Catching bugs with Mutation Testing

Date: 10 January 2024
Presenter: Mike Papadakis, University of Luxembourg
Abstract

Mutation testing realises the idea of injecting and observing the consequences of artificially seeded faults in given systems. The approach is powerful as it allows revealing real faults, identifying weaknesses, simulating the effects of failures, estimating systems robustness and performing controlled studies. Mutation testing has reached a maturity phase and gradually gains popularity both in academia and in industry. This talk will present the main concepts, will introduce the key advances and will discuss open challenges and future research directions around mutation testing.

Biography

Mike Papadakis is an Associate Professor at the University of Luxembourg where he leads the SERVAL (SEcurity, Reasoning and VALidation) research team. His research interests include software testing, code analysis and artificial inteligence for software engineering. He is best known for his work on Mutation Testing for which he has been awarded IEEE TCSE Rising Star Award 2020. He has been General Chair of the ACM SIGSOFT ISSTA 2025 and IEEE ICSME 2021 and Program Chair of SSBSE 2021 and IEEE ICST 2022. He also serves at the editorial and review boards of international Software Engineering journals (STVR, Empirical Software Engineering, ACM Transactions on Software Engineering and Methodology) and has co-authored of more than 100 publications in international peer-reviewed conferences and journals. His work has been supported by Facebook, FNR, CETREL (SIX group company), BGL (BNP Paribas), Microsoft and PayPal.

Second AI Open Seminar by UniAI

Date: 19 January 2024
Presenter: D. Spinellis, S. Karagiannakos
Abstract

• Machine Learning in Computer Vision, Sergios Karagiannakos, AI Summer Founder, Data Enginee, Author of Deep Learning in Production.
• Ai- cli- lib: A command- line copilot, Diomidis Spinellis, Professor of Software Engineering at Athens University of Economics and Business.

The SNDS Project

Date: 19 January 2024
Presenter: G. Xylomenos, Athens University of Economics and Business
Abstract

The Secure Named Data Sharing (SNDS) project, funded by the NGI Sargasso EU project, is building extensions to the Named-Data Networking (NDN) ICN architecture, to implement query-based content retrieval and fine-grained content storage and update operations. SNDS will provide decentralized content integrity and authenticity protections by leveraging Decentralized Identifiers, Verifiable Credentials and BBS+ digital signatures, a digital signature scheme that enables selective disclosure of data by supporting Zero-Knowledge Proofs. SNDS will allow large collections of content (for example, IoT sensor readings) to be queried using the NGSI-LD API, which supports queries based on content type and/or attribute values, translating them into NDN operations. Case studies for SNDS include Vehicular networks and Smart buildings.

https://mm.aueb.gr/projects/snds

API-driven Program Synthesis for Testing Static Typing Implementations

Date: 09 February 2024
Presenter: Thodoris Sotiropoulos, ETH
Abstract

We introduce a novel approach for testing static typing implementations based on the concept of API-driven program synthesis. The idea is to synthesize type-intensive but small and well-typed programs by leveraging and combining application programming interfaces (APIs) derived from existing software libraries. Our primary insight is backed up by real-world evidence: a significant number of compiler typing bugs are caused by small test cases that employ APIs from the standard library of the language under test. This is attributed to the inherent complexity of the majority of these APIs, which often exercise a wide range of sophisticated type-related features. The main contribution of our approach is the ability to produce small client programs with increased feature coverage, without bearing the burden of generating the corresponding well-formed API definitions from scratch. To validate diverse aspects of static typing procedures (i.e., soundness, precision of type inference), we also enrich our API-driven approach with fault-injection and semantics-preserving modes, along with their corresponding test oracles.

We evaluate our implemented tool, Thalia, on testing the static typing implementations of the compilers for three popular languages, namely, Scala, Kotlin, and Groovy. Thalia has uncovered 84 typing bugs (77 confirmed and 22 fixed), most of which are triggered by test cases featuring APIs that rely on parametric polymorphism, overloading, and higher-order functions. Our comparison with state-of-the-art shows that Thalia yields test programs with distinct characteristics, offering additional and complementary benefits.

