Protein language models (pLMs) typically tokenize sequences at the single-amino-acid level using a 20-residue alphabet, resulting in long input sequences and high computational cost. Sub-word tokenization methods such as Byte Pair Encoding (BPE) can reduce sequence length but are limited by the sparsity of long patterns in proteins encoded by the standard amino acid alphabet. Reduced amino acid alphabets, which group residues by physicochemical properties, offer a potential solution but their performances with sub-word tokenization have not been systematically studied. We investigate the combined use of reduced amino acid alphabets and BPE tokenization in protein language models. We pre-train RoBERTa-based pLMs de novo using multiple reduced alphabets and evaluate them across diverse downstream tasks. Our results show that reduced alphabets enable substantially shorter input sequences and faster training and inference. These findings suggest that alphabet reduction may facilitate more effective sub-word tokenization, enabling increased efficiency with marginal impact on predictive performance, and for specific tasks even improving accuracy. Models, tokenizers, and code are available at github.com/burstein-lab/BioTokenizers.
Sign language plays a vital role in enabling interaction for individuals with hearing and speech impairment, making accurate alphabet-level recognition a fundamental requirement for accessible human-computer interaction systems. This paper investigates visual deep learning approaches for interpreting sign language alphabets from image data. A systematic framework is developed to assess the effectiveness of transfer learning-based convolutional neural networks in capturing discriminative hand gesture features. Using a benchmark American Sign Language (ASL) alphabet dataset, several advanced architectures, including ConvNeXtXLarge, EfficientNet, VGG19, and ResNet-50, are examined under a unified experimental protocol. The comparative analysis reveals that ConvNeXtXLarge achieves the uppermost recognition performance, attaining an accuracy of 99.81%, while EfficientNet, VGG19, and ResNet-50 also demonstrate strong results with accuracy of 99.68%, 99.31%, and 97.29%, respectively. These findings emphasize the effectiveness of modern transfer learning strategies in enhancing visual representation learning for fine-grained motion recognition. The proposed evaluation framework offers practical insights into model selection for scalable and reliable sign language interpretation systems, contributing to the advancement of inclusive assistive technologies and real-world visual language understanding applications.
Lip reading technology has potential use across various fields, significantly enhancing communication for the deaf, aiding in noisy settings, and supporting information security through silent password entry. Although, notable progress has been made in constructing datasets of different types like digits, alphabets, words, phrases, and sentences levels for lip reading in various languages. However, developing a robust Urdu lip reading model remains a challenge due to the lack of a suitable dataset. Moreover, difficulties in adapting previous models, such as the LipNet model to Urdu. To address these barriers, we present the ULRA (Urdu lip reading alphabets) dataset, leverage advanced data augmentation techniques, and evaluate three cutting-edge DNN models: a LipNet-based 2D-CNN model, a Hybrid 2D_3D-CNN model, and a baseline 3D-CNN model. Each model undergoes rigorous testing in diverse environments, with both familiar and unfamiliar data. The results reveal that the LipNet-based 2D-CNN model achieves an impressive 81.97% accuracy on unknown data across diverse environments, while the Hybrid model excels in generalization, reaching 69.45% accuracy on unfamiliar data, thanks to its superior spatiotemporal feature extraction capabilities. Additionally, precision, recall, and F1-Score values of LipNet-Based 2D CNN are 0.83, 0.82, and 0.82 respectively. All three values of this model are also higher than the other two models. These findings underscore the strengths of various DNN architectures and the critical advancements made possible by the ULRA dataset, paving the way for future breakthroughs in Urdu lip reading research.
Sign language (SL) is mostly used by hearing-impaired people to communicate. Each country has its own SL. Usually, SL is structured in terms of its alphabet, words, and grammar. In Mexico, there are few investigations on Mexican Sign Language (MSL) automatic recognition, and published datasets are rare. To contribute to this research, we obtained a dataset of 21 static letters from the MSL alphabet. For each letter, 15 individuals performed the task once, and the data were collected using a 3D surface sensor to obtain the coordinates of the 3D points. All individuals used their right hand to perform the signs. The dataset contains 315 text files with 3D coordinates of the letters.
