Erion-Vasilis Pikoulis

Dr. Erion-Vasilis Pikoulis

orcid

Research Focus & Metrics

Academic Career Length

12Years (2015-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
27 November 2024 in IEEE Access
Verified
Journal Article
Open access
CAChristos Anagnostopoulos
Christos Anagnostopoulos
AGAlexandros Gkillas
Alexandros Gkillas
CMChristos Mavrokefalidis
Christos Mavrokefalidis
EPErion-Vasilis Pikoulis
Erion-Vasilis Pikoulis
NPNikos Piperigkos
Nikos Piperigkos
Aris LalosAris Lalos
Aris Lalos
The unprecedented technological advancements in Artificial Intelligence (AI) and the Internet of Things (IoT) have given rise to ecosystems of intelligent, interconnected devices, forming the Artificial Intelligence of Things (AIoT). These systems, due to security and privacy concerns, necessitate solutions that adhere to the data-driven learning paradigm of Federated Learning (FL) while simultaneously addressing data and resource heterogeneity. This survey, which focuses on smart environments within the transportation, agriculture, manufacturing, and medical sectors, begins with an in-depth review of FL methods and multimodal data-driven learning methodologies. Subsequently, various sensor modalities employed in these environments are presented, as well as the most common multimodal data fusion strategies and their associated fusion operators. The study then explains its shift in focus from data-level to model-level cooperation, delving into multimodal FL (MMFL) systems by categorizing architectures, data processing methods, and model aggregation rules. Emphasis is given to the case of heterogeneous MMFL, as it is identified as the most promising and relatively novel direction that has received insufficient attention due to its recent emergence. The paper introduces a novel classification of the strategies employed when agents have different sensing modalities and model architectures, offering an in-depth analysis of how various fusion approaches can be adapted to accommodate the diversity in data and models. Finally, it examimes the utilization of MMFL in the four application domains and concludes with an analysis of open challenges and future research directions in this promising field.

2 April 2021 in Applied Sciences
Verified
Journal Article
Open access
PDF file
EPErion-Vasilis Pikoulis
Erion-Vasilis Pikoulis
ZIZafeiria-Marina Ioannou
Zafeiria-Marina Ioannou
MPMersini Paschou
Mersini Paschou
ESEvangelos Sakkopoulos
Evangelos Sakkopoulos
Face morphing poses a serious threat to Automatic Border Control (ABC) and Face Recognition Systems (FRS) in general. The aim of this paper is to present a qualitative assessment of the morphing attack issue, and the challenges it entails, highlighting both the technological and human aspects of the problem. Here, after the face morphing attack scenario is presented, the paper provides an overview of the relevant bibliography and recent advances towards two central directions. First, the morphing of face images is outlined with a particular focus on the three main steps that are involved in the process, namely, landmark detection, face alignment and blending. Second, the detection of morphing attacks is presented under the prism of the so-called on-line and off-line detection scenarios and whether the proposed techniques employ handcrafted features, using classical methods, or automatically generated features, using deep-learning-based methods. The paper, then, presents the evaluation metrics that are employed in the corresponding bibliography and concludes with a discussion on open challenges that need to be address for further advancing automatic detection of morphing attacks. Despite the progress being made, the general consensus of the research community is that significant effort and resources are needed in the near future for the mitigation of the issue, especially, towards the creation of datasets capturing the full extent of the problem at hand and the availability of reference evaluation procedures for comparing novel automatic attack detection algorithms.

22 December 2021 in Seismological Research Letters
Verified
Journal Article
Closed access
PDF file
NTNikolaos Triantafyllis
Nikolaos Triantafyllis
IVIoannis E. Venetis
Ioannis E. Venetis
IFIoannis Fountoulakis
Ioannis Fountoulakis
EPErion-Vasilis Pikoulis
Erion-Vasilis Pikoulis
Efthimios SokosEfthimios Sokos
Efthimios Sokos
CEChristos Evangelidis
Christos Evangelidis
Automatic moment tensor (MT) determination is essential for real-time seismological applications. In this article, Gisola, a highly evolved software for MT determination, oriented toward high-performance computing, is presented. The program employs enhanced algorithms for waveform data selection via quality metrics, such as signal-to-noise ratio, waveform clipping, data and metadata inconsistency, long-period disturbances, and station evaluation based on power spectral density measurements in parallel execution. The inversion code, derived from ISOLated Asperities—an extensively used manual MT retrieval utility—has been improved by exploiting the performance efficiency of multiprocessing on the CPU and GPU. Gisola offers the ability for a 4D spatiotemporal adjustable MT grid search and multiple data resources interconnection to the International Federation of Digital Seismograph Networks Web Services (FDSNWS), the SeedLink protocol, and the SeisComP Data Structure standard. The new software publishes its results in various formats such as QuakeML and SC3ML, includes a website suite for MT solutions review, an e-mail notification system, and an integrated FDSNWS-event for MT solutions distribution. Moreover, it supports the ability to apply user-defined scripts, such as dispatching the MT solution to SeisComP. The operator has full control of all calculation aspects with an extensive and adjustable configuration. MT’s quality performance, for 531 manual MT solutions in Greece between 2012 and 2021, was measured and proved to be highly efficient.

