Computer vision (CV) and machine learning (ML) offer new tools for cultural heritage (CH) artifact analysis, but the CV/ML pipeline remains largely inaccessible to CH domain experts, who lack the background to configure, train, or assess models. We present AmalthAI, an open-source CV platform that bridges this gap, enabling non-ML CH experts to independently produce and validate archaeologically meaningful findings. The interface covers dataset management, training, and inference for classification, segmentation, and object detection, with Kubeflow and Katib handling scalable training and hyperparameter search. Grad-CAM localizes the image region behind a prediction, and a vision-language model (VLM) adds a text description of it for expert review. Since archaeological data is often state-owned or rights-encumbered and cannot leave institutional custody, AmalthAI's self-hostable deployment ensures sensitive data is kept within premises. We test the platform on an archaeological use case built on a custom dataset of clay textile imprints, where CH experts trained and validated segmentation, and classification models for hypothesis testing. We provide the implementation code at https://github.com/TEXTaiLES/AmalthAI.
Christos Chatzisavvas, Stelios Alvanos, Efstratios Politis et al.· 0 citations
Background Recent advances in artificial intelligence (AI) have accelerated the development of educational technologies intended to support students’ learning processes. However, classroom-based evidence on the educational impact of purpose-built AI applications remains limited, particularly across national contexts. Methods This study examined the effects of an AI conversational study assistant (Study Buddy) on secondary students’ academic performance, motivation, and engagement in authentic school settings. A quasi-experimental design was employed across two case studies conducted in Cyprus (Grade 10 Physics, N = 47) and Greece (Grade 7 History, N = 70). In each context, intact classes were assigned to experimental and control conditions. Academic performance was assessed using curriculum-aligned teacher-developed tests, while motivation and engagement were measured using the Motivation and Engagement Survey. In addition, system-generated usage logs were analyzed descriptively to document student access, interaction intensity, and patterns of tool use during the intervention. Statistical analyses included descriptive statistics and parametric or non-parametric group comparisons depending on distributional assumptions. Results Students who used Study Buddy demonstrated significantly higher learning gains in Physics and higher post-test scores in History compared to peers in control groups. In contrast, no statistically significant differences were observed for motivation or engagement in either case study. Usage log analysis indicated that students actively engaged with the application and that interaction patterns reflected the instructional design of each case study. Conclusions The findings suggest that purpose-built AI study assistants can support academic learning when integrated into regular classroom instruction. However, short-term exposure and predominantly task-focused interactions may limit their influence on motivational and engagement-related outcomes. The study contributes classroom-based, cross-national evidence on educational AI tools and highlights the importance of instructional design and teacher mediation in shaping both usage patterns and learning outcomes.
Theodoros Karafyllidis, Anna Vacalopoulou, S. Stamouli et al.· Open Research Europe· 0 citations
Automatic Lyric Transcription (ALT) remains substantially more challenging than speech recognition due to melodic variability, rhythmic irregularity, and accompaniment interference. This is heightened in low-resource languages like Greek, where no prior benchmark for ALT exists. We present the first controlled study of Whisper adaptation for Greek ALT, investigating model scaling effects, task composition via multitask training in transcribe-translate ratios, and two-stage speech-to-singing adaptation. We also curate a segment-level aligned singing dataset based on the Greek Audio Dataset (GAD) using source separation and CTC forced alignment. Results show that scaling consistently improves performance, while multitask learning acts as a beneficial regularizer primarily for smaller-capacity models. The 2-stage adaptation in Whisper Large-v3 achieves a Word Error Rate (WER) of 27.2%, a significant improvement over zero-shot baselines, establishing the first Greek ALT benchmark.
Maria Frangiadaki, Dimitrios Damianos, Kosmas Kritsis et al.· 0 citations
The Contrastive Routing Mechanism (CoRM) is proposed, which scores each expert by the gap between its affinity for the incoming token and its affinity for this shared reference state, interpreted through a distinct per-expert projection.
N. Xiros, Dimitrios Damianos, Maria-Eleni Zoumpoulidi et al.· 0 citations
A lifecycle model for LLM systems is proposed that supports security analysis by structuring it around security-relevant boundaries rather than workflow optimisation, and is supported by a 12-stage LLMOps pillar and a 9-category governance pillar.
Eleftherios Batzolis, George Drosatos, V. Katsouros et al.· ARES· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