Evaluating Automatic Program Repair Capabilities to Repair API Misuses

Date: 22 February 2024
Presenter: Maria Kechagia, UCL
Abstract

API misuses are well-known causes of software crashes and security vulnerabilities. However, their detection and repair is challenging given that the correct usages of (third-party) APIs might be obscure to the developers of client programs. This paper presents the first empirical study to assess the ability of existing automated bug repair tools to repair API misuses, which is a class of bugs previously unexplored. Our study examines and compares 14 Java test-suite-based repair tools (11 proposed before 2018, and three afterwards) on a manually curated benchmark (APIREPBENCH) consisting of 101 API misuses. We develop an extensible execution framework (APIARTY) to automatically execute multiple repair tools. Our results show that the repair tools are able to generate patches for 28% of the API misuses considered. While the 11 less recent tools are generally fast (the median execution time of the repair attempts is 3.87 minutes and the mean execution time is 30.79 minutes), the three most recent are less efficient (i.e., 98% slower) than their predecessors. The tools generate patches for API misuses that mostly belong to the categories of missing null check, missing value, missing exception, and missing call. Most of the patches generated by all tools are plausible (65%), but only few of these patches are semantically correct to human patches (25%). Our findings suggest that the design of future repair tools should support the localisation of complex bugs, including different categories of API misuses, handling of timeout issues, and ability to configure large software projects. Both APIREPBENCH and APIARTY have been made publicly available for other researchers to evaluate the capabilities of repair tools on detecting and fixing API misuses.

Preprint:

https://discovery.ucl.ac.uk/id/eprint/10124292/1/api_repair_tools_study_camera_ready_submitted.pdf

Biography

Dr Maria Kechagia is a Research Fellow at University College London. Previously, she was a postdoctoral researcher at the Delft University of Technology. She obtained a PhD degree from the Athens University of Economics and Business and an MSc degree from Imperial College London. Her research interests include static and dynamic analysis, automated program repair, software analytics, and software optimisation (e.g., energy efficiency). She has been a programme committee member of the research track of top software engineering venues including ICSE, ASE, ISSTA, MSR, ICSME, and SANER, and a reviewer for top software engineering journals including TSE, TOSEM, EMSE, and JSS.

nvshare: Practical GPU Sharing Without Memory Size Constraints

Date: 27 March 2024
Presenter: Georgios Alexopoulos, University of Athens
Abstract

GPUs are essential for accelerating Machine Learning (ML) workloads. A common practice is deploying ML jobs as containers managed by an orchestrator such as Kubernetes. Kubernetes schedules GPU workloads by exclusively assigning a device to a single job, which leads to massive GPU underutilization, especially for interactive development jobs with significant idle periods.

Current GPU sharing approaches assign a fraction of GPU memory to each colocated job to avoid memory contention and out-of-memory errors. However, this is impractical, as it requires a priori knowledge of memory usage and does not fully address GPU underutilization.

We propose nvshare, which transparently enables page faults (i.e.,exceptions that are raised when an entity attempts to access a resource) to allow virtual GPU memory oversubscription. In this way we permit each application to utilize the entire physical GPU memory (Video RAM).

To prevent thrashing (a situation in which page faults dominate execution time) in a reliable manner, nvshare serializes overlapping GPU bursts from different applications. We compared nvshare with KubeShare, a state-of-the-art GPU sharing solution.

Our results indicate that both perform equally well in convential sharing cases where total GPU memory usage fits into VRAM. For memory oversubscription scenarios, which KubeShare does not support, nvshare outperforms the sequential execution baseline by up to 1.35x.

Our tool has been adopted in both research and industrial settings.