This data article presents two complementary datasets that together cover the complete 27-letter Mexican Sign Language (Lengua de Señas Mexicana, LSM) alphabet, encompassing both static and dynamic signs. The first dataset contains 279,716 JPEG images of the 21 static signs (letters A-I, L-P, R-U, W, Y) captured from 20 participants across three levels of hand-pose variation. The second dataset contains 1,200 MP4 videos of the 6 dynamic signs (letters J, K, Ñ, Q, X, Z) recorded from 20 different participants in two camera views (frontal and profile), with five repetitions per sign. Both datasets were collected under controlled conditions using a uniform green background and constant illumination, totaling approximately 8.4 GB. Each dataset includes predefined participant-based train/test partitions for signer-independent evaluation. The datasets have been used in prior work on fingerspelling recognition [3] and unified alphabet classification [4]. The data are publicly available in the Zenodo repository under a CC-BY 4.0 license.
The integration of [177Lu]Lu-prostate-specific membrane antigen (PSMA) therapy into the treatment paradigm for advanced prostate cancer has improved survival and quality of life; however, resistance and disease progression remain inevitable. Radionuclide combination strategies, particularly α-/β-emitter pairings, aim to overcome biological and dosimetric limitations of [177Lu]Lu-PSMA monotherapy. The AlphaBet trial demonstrated the feasibility and safety of combining the bone-seeking α-emitter radium-223 with [177Lu]Lu-PSMA in metastatic castration-resistant prostate cancer, underpinned by complementary spatial targeting and radiophysical properties. Ongoing studies are evaluating other novel combinations to overcome resistance, earlier integration to define optimal sequencing, and biomarker-driven approaches to refine patient selection. For patients, these strategies may enable more effective and durable treatments by combining different types of targeted radiation or systemic therapies; however, further studies are needed to confirm long-term benefits and safety.
Languages provide social-category markers that tag people as one or another social group. How does the brain sort words into different language categories as a basis of the social-categorization function of language? The current work addressed this issue by testing neural categorization of visual words of different writing systems in nine studies using electroencephalography, magnetoencephalography, and a repetition suppression paradigm. This work showed that a neural network, including the anterior temporal, insular, orbital frontal, and ventral occipito-temporal cortices in both hemispheres, was engaged in computations of correlation distances between two words to represent intra-language similarity and inter-language difference during categorization of visual words of alphabetic and non-alphabetic languages. These processes occurred as early as 150 ms post-stimulus, recruited within-hemisphere functional connections, operated independently of words' semantic meanings and pronunciations, and exhibited consistently across individuals with diverse language backgrounds. These findings highlight the neural mechanisms of language-based spontaneous neural categorization of visual words as a basis of the social-categorization function of language.
The sequence-specific recognition of double-stranded DNA by biocompatible molecules is fundamental to molecular medicine and synthetic biology. Triplex-forming oligonucleotides (TFOs) enable programmable major groove recognition via Hoogsteen base pairing; however, the limited repertoire of natural nucleobases imposes strict constraints on target sequences and parallel motif triplexes require acidic conditions for stability. Here, we have expanded the triplex recognition space using nucleobases from an artificially expanded genetic information system (AEGIS). Through a systematic evaluation of 120 base triad combinations, we identify at least 12 modular triads that can be combined interchangeably to target duplex DNA containing standard, damaged, or synthetic base pairs with nanomolar affinity at neutral pH. We further demonstrate the versatility of this expanded recognition code by detecting oxidative lesions or AEGIS base pairs in enzymatically assembled duplex constructs using both chemically and enzymatically synthesized TFOs. This generalized framework provides a robust platform for precision gene-targeting, molecular sensing, and nucleic acid nanotechnology.
CLN3 disease is an inherited neurodegenerative disease, typically with childhood onset, and characterized by vision loss, seizures, cognitive decline, and difficulties. The CLN3 Staging System (CLN3SS) characterizes disease progression. Our aim was to assess differences in cognitive test scores in relation to CLN3SS among individuals with CLN3 disease. We evaluated the relationship between cognitive test performance and the CLN3SS in individuals with genetically confirmed CLN3 disease. Participants completed tasks of verbal reasoning, vocabulary knowledge, attention, fund of information, and ability to recite the alphabet. One-way ANOVA testing assessed differences in mean cognitive test score among CLN3SS score groups, and Chi-square testing was used to compare the proportion in each CLN3SS group that could recite the alphabet. Data were evaluated from a sample of 85 individuals with a total 245 CLN3SS assessments conducted within 6 months of their cognitive testing, A significant decrease in test scores was found between CLN3SS Stages 1 (vision loss present) and 2 (vision loss and seizures present) for each of the cognitive tests. The proportion of participants able to recite the alphabet also decreased from Stage 1 to Stage 2 (Χ 2 =12.1, p<.01). Cognitive ability declines with advanced disease severity in CLN3 disease, though motor disability in Stage 3 likely contributes to difficulty participating in cognitive assessment at this later disease stage. Understanding the relationship between cognition and CLN3 disease stage may help guide decision making, i.e., determining who could or should undergo cognitive assessment for clinical care or for group stratification in disease modifying clinical trials. Cognitive ability declines with advanced disease severity in CLN3 disease.
