News
Feedzai Research regularly publishes a variety of updates, interviews, talks, and other presentations. Think of this section as a series of post cards from our journey to continuously explore new ideas and technologies to shape the next generation of how we reinvent the fight against fraud and financial crime.
Ricardo Ribeiro gives a presentation about Transfer Learning at the DSPT meetup in Lisbon.
Ricardo Pereira presented "Hey, I've Seen This One!", where he showed how transfer learning techniques are applied to augment fraud detection systems. By leveraging knowledge across different markets and products, these methods help models adapt faster, spot attacks earlier, and reduce retraining costs. The talk covered practical considerations for deploying these techniques in production, as well as ongoing research that will shape the next generation of fraud prevention.
Ricardo Ribeiro presents our evaluation framework for transfer learning on ECML 2025.
Our AI team presents at the ECML PhD Forum.
Inês Silva, Ricardo Ribeiro, and Sérgio Jesus shared the work they have done for their PhD at the PhD Forum of ECML 2025.
Sérgio Jesus defended his PhD thesis at FCUP.
Sérgio Jesus successfully defended his PhD thesis at Faculdade de Ciências da Universidade do Porto, with the title “Robust Evaluation of Fairness and Explainability in AI Systems”.
The paper and dataset “SARSum: An Abstractive Summarization Dataset for Suspicious Activity Reports” was accepted in ECAI-2025.
The paper “High Probability Risk Control Under Covariate Shift” was accepted in COPA.
The paper “Evaluating Transfer Learning Methods on Real-World Data Streams” was accepted in ECML 2025.
The paper “A benchmarking framework and dataset for learning to defer in human-AI decision-making” was published in the Nature Scientific Data journal.
Hugo Ferreira was in a Workshop in The Global Anti-Scam Summit London 2025.
Hugo Ferreira participated in the “AI Against Scams” Workshop in The Global Anti-Scam Summit London 2025, together with speakers from Google, Bitdefender, ScamAdviser and PwC UK.
The paper “Mind the truncation gap: challenges of learning on dynamic graphs with recurrent architectures” was accepted in TMLR.
Vasco Pearson and Duarte Almeida defend their Master thesis at IST.
Vasco Pearson and Duarte Almeida both successfully defended their master thesis at Instituto Superior Técnico. Vasco's thesis title is "Uncertainty-Aware Systems for Human-AI Collaboration: Generative and Conformal Models in Dynamic, Resource-Constrained Environments", and Duarte's thesis title is "High-Probability Risk Control Under Covariate Shift".
Deep-Graph-Sprints: Accelerated Representation Learning in Continuous-Time Dynamic Graphs paper was accepted in TMLR
Show Me What’s Wrong!: Combining Charts and Text to Guide Data Analysis paper accepted at NLVIZ a workshop at IEEE VIS 2024
Aequitas Flow: Streamlining Fair ML Experimentation paper was accepted in the journal JMLR
Cost-Sensitive Learning to Defer to Multiple Experts with Workload Constraints paper was accepted in the journal TMLR
RIFF: Inducing Rules for Fraud Detection from Decision Trees paper was accepted at RuleML+RR 2024, the leading international joint conference in the field of rule-based reasoning
Fair-OBNC: Correcting Label Noise for Fairer Datasets paper was accepted at ECAIF 2024
AutiVizuA11y paper now available in the EuroVis’2024 proceedings
DiConStruct: Causal Concept-based Explanations through Black-Box Distillation
AutoVizA11y EuroVis’2024 presentation available in YouTube channel
Prof. Pedro M. Cruz (Northeastern University) starts part of his sabbatical at Feedzai
Cruz will be a visiting professor for a couple of months with the Data Visualization group
3rd edition of Women in Science Scholarships at Técnico now open
MSc applications from May 27 to Jun 21; PhD applications from May 27 to Jul 12
DiConStruct paper now available in the CLear2024 proceedings
DiConStruct: Causal Concept-based Explanations through Black-Box Distillation
Two patents granted
Patents on semantic AutoML and efficient RNN inference
Pedro Bizarro is in a panel @ AI Act – The Future is Now: Leveraging AI Innovation
Panel: Legal, Ethical and Economic Landscape of AI: Challenges and Strategies
The Women in Technology magazine ‘Trust in Tech’ features our accessibility work.