• Video demonstration: https://www.youtube.com/watch?v=9n-5sc5AICY
• Github repository: https://github.com/grgalex/nvshare

Introduction to Dynamic Analysis of Android Applications

Date: 10 April 2024
Presenter: Simon Althaus, University of Darmstadt
Abstract

Smartphones are ever-present in our daily lives and handle a wealth of sensitive information like text messages and photos. Malicious applications can obtain access to such data and leak them to third parties with potentially grave consequences (i.e., theft, blackmail, etc.). Therefore, the study of malware, with the goal of identifying such malicious applications, has attracted significant research interest in recent years, especially focusing on Android applications. Malware research based on a combination of static and dynamic approaches has been shown to be effective in identifying a range of typical malware types like example keyloggers and ransomware. Moreover, it has been shown that applications that are not malware in the traditional sense, e.g. social networking applications or even over-privileged system applications, might leak significant amounts of data without user consent or notification. Therefore, this talk will provide an overview of trends in research regarding static and dynamic analysis for Android applications with the goal of identifying such malicious or information-leaking behaviors. We will discuss different approaches based on taint analysis, system call analysis, provenance tracking, network traffic analysis and more. The talk will also cover the extended Berkeley Packet Filter (eBPF) and how it can be useful for tracing and dynamic analysis.

Biography

Simon Althaus is a Research Associate at the Telecooperation Lab at Technical University of Darmstadt. He obtained a MSc degree from Technical University of Darmstadt working on botnets. Simon is currently researching in the field of Android security. His research interests include dynamic analysis, eBPF, provenance tracking, and privacy-enhancing technologies.

Impact Analysis of Greece’s OpenGov Public Consultation Contributions on Final Legislation

Date: 22 April 2024
Presenter: Antonis Athanasiou, AUEB
Abstract

Public consultations and eParticipation initiatives have been a subject of research for many years. The Greek ICT-enabled consultation platform, Opengov, is used as a case study to measure consultation impact on voted law. Literature generally agrees that the goal of consultations is — among others — impact on developed policy, however there is not a universally accepted way of measuring this impact. This thesipos takes its own approach to defining impact as the similarity between draft legislation and voted legislation. Work from the public consultation field and the Natural Language Processing is presented. An extensive data collection and extraction process was necessary for this analysis which is also shown. Ordinary Least Squares (OLS) analysis was used to assert and quantify impact. Results show, among other explanatory variables, that the more contributions draft legislation gets, the more it changes on average. The thesis finishes with discussion of the results, the limitations of the selected approach and suggests ideas for improvement.

Data Analytics with Blockly and Unix Tools

Date: 29 April 2024
Presenter: Pantelis Kakavas and Vagelis Tsipis
Abstract

We introduce a new approach to integrating Unix command-line tools with visual programming, aiming to enhance accessibility and usability in data processing pipelines. Leveraging Blockly, a visual programming language, Unix commands are represented as graphical blocks, removing the need for advanced programming skills and making them more accessible to a broader audience. The creation of a visual programming environment enables users to construct data processing pipelines through drag-and-drop actions, facilitated by JSON definition files that create Unix command line tool abstractions linking visual blocks to Unix commands. The design is user friendly and deepens users' comprehension of Unix commands, improving accessibility for difficult data manipulation tasks. The project highlights the importance of visual programming in bridging the gap between complex command-line operations and user-friendly interfaces, thereby expanding the toolkit available for data scientists and researchers of various fields, while also offering new opportunities for interactive educational tools in command-line interface learning.

SoK: What don't we know? Understanding Security Vulnerabilities in SNARKs

Date: 10 May 2024
Presenter: Stefanos Chaliasos, Imperial College London
Abstract

Zero-knowledge proofs (ZKPs) have evolved from being a theoretical concept providing privacy and verifiability to having practical, real-world implementations, with SNARKs (Succinct Non-Interactive Argument of Knowledge) emerging as one of the most significant innovations. Prior work has mainly focused on designing more efficient SNARK systems and providing security proofs for them. Many think of SNARKs as "just math," implying that what is proven to be correct and secure is correct in practice. In contrast, this paper focuses on assessing end-to-end security properties of real-life SNARK implementations. We start by building foundations with a system model and by establishing threat models and defining adversarial roles for systems that use SNARKs. Our study encompasses an extensive analysis of 141 actual vulnerabilities in SNARK implementations, providing a detailed taxonomy to aid developers and security researchers in understanding the security threats in systems employing SNARKs. Finally, we evaluate existing defense mechanisms and offer recommendations for enhancing the security of SNARK-based systems, paving the way for more robust and reliable implementations in the future.