Genomic regulation depends on coordinated interactions among DNA sequence, chromatin state, epigenetic marks, three-dimensional architecture, and cellular context. Yet formal approaches for representing higher-order regulatory organization across these layers remain limited. Here, we propose a discrete hierarchical coding framework to examine whether regulatory systems exhibit reusable state constraints beyond sequence-level coding. The model defines an eight-symbol alphabet, including a re-entry symbol DO' for hierarchical closure, and yields a 64-state space of ordered dominant-contextual pairs. This structure captures directionality and separates coarse-grained regulatory identity from modulatory context. From this construction, we derive formal properties including closure, non-commutativity, modular partitions, self-consistency classes, and entropy constraints. We outline how dominant components may correspond to regulatory identities, whereas contextual components may represent epigenetic, chromatin, architectural, or environmental modulation. The framework yields empirically testable expectations, including constrained state-space occupancy, recurrent state-symbols, dominant-contextual separability, and structured transitions during differentiation, reprogramming, and system-level reorganization involving changes in three-dimensional genome architecture or karyotype. We propose the model as a complementary and falsifiable scaffold for studying higher-order coding in genomic regulation. If empirically supported, the proposed alphabet may constitute an initial symbolic layer of a broader regulatory grammar governing transitions among regulatory states.
Music cognition research has explored whether and how music training improves word reading. However, extant theories and empirical studies have only focused on alphabetical languages, limiting the current theoretical understanding to alphabetical languages alone. This cross-sectional study evaluated (1) whether musically trained children outperformed untrained children in Chinese word reading and (2) the relative mediating roles of segmental phonological and tone awareness in the relation between music perception and Chinese word reading in musically trained children. We recruited 86 musically trained children and untrained children with similar family income. They were tested on Chinese word reading, segmental phonological awareness, tone awareness, and music perception. Musically trained children outperformed musically untrained children on Chinese word reading, segmental phonological awareness, and tone awareness. Segmental phonological awareness mediated the relation between music perception and word reading in musically trained children, whereas tone awareness did not. This study provides initial evidence supporting the potential of music training to improve Chinese word reading, laying an empirical foundation for future music training studies. Moreover, the findings reflect a possible intervention mechanism in which music training improves Chinese word reading by enhancing segmental phonological awareness. From a theoretical perspective, the findings support the segmental phonological awareness account but not the tone awareness account.
Artificial intelligence (AI) tools based on natural language, such as ChatGPT 4.1 mini (OpenAI Group PBC) and Gemini 2.5 Flash (Alphabet Inc.), are used by patients as sources of medical information. The current study aimed to evaluate and compare the quality and readability of responses provided by these AIs, in Brazilian Portuguese, regarding rotator cuff surgery. The present cross-sectional, descriptive, and comparative study followed qualitative and quantitative approaches. A total of 24 frequently-asked patient questions were used, classified according to Rothwell. Each question was entered individually into both platforms, and only the first response was considered. The quality assessment used the DISCERN instrument, developed by the University of Oxford and the British Library, and the Journal of the American Medical Association (JAMA) benchmark criteria. Readability was estimated using Análise de Legibilidade Textual (ALT, "Text Readibility Anallysis", in Portuguese) software, validated for Brazilian Portuguese. The statistical analyses included the Wilcoxon and Friedman tests, repeated-measures analysis of variance (ANOVA), and the Conover post-hoc test with