The magazine ‘Trust in Tech’ - EMEIA Financial Services is produced by EY
Release of Aequitas Flow
Aequitas Flow is the newest version of the Aequitas Bias Auditing & Fair ML Toolkit
DiConStruct paper accepted at CLeaR’2024
Explainability paper accepted at AAAI’2024
FiFAR dataset release
Three papers presented at ICAIF, the ACM International Conference on AI in Finance
AutoVizuA11y is available in open-source
The Feedzai Women Science Scholarships at Técnico awards
The Feedzai Women Science Scholarships at Técnico are back for the 2nd edition
One tutorial and two papers presented at KDD 2023
Data Viz Lisboa meetup by our Data Viz group, Jun 19, 6 PM
Accesibility in DataVis: Beyond the Visual, a talk by Rita Costa and Diogo Duarte
Montreal AI Ethics Institute highlights Feedzai Research paper
MAIEI highlights our ICML'2022 paper, "Human-AI Collaboration in Decision-Making: Beyond Learning to Defer"
FairGBM paper accepted at ICLR
FairGBM: Gradient Boosting with Fairness Constraints
Feedzai leads Portuguese organizations in European patent submissions
With 20 patent submissions, Feedzai tops the ranking for the 2nd year in a row
“Turning the tables” paper and fairness testing datasets presented at NeurIPS’2022
Paper, datasets, datasheets, and code are available
Our LaundroGraph algorithm hits the news
TechXplore.com posts news article on LaundroGraph
Winners revealed for the Feedzai Women in Science scholarships
Msc Scholarships: Mariana Lima, Ana Gutiérrez, and Isabel Portugal. Phd scholarship: Maria Osório
Feedzai releases FairGBM in open-source for non-commercial uses
Another year with multiple papers at KDD
"Lightweight Automated Feature Monitoring […]" @KDD-AutoML and "Understanding Unfairness in Fraud Detection […]" @KDD-MLF workshops
Two papers accepted at ICML workshops
“Human-AI Collaboration […] Learning to Defer”@ICML-HMCaT + “Prisoners of Their Own Devices: How Models Induce Data Bias […]”@ICML-RDMDE
Feedzai Women in Science scholarships at Técnico now open
The scholarships support women, students or future students, for either Msc or Phd programs at Instituto Superior Técnico.
Paper and poster accepted at EuroVis 2022
"Data+Shift: Supporting visual investigation of data distribution shifts by data scientists", helps to identify what changed and when.
Summer internships open
Vacancies in the Research department in Machine Learning, Systems Research and Data Visualization
Paper accepted at the ICLR 2022 PAIRStruct workshop
Our work "ConceptDistil: Model-Agnostic Distillation of Concept Explanations" presents an approach to bring concept explanations to any black-box classifier, being a deep learning model or gradient boosting.
Feedzai is the Portuguese organization with more European patent applications in 2021
After Feedzai, the other organizations in the top 5 are Universidade do Minho, SWORD Health, Delta Cafés, and Universidade do Porto
Open-source release of TimeSHAP
Feedzai releases TimeSHAP, a world's first (afawk) way of producing Shapley explanations for recurrent neural networks.
Feedzai Women in Science scholarships at IST
Feedzai and Instituto Superior Técnico announce the "Feedzai Women in Science scholarships at IST" to sponsor Msc and Phd work for women.
Paper accepted at the AAAI 2022 – AI in Financial Services workshop
"Anti-Money Laundering Alert Optimization Using Machine Learning with Graphs" accepted at the AAAI 2022 - AI in Financial Services workshop.
Page printed in 22 Oct 2025. Plase see https://research.feedzai.com/news for the latest version.