Security needs Trust needs Security

Date: 17 May 2024
Presenter: Prof. Dr. Max Mühlhäuser, Technical University of Darmstadt
Abstract

The close relationship between trust and (IT/cyber) security has often been addressed in debates and publications, but remains underrepresented in mainstream research. There have been attempts to harmonize both areas and consider them as two sides of the same medal—as soft security and hard security or conversely as soft trust and hard trust, but these terminology focused discussions did not find common acceptance either. In this talk, we will briefly review the arguments about why trust and security are intimately connected, and outline the broader landscape of corresponding approaches. We will then look at computational trust as a particular family of trust-related approaches. Next, we will discuss how trust and security can act in service of each other. Under the heading "Security needs trust", we will look at how the trust aspects of IT security scenarios can be assessed and operationalized using computational trust; under the heading "Trust needs security", we will outline some approaches that can help to protect the assessment of computational trust against various threats. A brief outlook on open challenges will conclude the presentation.

Biography

Max Mühlhäuser is a full professor at the Technical University of Darmstadt and head of Telecooperation Lab. He holds key positions in several large collaborative research centers and is leading the Doctoral School on Privacy and Trust for Mobile Users. He and his lab members conduct research on the future Internet, Human Computer Interaction, Intelligent Systems, and PST (Privacy, Security & Trust). Max founded and managed industrial research centers, and worked as either professor or visiting professor at universities in Germany, the US, Canada, Australia, France, and Austria. He is a member of acatech, the German Academy of the Technical Sciences, ACM Distinguished Member, and IEEE Fellow. He published over 700 peer-reviewed articles and was and is active in numerous conference program committees, as organizer of several annual conferences, and as a member of editorial boards or guest editor for journals such as ACM IMWUT, ACM ToIT, Pervasive Computing, ACM Multimedia, and Pervasive and Mobile Computing.

Security in a World of Software Supply-Chain Vulnerabilities

Date: 20 May 2024
Presenter: Nikos Vasilakis, Brown University
Abstract

Modern software incorporates thousands of third-party components. Bugs or security vulnerabilities in these components can seriously compromise the integrity of incorporating applications. Because of their widespread use, and the difficulty of vetting the enormous number of integrated components for vulnerabilities, they comprise a compelling target for attackers, who purposefully insert vulnerabilities into widely used components with the goal of compromising the integrity of entire software ecosystems. I will present a series of systems that leverage component boundaries to offer automated solutions to vulnerabilities that appear in the software component supply chain. These solutions leverage system- and language-level containment techniques to prevent different classes of attacks from affecting these applications and the broader system in which they execute. Combined, they provide a holistic and in-depth transformation-based approach to securing entire software ecosystems.

Biography

Nikos Vasilakis is an Assistant Professor of Computer Science at Brown University. His research encompasses systems, programming languages, and security — and has been recognized by several distinguished paper awards. His current focus is on automatically transforming systems to add new capabilities such as parallelism, distribution, and security — against a variety of threat models. Nikos is also the chair of the Technical Steering Committee behind PaSh, a shell-script optimization system hosted by the Linux Foundation. More: https://nikos.vasilak.is

AppStack: An Agile Platform for Running Digital Public Services

Date: 03 June 2024
Presenter: Dimitris Mitropoulos, BALab
Abstract

From 2019 and onwards, the Greek National Infrastructures for Research and Technology (GRNET) is responsible for the development, operation and maintenance of several governmental services, thus playing an important role in Greece's digital transformation. To address the different challenges related to this role, GRNET teams developed AppStack, a cloud-native platform for running gov.gr horizontal services such as the gov.gr portal, the electronic issuance of documents signed by the Greek state, and gov wallet among others. AppStack comes with an enabling environment for integrating open-source software components. By taking advantage of this feature, DevOps can incorporate suitable tools to tackle scalability and security issues. Currently, AppStack hosts workloads that serve more than 8 million Greek citizens, are able to handle more than 20K requests per second, and can generate hundreds of digital documents signed by the Greek state per second. In this talk we will present AppStack and highlight the significance of its different components through a number of use cases. Finally, we will describe some key experiences from production.