Bonferroni correction. ChatGPT achieved a mean DISCERN score of 58.7 ± 4.0, and Gemini, 56.3 ± 3.5, with no significant difference ( p  = 0.174), but with a maximum effect size (rank-biserial correlation [rrb] = 1.0). Both models showed a mean readability corresponding to 13.3 years of schooling ( p  = 1.000). No response met the JAMA benchmark criteria. Value-based questions achieved the highest quality scores, whereas policy-related questions were the most complex in terms of readability. The correlation between quality and readability was moderate (ρ = 0.73; p  = 0.099). ChatGPT 4.1 mini and Gemini 2.5 Flash do not yet provide adequate medical information in Brazilian Portuguese regarding editorial reliability, quality, and textual accessibility for the general public. Ferramentas de inteligência artificial (IA) baseadas em linguagem natural, como ChatGPT-4.1 mini (OpenAI Group PBC) e Gemini 2.5 Flash (Alphabet Inc.), são utilizadas por pacientes como fonte de informação médica. Este estudo avaliou e comparou a qualidade e a legibilidade das respostas fornecidas por essas IAs, em português brasileiro, sobre cirurgia do manguito rotador. Estudo transversal, descritivo e comparativo, com abordagem qualiquantitativa. Foram utilizadas 24 perguntas frequentes de pacientes, classificadas segundo Rothwell. Cada pergunta foi inserida individualmente nas plataformas dos dois modelos, sendo considerada apenas a primeira resposta. A qualidade foi avaliada por meio do instrumento DISCERN, desenvolvido pela University of Oxford e pela British Library, e dos critérios editoriais da Journal of the American Medical Association (JAMA). A legibilidade foi estimada com o programa Análise de Legibilidade Textual (ALT), validado para o português brasileiro. As análises estatísticas incluíram os testes de Wilcoxon, Friedman, análise de variância ([ analysis of variance , ANOVA, em inglês] para medidas repetidas) e post hoc de Conover com correção de Bonferroni. O ChatGPT obteve escore médio DISCERN de 58,7 ± 4,0, e o Gemini, 56,3 ± 3,5, sem diferença significativa ( p  = 0,174), mas com efeito máximo ( rank-biserial correlation [rrb, em inglês] = 1,0). Ambos os modelos apresentaram legibilidade média correspondente a 13,3 anos de escolaridade ( p  = 1,000). Nenhuma resposta atendeu aos critérios editoriais da JAMA. Perguntas relacionadas a valores obtiveram os maiores escores de qualidade, ao passo que as perguntas sobre política foram as mais complexas em termos de leitura. A correlação entre qualidade e legibilidade foi moderada (ρ = 0,73; p  = 0,099). ChatGPT–4.1 mini e Gemini 2.5 Flash ainda não oferecem informação médica, em português brasileiro, adequada quanto à confiabilidade editorial, qualidade e acessibilidade textual para o público leigo.
Artificial intelligence (AI) tools based on natural language, such as ChatGPT 4.1 mini (OpenAI Group PBC) and Gemini 2.5 Flash (Alphabet Inc.), are used by patients as sources of medical information. The current study aimed to evaluate and compare the quality and readability of responses provided by these AIs, in Brazilian Portuguese, regarding rotator cuff surgery. The present cross-sectional, descriptive, and comparative study followed qualitative and quantitative approaches. A total of 24 frequently-asked patient questions were used, classified according to Rothwell. Each question was entered individually into both platforms, and only the first response was considered. The quality assessment used the DISCERN instrument, developed by the University of Oxford and the British Library, and the Journal of the American Medical Association (JAMA) benchmark criteria. Readability was estimated using Análise de Legibilidade Textual (ALT, "Text Readibility Anallysis", in Portuguese) software, validated for Brazilian Portuguese. The statistical analyses included the Wilcoxon and Friedman tests, repeated-measures analysis of variance (ANOVA), and the Conover post-hoc test with Bonferroni correction. ChatGPT achieved a mean DISCERN score of 58.7 ± 4.0, and Gemini, 56.3 ± 3.5, with no significant difference ( p  = 0.174), but with a maximum effect size (rank-biserial correlation [rrb] = 1.0). Both models showed a mean readability corresponding to 13.3 years of schooling ( p  = 1.000). No response met the JAMA benchmark criteria. Value-based questions achieved the highest quality scores, whereas policy-related questions were the most complex in terms