Will computer science Ph.D. students and professors soon be replaced by “thinking machines”? - I did my Ph.D. in Innsbruck in 1966 (with the invention of the algorithmic theory of “Gröbner Bases”), and I was a professor most time of my life (at the Johannes Kepler University in Linz). In 2004, I managed to generate the algorithmic idea of Gröbner bases by an algorithm invention algorithm. Recent LLM technology promises much more (?). Fortunately, a Ph.D. student in 1931 (Kurt Gödel), 10 years before the first thinking machines were physically built, showed by “pure thinking” that future mathematicians and computer scientists would never be jobless! – From the various adventures and stages in my life as a Ph.D. student, professor, and technology manager, I will derive some conclusions for the future careers of computer science Ph.D. students and junior and senior professors. Speaking at the place where I experienced the excitement of studying mathematics at the dawn of the computer age and working as one of the first programmers on the university’s first computer (a ZUSE Z23), I probably will not be able to avoid getting emotional.

Bruno Buchberger is Professor Emeritus of Computer Mathematics at Johannes Kepler University (JKU), Linz, Austria. He has been a visiting professor at numerous universities worldwide and a keynote speaker at international conferences over 150 times.

At the age of 23, in his Ph.D. thesis at the University of Innsbruck, Austria, he invented the theory of “Groebner Bases“, a general algorithmic method for handling multivariate polynomial systems. The theory has numerous applications in robotics, cryptography, hardware design, automated reasoning, etc., and is now a standard tool in all major mathematical software systems like Mathematica, Maple, GeoGebra, etc.. “Gröbner Bases” is now also an extra entry in the AMS Mathematical Subject Classifcation Index (13P10).

His current research focuses on the next level of AI, which combines symbolic computation, notably the automated reasoning methods in his Theorema system, with machine learning.

Buchberger founded the Journal of Symbolic Computation, the Research Institute for Symbolic Computation (RISC), the Softwarepark Hagenberg, and the University of Applied Sciences in Hagenberg.

His awards include six honorary doctorates in Europe, the UK, and Canada); membership in the Academy of Europe; the ACM Kanellakis Award for Theory and Practice (2008), the Herbrand Award for Automated Reasoning (2018, CADE), and Austrian of the Year in Research (2010).

How will future microarchitectures impact the security of existing cryptographic implementations? As we cannot keep reducing the size of transistors, chip vendors have started developing new microarchitectural optimizations to speed up computation. A recent study (Sanchez Vicarte et al., ISCA 2021) suggests that these optimizations might open the Pandora’s box of microarchitectural attacks. However, there is little guidance on how to evaluate the security impact of future optimization proposals. To help chip vendors explore the impact of microarchitectural optimizations on cryptographic implementations, we develop (i) an expressive domain-specific language, called LmSpec, that allows them to specify the leakage model for the given optimization and (ii) a testing framework, called LmTest, to automatically detect leaks under the specified leakage model within the given implementation. Using this framework, we conduct an empirical study of 18 proposed microarchitectural optimizations on 25 implementations of eight cryptographic primitives in five popular libraries. We find that every implementation would contain secret-dependent leaks, sometimes sufficient to recover a victim’s secret key, if these optimizations were realized. Ironically, some leaks are possible only because of coding idioms used to prevent leaks under the standard constant-time model.

Graph-Massivizer (https://graph-massivizer.eu/) is a Horizon Europe project coordinated by the University of Klagenfurt that researches and develops a holistic neuro-symbolic platform for processing and analytics of extreme data represented as semantic knowledge graphs with billions of nodes and edges. The talk presents a case study of using the platform for anomaly prediction in the CINECA supercomputing center using graph neural networks supported by machine learning-driven sampling algorithms for scalable training and inference on resource-constrained devices.

Behaviour distances have been studied as quantitative semantical counterpart to behavioural equivalences like bisimilarity. Instead of proving that states in probabilistic transition systems behave equivalently, they quantify how different/similar such states are. The most studied and accepted behaviour distance is one based on the Kantorovich distance between distributions. It comes with many beautiful theoretical results.

Another one, ϵ-distance is based on ϵ-bisimilarity, an approximate notion of behavioural equivalence that unifies both worlds: it gives an equivalence that relates states at distance at most ϵ. This distance has the advantages that it is intuitively easy to understand, relates systems that have close probabilities even if these differences can imply very different behaviour in the long run (for example, an imperfect implementation is close to its specification), and it is easy to compute. However, until recently it was not clear whether this distance shares any of the nice properties of the Kantorovich-based distance.

Recently, together with Joseé Desharnais, we showed that ϵ-distance indeed shares the useful properties of the Kantorovich distance, most notably it is the greatest fixpoint of a suitable functional. At the core of these results is the observation that replacing the Kantorovich distance with the Lévy-Prokhorov distance on distributions yields ϵ-distance. In addition, we see that ϵ-bisimulations have an appealing coalgebraic characterization.

Mobile malware continues to pose a serious threat to the security and privacy of mobile device users. In response, the research community has developed a wide range of machine learning-based detection approaches over the past decade, aiming to overcome the limitations of traditional signature-based techniques. While these learning-based methods have demonstrated strong potential, the field still faces a number of unresolved challenges—such as concept drift and evolving adversarial behaviors—that must be addressed to ensure sustained effectiveness in real-world environments. In this talk, we reflect on a decade of research in machine learning-based mobile malware detection, discuss key lessons learned, and highlight ongoing challenges that present opportunities for future work.