108495 research outputs found
Sort by
Voices of Luxembourg: Tackling Dialect Diversity in a Low-Resource Setting
Get PDFDialect classification is essential for preserving linguistic diversity, particularly in low-resource languages such as Luxembourgish. This study introduces one of the first systematic approaches to classifying Luxembourgish dialects, addressing phonetic, prosodic, and lexical variations across four major regions. We benchmarked multiple models, including state-of-the-art pre-trained speech models like Wav2Vec2, XLSR-Wav2Vec2, and Whisper, alongside traditional approaches such as Random Forest and CNN-LSTM. To overcome data limitations, we applied targeted data augmentation strategies and analyzed their impact on model performance. Our findings highlight the superior performance of CNN-Spectrogram and CNN-LSTM models while identifying the strengths and limitations of data augmentation. This work establishes foundational benchmarks and provides actionable insights for advancing dialectal NLP in Luxembourgish and other low-resource languages.https://aclanthology.org/2025.resourceful-1.0
VerbCraft: Morphologically-Aware Armenian Text Generation Using LLMs in Low-Resource Settings
No full textUnderstanding and generating morphologically complex verb forms is a critical challenge in Natural Language Processing (NLP), particularly for low-resource languages like Armenian. Armenian's verb morphology encodes multiple layers of grammatical information, such as tense, aspect, mood, voice, person, and number, requiring nuanced computational modeling. We introduce VerbCraft, a novel neural model that integrates explicit morphological classifiers into the mBART-50 architecture. VerbCraft achieves a BLEU score of 0.4899 on test data, compared to the baseline's 0.9975, reflecting its focus on prioritizing morphological precision over fluency. With over 99\% accuracy in aspect and voice predictions and robust performance on rare and irregular verb forms, VerbCraft addresses data scarcity through synthetic data generation with human-in-the-loop validation. Beyond Armenian, it offers a scalable framework for morphologically rich, low-resource languages, paving the way for linguistically informed NLP systems and advancing language preservation efforts
Case error corrections for noun phrases containing deverbal attributive nouns in Greenlandic
No full textThis paper contains preliminary findings using Constraint Grammar (CG) in semantic annotation in a specific type of noun phrases in Greenlandic, in which the attributive noun is a nominalized predicative verbal stem. The annotation is used in a grammar checker pipeline for the purpose of making case error correction suggestions
Sõnaraamat kui üks muukeelsete laste õppevahend vene keele kui võõrkeele tundides põhikoolis: harjutusvara suulise ja kirjaliku väljendusoskuse arendamiseks (A1-A2 tasemel)
No full texthttps://www.ester.ee/record=b574033
Margins in Contrastive Learning: Evaluating Multi-task Retrieval for Sentence Embeddings
No full textThis paper explores retrieval with sentence embeddings by fine-tuning sentence-transformer models for classification while preserving their ability to capture semantic similarity. To evaluate this balance, we introduce two opposing metrics – polarity score and semantic similarity score – that measure the model's capacity to separate classes and retain semantic relationships between sentences. We propose a system that augments supervised datasets with contrastive pairs and triplets, training models under various configurations and evaluating their performance on top- sentence retrieval. Experiments on two binary classification tasks demonstrate that reducing the margin parameter of loss functions greatly mitigates the trade-off between the metrics. These findings suggest that a single fine-tuned model can effectively handle joint classification and retrieval tasks, particularly in low-resource settings, without relying on multiple specialized models
Mitokondriaalse dünaamika roll Wolframi sündroomi korral
Get PDFVäitekirja elektrooniline versioon ei sisalda publikatsiooneNeuroarengulised ja neurodegeneratiivsed haigused on tänapäeva ühiskonnas oluliseks elukvaliteedi vähendajaks ja majanduslikuks koormaks. Seniste uuringute jooksul on saanud selgemaks, et ajuhaiguste väljakujunemises võib olla oluline osa mitokondrite ehituse ja/või talitluse kõrvalekalletel. Mitokondrid on dünaamilised organellid, mis tähendab, et nad liiguvad aktiivselt mööda närvirakkude jätkeid ringi, ühinevad (fusioon) ja lahknevad (fission) omavahel ning vigased/üleliigsed mitokondrid suunatakse lagundamisele (mitofaagia). Wolframi sündroom (WS) on harvikhaigus, mis avaldub progressiivsete neuroendokriinsete kõrvalekalletena, nagu magediabeet, diabeet, optiline atroofia ja kurtus. Haiguse avastamise ja kirjeldamise järel peeti seda kauaaegselt mitokondriaalseks haiguseks. Hiljem avastati, et WS põhjustavad mutatsioonid WFS1 (WS1) ja CISD2 (WS2) geenides, mille poolt kodeeritud valgud paiknevad endoplasmaatilise retiikulumi (ER) ja mitokondrite ühenduskohtades. Senini ei ole teada, millised on WFS1 ja CISD2 täpsed ülesanded rakkudes. WS1 peetakse peamiselt ER haiguseks, sest varem on leitud WFS1 osalemist ER stressi ning rakusisese kaltsiumi tasakaalu regulatsioonis.