of readability. The correlation between quality and readability was moderate (ρ = 0.73; p  = 0.099). ChatGPT 4.1 mini and Gemini 2.5 Flash do not yet provide adequate medical information in Brazilian Portuguese regarding editorial reliability, quality, and textual accessibility for the general public. Ferramentas de inteligência artificial (IA) baseadas em linguagem natural, como ChatGPT-4.1 mini (OpenAI Group PBC) e Gemini 2.5 Flash (Alphabet Inc.), são utilizadas por pacientes como fonte de informação médica. Este estudo avaliou e comparou a qualidade e a legibilidade das respostas fornecidas por essas IAs, em português brasileiro, sobre cirurgia do manguito rotador. Estudo transversal, descritivo e comparativo, com abordagem qualiquantitativa. Foram utilizadas 24 perguntas frequentes de pacientes, classificadas segundo Rothwell. Cada pergunta foi inserida individualmente nas plataformas dos dois modelos, sendo considerada apenas a primeira resposta. A qualidade foi avaliada por meio do instrumento DISCERN, desenvolvido pela University of Oxford e pela British Library, e dos critérios editoriais da Journal of the American Medical Association (JAMA). A legibilidade foi estimada com o programa Análise de Legibilidade Textual (ALT), validado para o português brasileiro. As análises estatísticas incluíram os testes de Wilcoxon, Friedman, análise de variância ([ analysis of variance , ANOVA, em inglês] para medidas repetidas) e post hoc de Conover com correção de Bonferroni. O ChatGPT obteve escore médio DISCERN de 58,7 ± 4,0, e o Gemini, 56,3 ± 3,5, sem diferença significativa ( p  = 0,174), mas com efeito máximo ( rank-biserial correlation [rrb, em inglês] = 1,0). Ambos os modelos apresentaram legibilidade média correspondente a 13,3 anos de escolaridade ( p  = 1,000). Nenhuma resposta atendeu aos critérios editoriais da JAMA. Perguntas relacionadas a valores obtiveram os maiores escores de qualidade, ao passo que as perguntas sobre política foram as mais complexas em termos de leitura. A correlação entre qualidade e legibilidade foi moderada (ρ = 0,73; p  = 0,099). ChatGPT–4.1 mini e Gemini 2.5 Flash ainda não oferecem informação médica, em português brasileiro, adequada quanto à confiabilidade editorial, qualidade e acessibilidade textual para o público leigo.
Children with attention deficit/hyperactivity disorder and comorbid developmental dyslexia (ADHD + DD) demonstrate both core symptoms of ADHD and pronounced reading difficulties, which further compromise executive function (EF). This study compared the effects of open-versus closed-skill exercises (OSE vs. CSE) on EF in children with ADHD + DD. Thirty children (6.9-8.5 years) with ADHD + DD were randomly assigned to OSE (n = 15) or CSE (n = 15) training, with 15 typically developing (TD) children receiving CSE. All completed 12-week, thrice-weekly, moderate-to-high intensity (> 60% VO₂ max) sessions. OSE performed table tennis; CSE did track-and-field. EF was assessed with SCWT, CFT, and TMT; visual perception used DTVP-3. Data were analyzed by two-way repeated-measures ANOVA, with ANCOVA used to adjust baseline differences. Compared with TD peers, ADHD + DD children exhibited impairments in inhibitory control (including Stroop A/B/C/D RT, Stroop B/D error count, and word interference time), all working memory indices, cognitive flexibility (digit-alphabet linking time), and all visual perception measures (all p < 0.05). After intervention the outcomes revealed significant Group × Time differences in EF, including inhibitory control (Stroop B/D RT, word interference time), working memory (delayed detail and structural memory), and cognitive flexibility (digit-alphabet linking time) (p < 0.05). Post-hoc tests confirmed that OSE yielded significantly greater improvements than CSE. For secondary visual perception outcomes, all measures showed significant Group × Time interactions (all p < 0.05). Further adjusted analyses demonstrated that OSE outperformed CSE significantly in visual-motor integration (VMI), copying ability, motor-reduced visual perception (MRP), form consistency, and general visual perception (GVP) (all p < 0.05). OSE (table tennis training) has superior efficacy in improving EF and visual perception in children with ADHD + DD compared with CSE (track-and-field training), providing preliminary empirical evidence for a feasible non-pharmacological intervention strategy for this population.