LLMs for Code: The Potential, Prospects, and Problems

Date: 12 June 2024
Presenter: Tushar Sharma, Dalhousie University
Abstract

With the introduction of Large Language Models (LLMs) and their integration with software development tasks, the software development landscape has changed drastically in the last couple of years. In this session, we delve into the intricate world of large language models for code (LLM4Code) and explore their benefits, challenges, and threats. On one hand, these models have revolutionized code completion, bug detection, and even generated entire sections of code with remarkable accuracy. However, on the other side, several concerns have emerged surrounding inaccurate, buggy, and vulnerable code generation, biases, implications for climate, and the potential for unintended consequences. The talk promises an exploratory take that not only seeks to harness the potential of LLMs4Code but also ensures a conscientious and mindful approach toward their integration into our coding practices.

Biography

Tushar Sharma is a tenure-track assistant professor at Dalhousie University, Canada. He leads the Software Maintenance and Analytics Research Team (SMART) lab in the Faculty of Computer Science. Topics related to software engineering, sustainable artificial intelligence, and machine learning for software engineering (ML4SE) define his career interests. He earned a PhD from the Athens University of Economics and Business, Athens, Greece, specializing in software engineering in 2019. He has ten years of industry experience, mainly with Siemens Research, USA and India. He co-authored Refactoring for Software Design Smells: Managing Technical Debt and two Oracle Java certification books. He founded Designite, offering code quality assessment tools that many practitioners and researchers use worldwide. He is an IEEE Senior Member.

Bloat beneath Python's Scales: A Fine-Grained Inter-Project Dependency Analysis

Date: 03 July 2024
Presenter: Georgios Drosos, ETH Zurich
Abstract

Modern programming languages promote software reuse via package managers that facilitate the integration of inter-dependent software libraries. Software reuse comes with the challenge of dependency bloat, which refers to unneeded and excessive code incorporated into a project through reused libraries. Such bloat exhibits security risks and maintenance costs, increases storage requirements, and slows down load times.

In this work, we conduct a large-scale, fine-grained analysis to understand bloated dependency code in the PyPI ecosystem. Our analysis is the first to focus on different granularity levels, including bloated dependencies, bloated files, and bloated methods. This allows us to identify the specific parts of a library that contribute to the bloat. To do so, we analyze the source code of 1,302 popular Python projects and their 3,232 transitive dependencies. For each project, we employ a state-of-the-art static analyzer and incrementally construct the fine-grained project dependency graph (FPDG), a representation that captures all inter-project dependencies at method-level. Our reachability analysis on the FPDG enables the assessment of bloated dependency code in terms of several aspects, including its prevalence in the PyPI ecosystem, its relation to software vulnerabilities, its root causes, and developer perception.

Our key finding suggests that PyPI exhibits significant resource underutilization: more than 50% of dependencies are bloated. This rate gets worse when considering bloated dependency code at a more subtle level, such as bloated files and bloated methods. Our fine-grained analysis also indicates that there are numerous vulnerabilities that reside in bloated areas of utilized packages (15% of the defects existing in PyPI). Other major observations suggest that bloated code primarily stems from omissions during code refactoring processes and that developers are willing to debloat their code: Out of the 36 submitted pull requests, developers accepted and merged 30, removing a total of 35 bloated dependencies. We believe that our findings can help researchers and practitioners come up with new debloating techniques and development practices to detect and avoid bloated code, ensuring that dependency resources are utilized efficiently.

Biography

Georgios Drosos is currently a Direct Doctorate student studying Computer Science at ETH Zurich. His academic journey began at the Athens University of Economics and Business (AUEB), where he earned his Bachelor’s Degree in 2022 from the Department of Management Science and Technology, majoring in Software Engineering and Data Science. Since 2021, Georgios has been a member of the Business Analytics Laboratory. During his involvement with the lab, he worked as Research Software Engineer on the FASTEN and SecOPERA Research Projects. His research interests involve Software Testing, Software Ecosystems and Program Analysis.