30 March 2022 in Sensors
Verified
Journal Article
Open access
PDF file
EPErion-Vasilis Pikoulis
Erion-Vasilis Pikoulis
AMAristeidis Mpifis
Aristeidis Mpifis
Maria TrigkaMaria Trigka
Maria Trigka
CCConstantinos Constantinopoulos
Constantinos Constantinopoulos
DKDimitrios Kosmopoulos
Dimitrios Kosmopoulos
Sign language (SL) translation constitutes an extremely challenging task when undertaken in a general unconstrained setup, especially in the absence of vast training datasets that enable the use of end-to-end solutions employing deep architectures. In such cases, the ability to incorporate prior information can yield a significant improvement in the translation results by greatly restricting the search space of the potential solutions. In this work, we treat the translation problem in the limited confines of psychiatric interviews involving doctor-patient diagnostic sessions for deaf and hard of hearing patients with mental health problems.To overcome the lack of extensive training data and be able to improve the obtained translation performance, we follow a domain-specific approach combining data-driven feature extraction with the incorporation of prior information drawn from the available domain knowledge. This knowledge enables us to model the context of the interviews by using an appropriately defined hierarchical ontology for the contained dialogue, allowing for the classification of the current state of the interview, based on the doctor’s question. Utilizing this information, video transcription is treated as a sentence retrieval problem. The goal is predicting the patient’s sentence that has been signed in the SL video based on the available pool of possible responses, given the context of the current exchange. Our experimental evaluation using simulated scenarios of psychiatric interviews demonstrate the significant gains of incorporating context awareness in the system’s decisions.

4 April 2020 in Journal of Geophysical Research: Solid Earth
Verified
Journal Article
Open access
PDF file
EPErion-Vasilis Pikoulis
Erion-Vasilis Pikoulis
OKOlga Ktenidou
Olga Ktenidou
EPEmmanouil Z. Psarakis
Emmanouil Z. Psarakis
NANorman A. Abrahamson
Norman A. Abrahamson
We present a framework for stochastically modeling the Fourier spectrum of the noisy seismic recording based on the fundamental assumption that the latter constitutes a random rather than a deterministic quantity. First we demonstrate (mathematically and through simulations) that, under the stochastic‐signal assumption, the periodogram ordinates of the noisy recording can be considered independent exponential random variables with a frequency‐dependent mean. Using this finding, the estimation of seismological parameters is translated into a well‐defined maximum likelihood (ML) problem, allowing a fast, accurate, and robust solution. Although the proposed ML methodology constitutes a general estimation framework that we believe can improve any spectral analysis, here we specifically apply it to the high‐frequency attenuation parameter (κ), crucial for understanding rock ground motion. Other seismological parameters could be estimated by appropriately adapting the theoretical model and frequency band used. The greatest advantage of the proposed method is its ability to account for the presence of noise rather than simply try to avoid it, as is the case with all conventional κ approaches and many other spectral analysis approaches. This means that the proposed technique achieves acceptable results even for very low signal‐to‐noise ratios (SNR), thus pushing the boundary of what can be considered usable recording quality. The technique's superior performance in estimation is demonstrated through a series of experiments involving both synthetic and real seismic recordings. Our results also indicate that, compared to this new “noise‐modeling” approach, conventional “noise‐avoiding” approaches are likely to systematically underestimate the “true” kappa value in recordings of moderate‐to‐low SNR.

13 January 2026 in Natural Hazards and Earth System Sciences
Verified
Journal Article
Open access
PDF file
OKOlga Ktenidou
Olga Ktenidou
EPErion-Vasilis Pikoulis
Erion-Vasilis Pikoulis
APAntonia Papageorgiou
Antonia Papageorgiou
FGFevronia Gkika
Fevronia Gkika
SLSpyros Liakopoulos
Spyros Liakopoulos
CEChristos Evangelidis
Christos Evangelidis
The assumption of reference station conditions is investigated for the first time across 60 rock stations belonging to the broadband and accelerometric networks of the National Observatory of Athens. We include in our assessment all stations that have some probability of lying on rock, based on existing data or beliefs, provided their data have been publicly available for long enough to yield a substantial number of recordings. No studies on site effects have been conducted before for the ensemble of stations. Furthermore, almost no ad hoc field campaigns have been performed to characterise them. The first step is to compile all readily available information per station from publicly available external sources, i.e. geology, topography, housing, Vs30 estimates, and any other known metadata. The second step is to analyse geological maps to derive the geological unit and age; to combine this external information with internal information (namely questioning network staff to access the operator's first-hand experience of the sites); and to better describe geology, geomorphology, and station installation details. The third and largest step is to compile the first Greek ground motion dataset on rock and to perform a detailed analysis of the recordings to estimate site-specific amplification to assess the local site response for each station. A strong-motion dataset of over 7500 recordings is developed and curated for this purpose, dating from 2012 to 2023. It is visually inspected and meticulously processed on a waveform-specific basis in the time and frequency domains, paying special attention to signal quality and strength. Single-station amplification functions using horizontal-to-vertical spectral ratios (HVSRs) are then estimated from the database. Considering that “true” reference sites should have low, flat amplification with no directional dependence, the analysis goes beyond the usual path of combining the two horizontal components into a mean HVSR. It also assesses the directional sensitivity of the HVSR to identify departure from the 1D assumption, corrects the HVSR for the vertical amplification effect, and uses clustering techniques to select groups of stations with different response characteristics. This data-derived characterisation is combined with the previously compiled station metadata to evaluate the stations' overall capacity as reference sites. This results in a qualitative ranking of the stations. The least and most adequate reference stations are showcased to facilitate a better use of seismic data in future seismological and hazard applications.