Maximal unique matches (MUMs) are a fundamental primitive in genome comparison, where they serve as high-confidence anchors for downstream multiple genome alignment. However, because MUMs rely on exact string matching, their effectiveness degrades with increased genome divergence and larger sets of genomes, inhibiting their ability to recover long homologous regions and reducing the number of base pairs covered by the multiple genome alignment. Additionally, existing approaches that improve robustness to mutation, such as spaced seeds or translated alignment methods, introduce trade-offs in specificity, scalability, or computational complexity. To address this gap, we introduce the Min-Frame Transformation (MFT), a deterministic encoding of nucleotide sequences to sequences over a transformed alphabet that preserves the coordinate structure of the original sequence. At each position, the MFT selects a k-mer from a local window according to a fixed global ordering and assigns it a character in the transformed alphabet via a predefined mapping. This process captures local sequence context and can mask the impact of mutations, increasing the likelihood that homologous regions remain detectable as exact matches. The resulting transformed sequences can be indexed using standard string data structures, such as suffix arrays and suffix trees, enabling efficient extraction of MUMs without modifying existing algorithms. The MFT is a novel computational approach for improving the robustness of MUM-based seeding for genome alignment by producing longer and more contiguous matches that span a greater fraction of the genome, leading to improved alignment coverage and SNP recall. Altogether, these improvements have the potential to result in improvements for downstream viral genome analysis applications such as phylogenetic inference and transmission analysis.
Urdu is spoken by over 230 million people worldwide, yet it remains significantly underrepresented in digital resources, with limited availability of large-scale, publicly accessible training datasets for optical character recognition (OCR). The diversity of Urdu font styles encountered in printed books, newspapers, and digital publications poses a substantial barrier to developing generalizable OCR systems, while the absence of standardized benchmarks hinders fair and reproducible comparison across recognition approaches. This data article presents FIPU-OCR-CHAR, a benchmark dataset of printed Urdu characters encompassing 48 classes: 38 alphabets and 10 numerals in their isolated forms. The dataset was constructed through a fully systematic pipeline comprising five sequential stages: font collection and validation, character set definition, base image rendering, augmentation, and dataset organization with split generation. Each character class was rendered from 201 distinct Urdu TrueType/OpenType font files, producing 9,648 base images (201 fonts × 48 classes). Each base image was subsequently processed through 34 augmentation operations encompassing geometric transforms, photometric adjustments, blur filters, noise injection, and morphological operations, producing 328,032 augmented images. The complete dataset totals 337,680 labeled PNG images, each stored at 28×28 pixel resolution with 24-bit color depth. The dataset is organized into three predefined splits: training (70%; 236,376 images), validation (20%; 67,536 images), and testing (10%; 33,768 images), each accompanied by a CSV annotation file mapping image filenames to integer class labels (0-47). The repository additionally contains a Jupyter Notebook implementing a ResNet-34 baseline classification pipeline, a results summary image, and a README file documenting dataset structure and label definitions. The dataset is publicly available on Mendeley Data under a CC BY 4.0 license and is intended for use in OCR model development, font-invariant classifier training, Urdu script digitization, transfer learning for word- and line-level recognition, and benchmarking of convolutional neural network and Vision Transformer architectures on low-resource script character recognition tasks.
Physical neural networks (PNNs) are neural-like computational frameworks that exploit the intrinsic dynamics of physical media to achieve ultrafast and energy-efficient information processing. However, the complex and strongly-coupled physical nature of PNNs in disordered environments makes them resistant to accurate differentiable modeling. Here, we propose a concept of computational space that empowers the chaotic environment itself with computational capabilities. This space constitutes a large-scale, model-agnostic PNN through distributed intelligent metasurfaces. To enable effective training, we develop a fully-forward learning framework that estimates zeroth-order gradients from in-situ measurable electromagnetic feedback, thereby circumventing the rigorous modeling requirements of conventional backpropagation. In experiments, we construct such computational space that achieves recognition accuracies of 97% for alphabetic characters and 99% for numeric patterns. Furthermore, the space exhibits the functionalities of enhanced focusing under disordered scattering conditions and reliable human position localization. This emerging paradigm of self-evolving physical intelligence holds potential for advancing embodied intelligence, autonomous cyber-physical systems, and next-generation human-machine interaction, marking a fundamental transition from computing the physics to computing with physics.
Text entry remains a bottleneck for productivity-oriented Virtual Reality (VR), especially in scenarios where optical hand tracking is unstable because of self-occlusion, poor lighting, or out-of-view interaction. We present SlideRing, a dual-thumb wearable text-entry method that senses thumb-to-finger micro-gestures with two miniature Inertial Measurement Units (IMUs). SlideRing defines a 30-command interaction space from two hands, three target fingers, and five gesture types, then maps these commands to a full alphabetic keyboard through two complementary strategies: an ergonomic layout optimized for low movement cost and a QWERTY-compatible layout optimized for learnability. To decode subtle inertial signals, we design a dual-stream recognition model with a Statistical Feature Encoder, a Temporal Feature Encoder, and a context-aware gating module for joint finger-action classification. In offline evaluation, the model reaches 96.5% target-finger accuracy and 94.2% action-type accuracy. In a five-day text-entry study, the ergonomic layout improves from 7.43 to 15.75 words per minute (WPM), while the QWERTY-compatible layout improves from 10.55 to 15.25 WPM. The ergonomic layout reduces physical demand, whereas the QWERTY-compatible layout lowers initial mental load. These results suggest that IMU-based thumb-to-finger input has the potential to provide robust, low-visual-demand text entry for constrained VR environments.
