MSc Thesis presentation of Vassiliki Strouthopoulou – Thursday, 2/7/2026
Student: “Vassiliki Strouthopoulou”
Program: “Data Science and Information Technologies”
Title: “Transformer-Based Architectures for Financial Time Series Forecasting: From Single-Stock Learning to Cross-Asset Generalization”
Abstract:
This thesis investigates the application of Transformer-based architectures to the problem of financial time series forecasting, with a particular focus on short-term stock price prediction. The study investigates model complexity, beginning with a Seq2One Transformer encoder, advancing through an Attention-Only configuration, and culminating in a Seq2Seq encoder–decoder architecture. Using daily prices from NASDAQ-listed companies, multiple experiments were conducted to evaluate the predictive accuracy, temporal consistency, and generalization ability of each architecture across varying sequence lengths and training approaches.
The obtained results demonstrate a performance improvement across models. The Seq2One architecture establishes a stable baseline but exhibits cumulative error in autoregressive forecasts. The Attention-Only model, despite its simplicity, achieves competitive short-term accuracy by effectively capturing localized temporal dependencies. The Seq2Seq Transformer outperforms both predecessors by jointly modeling multi-step dependencies and generating consistent five-day forecasts, achieving the lowest mean absolute and mean
squared errors among all configurations. Furthermore, a multi-stock Seq2Seq experiment shows that Transformers can generalize learned temporal patterns across different
assets, producing accurate predictions for unseen stocks within the NASDAQ domain.
The findings highlight the robustness and flexibility of attention-based models for financial forecasting tasks. They also underline the importance of sequence length selection, with short-to-moderate input windows (10–30 days) offering the best trade-off between responsiveness and noise reduction. Finally, the cross-stock experiments reveal promising evidence of transfer learning capabilities of Transformer architectures in the context of multi-asset forecasting systems. Future work may extend this framework through multi-feature integration, adaptive fine-tuning, and volatility-aware embeddings to further enhance model interpretability and resilience to market shifts.
Date-Time: 2/7/2026 – 12:00 PM
Examination Committee:
Associate Professor Aggelos Pikrakis, University of Piraeus
Dr. Konstantinos Koutroumbas, National Observatory of Athens
Dr. Kosmas Kritsis, ILSP/Athena Research Center
MSc Thesis presentation of Eirini Baltzi – Thursday, 2/7/2026
Student: “Eirini Baltzi”
Program: “Data Science and Information Technologies”
Title: “Transformer-Based Modeling of Irregular Clinical Time Series: Bi-Axial Attention and Temporal Encoding”
Abstract:
In this work, we study transformer-based modeling of irregular clinical time series derived from Electronic Health Records, with particular focus on the Bi-Axial Transformer (BAT) architecture. Clinical prediction from EHR data remains challenging due to irregular sampling, sparse observations, informative missingness and the coexistence of dynamic and static patient information. Within this setting, our goal is to first, reproduce the BAT architecture and verify that the reconstructed implementation is consistent with the original pa- per and second, to investigate a modification of its temporal encoding mechanism through Rotary Positional Encoding (RoPE). The experimental evaluation is conducted on three clinical benchmarks: PhysioNet 2019 for sepsis prediction, PhysioNet 2012 for in-hospital mortality prediction, and MIMIC-III, where additional experiments are performed under a separate preprocessing pipeline. In addition to the reproduced BAT and the proposed BAT-RoPE variant, two baseline models are considered, namely GRU-D and a standard Transformer. The results show that BAT is a strong and competitive architecture across both datasets. The original BAT achieves the best AUPRC on PhysioNet 2019, while BAT-RoPE performs best on PhysioNet 2012. Additional attention and ablation analyses suggest that BAT captures structured temporal and inter-variable relationships and relies primarily on the dynamic clinical time series, while focal loss does not provide a consistent improvement.
Date-Time: 2/7/2026 – 11:00 AM
Examination Committee:
Associate Professor Aggelos Pikrakis, University of Piraeus
Dr. Kosmas Kritsis, ILSP/Athena Research Center
Dr. Vassilis Psomas, Czech Technical University, Prague
Msc Thesis Presentation
Thursday, July 2 · 11:00am – 12:00pm
Time zone: Europe/Athens
Webex joining info
Video call link: https://unipi.webex.com/meet/webex-host2
MSc Thesis presentation of Georgios Xydias – Thursday 25/06/2026
Title: “Modular LLM Pipeline for Task-Oriented Customer Service Dialogue”
Abstract:
Task-oriented dialogue (TOD) systems are a cornerstone of modern customer-service
automation, promising to handle structured user goals such as searching, booking, and
information requests through natural conversation. The recent rise of instruction-tuned Large
Language Models (LLMs) has reopened a fundamental design question: should a single
general-purpose LLM handle the entire turn, or should the responsibilities of a dialogue agent
be split into a modular pipeline of smaller, role-specific components? Existing work explores
both ends of this spectrum, but rarely compares them head-to-head under a setup that uses the
same dataset, the same evaluation procedure, and the same model families across
architectures, leaving the practical trade-offs between architectural simplicity and architectural
decomposition never brought together under a single controlled comparison.
