We wish to congratulate Prof. Dr. Álvaro Figueira and Dr. Bruno Vaz for winning the Mathematics 2022 Best Paper Award.

Name: Prof. Dr. Álvaro Figueira
Affiliation: CRACS-INESCTEC and Faculty of Sciences, University of Porto, 4169-007 Porto, Portugal
Research interests: Computer Science; Generative Adversarial Networks; Synthetic Data Generation; Disinformation Detection; Large Language Models.

Name: Dr. Bruno Vaz
Affiliation: Faculty of Sciences, University of Porto, Rua do Campo Alegre, s/n, 4169-007 Porto, Portugal; DareData Engineering, 1050-117 Lisbon, Portugal
Research interests: Large Language Models; Agentic Systems; Generative Adversarial Networks

The following is from an interview with Prof. Dr. Álvaro Figueira and Dr. Bruno Vaz:

1. Could you give a brief introduction of yourself to the readers? Could you introduce your current research direction and provide an update on your progress?

BV: I was a student at the Faculty of Sciences, University of Porto. I am currently an AI engineer working at a company from the private sector that builds custom AI-powered solutions. I work mainly with generative AI systems, such as chatbots.

AF: I graduated in 1995 with a degree in applied mathematics and computer science from the Faculty of Sciences, University of Porto. I earned an M.Sc. in advanced information technology from Imperial College London in 1997 and a Ph.D. in computer science from the University of Porto in 2004. Currently, I am a tenured Assistant Professor at the Faculty of Sciences, University of Porto. My research focuses on disinformation detection, social media analysis, and data visualization. As a researcher at the CRACS/INESCTEC unit, I have led international projects with institutions such as the University of Texas at Austin, University of Coimbra, and University of Aveiro, specializing in information extraction from social networks using machine learning and AI techniques.

2. Could you please briefly introduce the main content of the winning paper?

AF+BV: We wanted to provide an extensive review of synthetic data generation techniques, with a focus on Generative Adversarial Networks (GANs). The paper explores the need for synthetic data, especially in situations where real data is scarce, of poor quality, or involves privacy concerns, such as medical or financial data. We present various methods of generating synthetic data, both traditional and deep learning-based, with GANs being the primary focus due to their state-of-the-art performance in generating realistic data samples.

3. Could you describe the difficulties and breakthrough innovations in this research field?

AF+BV: The field of synthetic data generation and GANs faces several key challenges, including training instability, mode collapse, and difficulty generating diverse, high-quality data, especially for tabular datasets. Evaluating the quality of synthetic data remains difficult, as different applications require varying metrics, and hyperparameter tuning can be complex and resource-intensive. Moreover, privacy concerns arise when synthetic data generation risks revealing sensitive information from the original dataset.

Despite these challenges, breakthrough innovations have improved the performance and stability of GANs. Techniques like Wasserstein GAN (WGAN) and Progressive Growing GAN (ProGAN) have enhanced training stability and image resolution, while architectures like TGAN, CTGAN, and TabFairGAN have adapted GANs to generate high-quality tabular data. These innovations have significantly expanded the applicability and robustness of synthetic data generation.

4. What appealed to you about the journal that made you want to submit your paper? How was your experience submitting to Mathematics?

AF+BV: We were both drawn to Mathematics because of its reputation for efficiency and a well-structured review process. The journal ensures that reviewers are promptly engaged after submission, with a clear and expedited timeline—typically ranging from one week to a month, even during busy periods. This approach minimizes delays and keeps the process moving smoothly. We also appreciated the user-friendly submission system, which made navigating the various stages straightforward. The blind review process was particularly noteworthy for its clear and effective communication, allowing us to quickly understand and address the reviewers' feedback. Furthermore, the journal's strong impact factor and the consistently high quality of papers, especially in the Computer Science section, reinforced our decision to submit our work to Mathematics.

5. Which research topics do you think will be of particular interest to the research community in the coming years?

BV: In the last few months, LLMs have gained a lot of traction and have become more and more proficient, not only at natural language tasks but also in vision and audio. Even though the advances are astonishing, such models still suffer from hallucinations—they return content that, even though well-written, might not be factual. I believe this will be a clear topic of research in the coming years. Moreover, as AI systems become more autonomous, questions about ethical decision-making, accountability, and transparency will intensify. So, I believe research into AI fairness, bias mitigation, and explainability is expected to grow.

AF: I would also emphasize that the future likely holds a trend toward the specialization of LLMs in specific domains or for individual users, often referred to as personalized LLMs. This specialization could significantly enhance the relevance and effectiveness of AI interactions. Additionally, as LLMs become more tailored to individual needs, we can expect a growth in interactions between different LLMs, opening up new possibilities for making these exchanges mutually beneficial for both the LLMs and their users. This evolution could lead to more dynamic, personalized, and productive AI-driven conversations in the future.

6. Do you have any advice for aspiring young researchers looking to make a meaningful impact in their respective fields?

BV: More and more often I hear about FOMO (fear of missing out). With so many advances in generative AI, it is easy to feel like you are missing out. This has happened to me personally, so the advice that I would give would be to have JOMO (joy of missing out). If one can focus on a specific thing and do it well, I believe they would go a long way.

AF: In today’s fast-paced world of scientific advancement, it can be challenging to keep track of the latest developments and where the cutting-edge lies. To navigate this complex landscape, we rely on an enormous quantity of scientific publications that together with current advanced large language models help us map out the current state of knowledge. However, pushing beyond these boundaries still demands human effort, creativity, sensitivity, and insight. While these tools can guide us, it is ultimately our unique human qualities that drive innovation and meaningful progress.

7. As the winner of this award, is there something you want to express or someone you wish to thank most?

BV: Like Newton once wrote, "If I have seen further, it is by standing on the shoulders of Giants". If I was able to write this paper, it is because of the giants in my life. So, I would like to thank my parents, Albino and Natércia, and girlfriend, Rita, for being my pillars in life. Also, a special thanks to Professor Álvaro Figueira, who has guided me through the whole writing of this paper—without his help, I would still be floating on uncharted waters.

AF: The pursuit of methods to identify fake news led me to the creation of synthetic data (for balancing the datasets). While this was my initial focus, working with Bruno greatly expanded the scope of the research beyond my original goals. I am deeply grateful for his dedication and invaluable contributions to this project, and I want to express my heartfelt thanks for his commitment.

8. What is your opinion of the open access model of publishing?

AF+BV: Open access publishing can remove paywalls, allowing anyone with Internet access to read the published articles. This is particularly beneficial for researchers, students, and professionals in low-income or under-resourced institutions who cannot afford expensive journal subscriptions. Moreover, open access platforms often provide a faster route to publication and allow for the immediate dissemination of research, which is crucial in fast-moving fields like medicine, climate science, and technology.