Fixed poetic forms such as the sonnet, ottava rima, or terza rima are an important feature of European literary traditions, yet large-scale empirical research on their cross-lingual distribution and evolution has been limited so far. This paper introduces a fully language-independent, unsupervised method for identifying recurrent rhyme-based forms using local sequence alignment. Drawing on 187,719 poems from six European traditions (Czech, English, French, German, Italian, Russian) in the PoeTree collection, we encode rhyme schemes in a compact eight-symbol alphabet and apply the Smith-Waterman algorithm via the Metronome package to compute pairwise distances. Dimensionality reduction (UMAP) and density-based clustering (HDBSCAN) yield 61 clusters, many of which align with known fixed forms. Evaluation against existing Czech and Russian annotations shows strong recall, while supervised classification experiments-both within and across languages-demonstrate that form categories are robustly learnable in the induced vector space. We illustrate the potential of such data for literary research in three showcases: cross-tradition influence in 19th-century Czech poetry, topical affinities of selected forms using multilingual topic modeling, and geographic associations revealed through geonym analysis.
The Diptera genus-group names of Hermann Loew are reviewed and annotated. A total of 404 available genus-group names in 64 families of Diptera are listed alphabetically for each name giving author, year and page of original publication, originally included species, type species and method of fixation, current status of the name, family placement, and a list of any emendations of it that have been found in the literature. Remarks are given to clarify nomenclatural or taxonomic information. Additionally, a full list of all species-group names proposed by Loew (3,762, of which 3,689 are available names) is presented with original date of publication and page number. An appendix gives a full bibliography of all scientific works published by Loew. Type species are designated for: Apedilia Loew, 1850e [Ceratopogonidae]; Eupedilia Loew, 1850e [Ceratopogonidae]; Tanyrrhina Bezzi, 1912 [Blephariceridae]. Corrected or clarified type-species, clarified type-species fixations and/or earlier dates of availability for nominal genera are given for the following genus-group names: Achalcus Loew, 1857c [Dolichopodidae]; Anarmostus Loew, 1860d [Asilidae]; Anorostoma Loew, 1862g [Heleomyzidae]; Antipalus Loew, 1849a [Asilidae]; Antiphrisson Loew, 1849a [Asilidae]; Apospasmica Loew, 1868e [Ulidiidae]; Ataracta Loew, 1850e [Limoniidae]; Bathypogon Loew, 1851d [Asilidae]; Centor Loew, 1864m [Chloropidae]; Coniceps Loew, 1873b [Richardiidae]; Echthistus Loew, 1849a [Asilidae]; Epiplatea Loew, 1868a [Richardiidae]; Epitriptus Loew, 1849a [Asilidae]; Eriopogon Loew, 1847f [Asilidae]; Eutolmus Loew, 1848b [Asilidae]; Euxesta Loew, 1868a [Ulidiidae]; Habropogon Loew, 1847f [Asilidae]; Hammatorrhina Loew, 1869i [Blephariceridae]; Holopogon Loew, 1847a [Asilidae]; Isopogon Loew, 1847f [Asilidae]; Lamprozona Loew, 1851d [Asilidae]; Laparus Loew, 1851d [Asilidae]; Lasiopogon Loew, 1847f [Asilidae]; Merosargus Loew, 1855c [Stratiomyidae]; Mochlonyx Loew, 1844b [Chaoboridae]; Mochtherus Loew, 1849a [Asilidae]; Nygmatodes Loew, 1845a [Psychodidae]; Proagonistes Loew, 1858c [Asilidae]; Pycnopogon Loew, 1847f [Asilidae]; Scoliocentra Loew, 