Rumor Spreading and Mitigation in Random Environments

Date: 08 July 2024
Presenter: Nick Bambos, Stanford University*
Abstract

We model epidemic spreading of rumors/misinformation (or malware, pathogens) in a population of individuals (or devices) within a large region. We allow for mobility of infected individuals, who can infect others while roaming around. We distribute fact checking (testing) centers around the region, where individuals testing positive are "quarantined." Infected and recovered individuals cannot be reinfected in the current epidemic wave. The infection transmission rate (so called, infectivity) is a random field over the region.

We explore conditions under which the infection inherently spreads, as opposed to dying out. We observe that the average infectivity is not enough to characterize spreading, as fluctuations do matter. Indeed, even under subcritical infectivity averages, overcritical fluctuations can cause the infection to spread. We then focus on the impact of testing center density on suppressing an epidemic that has the inherent potential to spread. Finally, we discuss how to optimize testing vs. recovery resources.

• Joint work with Prof. Aris Moustakas (UoA, Physics) and Kyriakos Lotidis (Stanford)
• In memory of Petros Meramveliotakis (UoA, Physics), who was a key contributor to this work

Biography

Nick Bambos is R. Weiland Professor of electrical engineering and of management science & engineering at Stanford University. He received is PhD in electrical engineering and computer sciences from the University of California at Berkeley in 1989 and his engineering diploma from NTUA in 1984. His main research interests are in high-performance engineering of computer systems and networks, including comp/comm resource scheduling, power management of data centers, computing risk management, etc. His methodological interests are in stochastic modeling and control, machine learning and AI.

Data Management Innovation at Amazon Web Services

Date: 10 July 2024
Presenter: Ippokratis Pandis, AWS
Abstract

Amazon Web Services is the largest provider of data management services in the world, even though it only started providing such services in 2011. In this talk we will try to explain some of the reasons behind this success. First we'll talk about the anatomy of a data management service in the cloud, and then we'll show that AWS is leading the way and constantly building on all layers of these systems.

Biography

Ippokratis Pandis is a VP/Distinguished Engineer at Amazon Web Services responsible for the technical direction of AWS’s Analytics and Relational services. Ippokratis spends a lot of his time in Amazon Redshift. Redshift is Amazon's enterprise cloud data warehouse service. Previously, Ippokratis has held positions as software engineer at Cloudera, where he worked on the Impala SQL-on-Hadoop query engine, and as member of the research staff at the IBM Almaden Research Center, where he worked on IBM DB2 BLU. Ippokratis received his PhD from the Electrical and Computer Engineering department at Carnegie Mellon University. He is the recipient of Best Demonstration awards at ICDE 2006 and SIGMOD 2011, and Test-of-Time award at EDBT 2019. He has served as PC chair of DaMoN 2014, DaMoN 2015, CloudDM 2016, HPTS 2019, ICDE Industrial 2022 and SIGMOD Industrial 2024, as well as General Chair of SIGMOD 2023 and the president of HPTS.

Broken Agreement: The Evolution of Solidity Error Handling

Date: 02 October 2024
Presenter: Charalambos Mitropoulos, TUC
Abstract

Background. A smart contract is a computer program enclosing the terms of a legal agreement between two or more parties which is automatically verified and executed via a computer network called blockchain. Once a smart contract transaction is completed the blockchain is updated and the transaction cannot be changed anymore. This implies that any error codified in the smart contract program cannot be rectified. Therefore, it is of vital importance that developers of smart contracts properly exploit error handling to prevent issues during and after the contract execution. Existing programming languages for smart contracts, support developers in this task by providing a set of Error Handling (EH) features. However, it is unclear the extent to which developers effectively use EH in practice. Aims. Our work aims to fill this gap by em- pirically investigating the state of practice on the adoption of EH features of one of the most popular programming languages for smart contracts, namely Solidity. Method. We empirically analyse the usage of EH features in 283K unique open-source Solidity smart contracts for the Ethereum blockchain. Results. Our analysis of the documentation of the different versions of Solidity coupled with the empirical evaluation of the EH uses and misuses found in real-word smart contracts, indicate that, among other things, Solidity EH features have been changing frequently across versions, and that the adoption of most of the Solidity EH features has been limited in practice. However, we observe an upward trend in the usage of the require EH feature, which is specifically designed for smart contract development. Conclusions. The insights from our study could help developers improve their EH practice as well as designers of smart contract programming languages to equip their language with appropriate EH features.

