advances regarding the solutions found, perceived mainly when the mathematical expressions are unfamiliar to the students. The results of the eye tracking experiment showed that in addition to the complexity of the mathematical expression, it was necessary to propose a new formal concept that was called diversity, quantifying this subjective property of the structures of expressions because it was found that it also impacts during the cognitive processing of expressions. Data analysis provided clues for creating the interaction model that uses diversity to control the cognitive load in accessing mathematical expressions during the process. The evaluation of the model with visually impaired people showed an advance concerning existing works since students performed better when accessing mathematical expressions with the model. In the fourth stage, we made the final proposition of the model based on the assessment of people with visual impairments. The results achieved in this investigation allow greater autonomy in the reading of mathematical expressions; people with visual impairment can govern the interaction in the auditory access according to the need to reinforce their memory; in addition, it can reduce the time in the manipulation of mathematical expressions compared to traditional tools, improve the writing process, since reading is linked to writing and relieve the student’s memory. In addition to these contributions, we can also highlight the discovery of the new diversity parameter, which is strongly related to the cognitive processing of expressions. In general, these contributions make it possible to improve and develop mathematics education, particularly in the teaching-learning process of visually impaired people, making them more autonomous beings, which, in addition to scientific contributions, can also generate social and economic impacts arising from accessibility.”

Keywords: Synthesized Speech, Mathematics, Accessibility, Visual Impairment, Complexity, Diversity.

DEI TALKS | “From Numerical Libraries, To Efficient Matrix Multiplication Compiler-Only Code Generation, To a Modular Automated General Packing Data Transformation” by Prof. J. Nelson Amaral

Posted on

By the author:

“To support both Artificial Intelligence and High-Performance Computing workloads, new processors have introduced hardware acceleration for matrix multiplication. Examples include the Matrix Multiply Assist (MMA) in the IBM POWER10 and the Advanced Matrix Extensions (AMX) in the Intel Sapphire Rapids microarchitecture for Xeon servers. This talk describes how, in a collaboration between the University of Alberta, the University of Campinas, and IBM, we developed compiler technology to support such accelerators. An initial solution delivered a robust pattern matcher for General Matrix Multiplication (GEMM) computation operating at the LLVM intermediate representation that allows the replacement of the computation with an invocation of a high-performance library. A later solution delivered a compiler-only path for code generation by adapting the layered approach used in numerical libraries to the compiler code-generation process. Finally, a modular and automated general strategy for data packing, which can be applied to multiple algorithms, was developed for the Multi-Level Intermediate Representation (MLRI).”

“From Numerical Libraries, To Efficient Matrix Multiplication Compiler-Only Code Generation, To a Modular Automated General Packing Data Transformation” will be presented July 17, at 11:30, room B006, moderated by Prof. Pedro Diniz (DEI) and co-organized by DEI Talks and the University of Porto – Faculty of Engineering ACM Student Chapter.

“J. Nelson Amaral, a Computing Science professor at the University of Alberta with a Ph.D. from The University of Texas at Austin, has published in optimizing compilers and high-performance computing. Scientific community service includes general chair for the 23rd International Conference on Parallel Architectures and Compilation Techniques in 2014, for the International Conference on Performance Engineering in 2020, and for the International Conference on Parallel Processing in 2020. Accolades include ACM Distinguished Engineer, IBM Faculty Fellow, IBM Faculty Awards, IBM CAS “Team of the Year”, awards for excellence in teaching, the University of Alberta Graduate-Student Association Award for Excellence in Graduate Student Supervision, an University of Alberta Award for Outstanding Mentorship in Undergraduate Research & Creative Activities, and a recent University of Alberta 2020 COVID-19 Remote Teaching Award.”

https://webdocs.cs.ualberta.ca/~amaral/