In this work, we introduce a controlled comparative study of three task-oriented dialogue
architectures for hotel and restaurant customer service, evaluated on the MultiWOZ 2.2
benchmark. The first architecture is a single-LLM baseline that performs dialogue state tracking,
database lookup, and response generation in one prompt. The second is a zero-shot modular
pipeline that separates the dialogue state tracker from the response generator, connected by a
deterministic database lookup and a policy gate, with a lightweight guardrail on the generated
response. The third applies role-specific LoRA fine-tuning to the two LLM-based modules of the
pipeline (the dialogue state tracker, DST, and the response generator, ResponseGen) by
training small open-source models for their assigned role via QLoRA. All three architectures are
evaluated with the standardized MultiWOZ evaluator, reporting both dialogue-state metrics
(Joint Goal Accuracy, Slot F1) and end-to-end metrics (Inform, Success, BLEU, Combined),
complemented by a custom set of operational metrics covering hallucination, policy violations,
system correctness, latency, and cost.
Our findings show that a) architectural decomposition is not universally beneficial: it
substantially helps weaker models that struggle to manage every responsibility in a single
prompt, but it can hurt the strongest open-source models that already coordinate all
responsibilities reliably, and b) role-specific LoRA fine-tuning lifts dialogue-state tracking of
small open-source models close to the level of much larger commercial APIs, making them
cost-efficient, fast, and competitive alternative for slot tracking. These results show that
architectural decomposition and role-specific fine-tuning are effective tools, but only under the
right conditions, for building reliable LLM-based customer-service dialogue systems.
Date-Time: Thursday 25/06/2026 – 10:00 AM
Examination Committee:
Dr. Kosmas Kritsis
Prof. Aggelos Pikrakis
Dr. Georgios Paraskevopoulos
Msc Thesis Presentation
Thursday , June 25 · 10:00am – 11:00pm
Time zone: Europe/Athens
Microsoft Teams meeting
Join:
https://teams.microsoft.com/meet/318082679942220?p=ne3PcOyBsgIcxTjyGu
MSc Thesis presentation of Rafail N. Adam – Monday, 08/6/2026
Student: “Rafail N. Adam”
Program: “Data Science and Information Technologies”
Title: “MicroRNA target prediction with novel machine learning methods beyond quadratic attention”
Abstract:
MicroRNAs (miRNAs) are small non-coding RNA molecules that regulate
gene expression, primarily by suppressing mRNA translation through
binding to MicroRNA Response Elements (MREs). They are involved in
many physiological processes, and their dysregulation is associated with
disease. Advances in multi-omics and personalized medicine have
enabled miRNA discovery and analysis through high-throughput
techniques such as small RNA-seq, AGO-eCLIP, and AGO-CLASH, which
quantify miRNA–mRNA interactions. Because a single miRNA can target
multiple mRNAs, understanding their regulatory effects requires
advanced computational approaches. Consequently, numerous machine
learning and deep learning models have been developed for miRNA
target prediction, either by identifying valid miRNA–MRE interactions or
by predicting gene-level effects using transcriptomic data. This thesis
presents the development of a machine/deep-learning model that
predicts interactions between miRNAs and MREs. AGO2-eCLIP data were
used to train and evaluate several models, resulting in a weighted
ensemble approach capable of distinguishing valid from invalid
interactions with Average Precision performance comparable to current
state-of-the-art methods. The model also provides interpretability by
estimating the contribution of each feature to the final prediction.
Date-Time: 08/06/2026 – 11:00 AM
Examination Committee:
Prof. Martin Reczko
Prof. Artemis Hatzigeorgiou
Prof. Alexandros Dimopoulos
Msc Thesis Presentation
Monday, June 8 · 11:00am – 12:00pm
Time zone: Europe/Athens
Google Meet joining info
Video call link: https://meet.google.com/wfp-snxk-trn
Παρουσίαση των Μεταπτυχιακών Προγραμμάτων Σπουδών (ΠΜΣ) και Διατμηματικών Προγραμμάτων Μεταπτυχιακών Σπουδών (ΔΠΜΣ) του Τμήματος
Η Παρουσίαση των Μεταπτυχιακών Προγραμμάτων Σπουδών (ΠΜΣ) και Διατμηματικών/Διιδρυματικών Προγραμμάτων Μεταπτυχιακών Σπουδών (ΔΠΜΣ) του Τμήματος Πληροφορικής και Τηλεπικοινωνιών ΕΚΠΑ, θα πραγματοποιηθεί δια ζώσης, στο Αναγνωστήριο του Τμήματος Πληροφορικής και Τηλεπικοινωνιών την Πέμπτη 21 Μαΐου 2026, ώρα 11.00 έως 14:00.