Käesoleva uurimistöö eesmärgiks oli kirjeldada mitokondriaalse dünaamika rolli WS rakuliste mehhanismide hulgas kasutades katseklaasimudelit ja otsida võimalikke farmakoloogilisi sekkumisi, mis aitaksid leevendada tekkinud kõrvalekaldeid mitokondrite talitluses ja närvirakkude arengus.
Uurimistöös leiti, et WFS1 puuduse korral vähenes mitokondrite fusioonide sagedus ja mitokondrid olid lühenenud. Lisaks oli suurenenud mitokondrite lagundamine ja seeläbi vähenenud ka mitokondrite tihedus närvirakkude aksonites. Mitokondrite dünaamika muutuseid seostati membraanipotentsiaali languse ja ATP tootmise vähenemisega, mis omakorda viis närvirakkude arengu aeglustumiseni. Seejärel leiti, et kõrvalekalded mitokondrite dünaamikas ja bioenergeetikas olid seotud häirunud kaltsiumi tasakaaluga ER-s ja vähenenud kaltsiumi vooludega ER-st mitokondritesse. WFS1 puuduse korral esinevaid kõrvalekaldeid mitokondrite dünaamikas ja närvirakkude arengus aitasid leevendada keemilised ained, mis taastasid kaltsiumi tasakaalu ER-s ja mitokondrites.
Kokkuvõttes laiendavad antud doktoritöö tulemused praeguseid teadmisi WS rakulistest mehhanismidest ning võivad olla abiks tulevikus uute ravimite väljatöötamisel. Antud tulemustest järeldub, et neuroloogiliste sümptomite väljakujunemisel võib olla oluline roll vähenenud energiatootmisel närvirakkudes, mille põhjusteks on omakorda nii mitokondrite dünaamikahäired kui ka ER-mitokondrite kaltsiumi voolude vähenemine.Neurodevelopmental and neurodegenerative disorders are both growing major public health concerns causing substantial burdens. Interestingly, mitochondrial abnormalities are regularly described as an early pathological feature in common neurodegenerative diseases. Moreover, mitochondrial dynamics, including fusion and fission, trafficking, and mitophagy, are crucial for neuronal development. In this context, we investigated an in vitro neuronal model of Wolfram syndrome (WS), a progressive autosomal recessive neuroendocrine disorder characterized by diabetes mellitus, optic atrophy, diabetes insipidus, and deafness. Interestingly, WS was initially proposed to be a mitochondriopathy, but later studies located the two responsible proteins, WFS1 (WS1) and CISD2 (WS2), to endoplasmatic reticulum (ER) membranes forming contacts with mitochondria. WS1 is primarily considered an ER disease affecting Ca2+ homeostasis and ER stress regulation.
This thesis aimed to elucidate the role of mitochondrial dynamics in the pathogenesis of WS and find possible pharmacological interventions to alleviate the impaired mitochondrial phenotype and delayed neuronal development.
Our studies found decreased mitochondrial fusion rate and fragmentation alongside increased mitochondrial elimination and reduced density in axons of WFS1-deficient neurons. The disruption of mitochondrial dynamics adversely affected cellular bioenergetics and decreased axonal ATP levels leading to delayed neuronal development. The perturbations in mitochondrial dynamics and bioenergetics in an in vitro model of WS were connected to the impaired ER Ca2+ homeostasis and diminished ER-mitochondria Ca2+ flux. The chemical compounds activating Ca2+ uptake to the ER and the mitochondria were able to restore mitochondrial dynamics and rescue neuronal development.
In conclusion, novel insights into the role of mitochondrial dysfunction in the neuronal pathogenesis of Wolfram syndrome 1 broaden our understanding of the disease's mechanisms and provide new targets for pharmacological interventions.https://www.ester.ee/record=b573997
The 1st Workshop on Nordic-Baltic Responsible Evaluation and Alignment of Language Models. Proceedings of the Workshop
Get PDF
MorSeD: Morphological Segmentation of Danish and its Effect on Language Modeling
No full textCurrent language models (LMs) mostly exploit subwords as input units based on statistical co-occurrences of characters. Adjacently, previous work has shown that modeling morphemes can aid performance for Natural Language Processing (NLP) models. However, morphemes are challenging to obtain as there is no annotated data in most languages. In this work, we release a wide-coverage Danish morphological segmentation evaluation set. We evaluate a range of unsupervised token segmenters and evaluate the downstream effect of using morphemes as input units for transformer-based LMs. Our results show that popular subword algorithms perform poorly on this task, scoring at most an F1 of 57.6 compared to 68.0 for an unsupervised morphological segmenter (Morfessor). Furthermore, evaluate a range of segmenters on the task of language modeling