Lessons from Unix History

Date: 21 October 2024
Presenter: Diomidis Spinellis, AUEB
Abstract

This talk examines the enduring lessons from the evolution over the past fifty years of Unix and its deep impact on modern computing. By tracing Unix's development from the First Research Edition to modern FreeBSD releases, the presentation highlights key innovations such as prototyping, portability, modular design, and prioritizing developer efficiency over machine resources.

The presentation focuses on architectural principles that have been central to Unix, including aggressive partitioning, composition, layering, convention-based extensibility, and the extensive use of pipelines and filters. Drawing on research and case studies, this talk provides a technical retrospective on Unix's legacy and the principles that continue to guide effective system and software development today. It is aimed at developers, architects, and technology enthusiasts interested in exploring the historical and ongoing relevance of Unix's design principles.

Towards a Greek Proverb Atlas: Computational Spatial Exploration and Attribution of Greek Proverbs

Date: 04 November 2024
Presenter: Panos Louridas, AUEB
Abstract

Proverbs carry wisdom transferred orally from generation to generation. Based on the place they were recorded, this study introduces a publicly-available and machine-actionable dataset of more than one hundred thousand Greek proverb variants. By quantifying the spatial distribution of proverbs, we show that the most widespread proverbs come from the mainland while the least widespread proverbs come primarily from the islands. By focusing on the least dispersed proverbs, we present the most frequent tokens per location and undertake a benchmark in geographical attribution, using text classification and regression (text geocoding). Our results show that this is a challenging task for which specific locations can be attributed more successfully compared to others. The potential of our resource and benchmark is showcased by two novel applications. First, we extracted terms moving the regression prediction toward the four cardinal directions. Second, we leveraged conformal prediction to attribute 3,676 unregistered proverbs with statistically rigorous predictions of locations each of these proverbs was possibly registered in.

Co-authored with John Pavlopoulos and Panagiotis Filos.

Accepted for publication in The 2024 Conference on Empirical Methods in Natural Language Processing, November 12--16, Miami, Florida.

AI Open Seminar, Exploring Trustworthiness and Self-Supervised Learning in Artificial Intelligence

Date: 29 November 2024
Presenter: Stafilakis Themos, AUEB and Dimitris Apostolou, UniPi
Abstract

Adaptive and Personalised Education System using AI:
Exploring the process of designing an AI-powered system to tailor educational experiences to the unique needs of each student. By leveraging advanced algorithms and data analysis, the system adapts learning pathways, pace, and methods of experimental activities in online STEM labs and delivers targeted feedback based on real-time student performance to match individual capabilities and preferences, aiming to enhance engagement, understanding, and overall academic success. This approach emphasizes a student-centered approach, supporting critical thinking and problem-solving skills.

Deep and Self-Supervised Learning in Voice Technologies-Stafilakis Themos: Voice technologies are revolutionizing how we interact with machines, and at their core are breakthroughs in deep and self-supervised learning. This session unveils the latest advancements driving smarter, more intuitive voice applications. Discover how these innovations are unlocking new possibilities for businesses and redefining user experiences.

Biographies

Themos Stafylakis is an Associate Professor at the Athens University of Economics and Business and a voice recognition expert. He holds a PhD from NTUA and has conducted postdoctoral research in Canada and the UK. Since 2018, he has led the machine learning and voice biometrics departments at Omilia. He is also a collaborating researcher with the "Athena" Research Center. His work focuses on advancing voice and speaker recognition technologies.

Dr. Dimitris Apostolou is a Professor in the Informatics Department at the University of Piraeus and a Senior Researcher at the Institute of Communication and Computer Systems. He holds a PhD from NTUA and advanced degrees from UCL and NJIT. With over 140 publications and 3 best paper awards, his research focuses on decision support, knowledge management, and intelligent information systems. He has participated in more than 30 European R&D projects and has a decade of experience in management and technology consulting. He also serves on the editorial boards of international journals and contributes to major conferences worldwide.

A seminar series powered by UniAI