1862e [Heleomyzidae]; Stictocephala Loew, 1873e [Ulidiidae]; Symmictus Loew, 1858c [Nemestrinidae]; Synarthrus Loew, 1857d [Dolichopodidae]; Tachista Loew, 1864r [Hybotidae]; Tephrochlamys Loew, 1862e [Heleomyzidae]; Teratopus Loew, 1858c [Asilidae]; Triptotricha Loew, 1872a [Xylophagidae]; Xanthochlorus Loew, 1857c [Dolichopodidae]. The following genus-group names previously treated as available have been found through research conducted in this work to be unavailable: Chaetostoma Loew, 1873e [Tephritidae]; Cryomobia Frey, 1913 [Heleomyzidae]; Spathidogaster Loew, 1876b [Syrphidae]; Tanyrhina Loew, 1862b [Blephariceridae]. Acting as FirstReviser, the following correct original spellings for multiple original spellings are selected by us: Melanoloma Loew, 1873e [Richardiidae]; Platomodes Loew, 1855b [Bombyliidae]; Sphageus Loew, 1866e [Asilidae]. Also, acting as First Reviser, we select Antiphrisson sareptanus Lichtwardt, 1903 to have precedence over Antiphrisson thalhammeri Lichtwardt, 1903 as the valid name for the type species of Antiphrisson Loew, 1849 [Asilidae] and Synolcustenuiventris Loew, 1858 to have precedence over Synolcus signatus Loew, 1858 as the valid name for the type species of Synolcus Loew, 1858. The following correct original spellings for multiple original spellings selected by First Reviser actions were missed by previous workers: Acidogona Loew, 1873b [Tephritidae]; Electra Loew, 1850e [Xylophagidae]; Empyelocera Loew, 1866h [Ulidiidae]; Tachista Loew, 1864g [Hybotidae]; Xanthochlorus Loew, 1857c [Dolichopodidae]. The following nominal genera are new junior synonyms of their respective senior synonyms: Acanthoneura Loew, 1862b of Acanthonevra Macquart, 1844, n. syn. [Tephritidae]; Apedilia Loew, 1850e of Serromyia Meigen, 1818, n. syn. [Ceratopogonidae]; Chrysochlamys Loew, 1858h of Ferdinandea Rondani, 1844, n. syn. [Syrphidae]; Clinorrhyncha Bertkau, 1893 of Ozirhincus Rondani, 1840, n. syn. [Cecidomyiidae]; Criorrhina Loew, 1871a of Criorhina Meigen, 1822, n. syn. [Syrphidae]; Dirrhiza Bertkau, 1893 of Dirhiza Loew, 1850c, n. syn. [Cecidomyiidae]; Eupedilia Loew, 1850e of Forcipomyia Meigen, 1818, n. syn. [Ceratopogonidae]; Exeretoneura Loew, 1860d of Exeretonevra Macquart, 1846, n. syn. [Xylophagidae]; Heligmoneura Loew, 1860d of Heligmonevra Bigot, 1858b, n. syn. [Asilidae]; Homalocephala Loew, 1873e of Setellia Robineau-Desvoidy, 1830, n. syn. [Richardiidae]; Leptomidas Loew, 1872a of Leptomydas Gerstaecker, 1868, n. syn. [Mydidae]; Loxoneura Loew, 1873e of Loxonevra Macquart, 1835, n. syn. [Platystomatidae]; Megarhina Loew, 1862b of Lynchiella Lahille, 1904, n. syn. [Culicidae]; Microdromia Loew, 1864h of Microdromya Bigot, 1857, n. syn. [Empididae]; Myia Loew, 1858c of Calliphora Robineau-Desvoidy, 1830, n. syn. [Calliphoridae]; Naeeta Loew, 1862b of Noeeta Robineau-Desvoidy, 1830, n. syn. [Tephritidae]; Nygmatodes Loew, 1845a of Nemopalpus Macquart, 1838a, n. syn. [Psychodidae]; Oncodocera Loew, 1862b of Ogcodocera Macquart, 1840, n. syn. [Bombyliidae]; Pachyrrhina Loew, 1871a of Nephrotoma Meigen, 1803. n. syn. [Tipulidae]; Phlebotomus Loew, 1845a, of Phlebotomus Rondani & Berté, 1840, n. syn. [Psychodidae]; Polistoides Loew, 1873e of Polystodes Robineau-Desvoidy, 1830, n. syn. [Platystomatidae]; Polydromia Loew, 1864h of Chelifera Macquart, 1823, n. syn. [Empididae]; Pterotaenia Rondani, 1868a of Apospasmica Loew, 1868e, n. syn. [Ulidiidae]; Ramphomyia Loew, 1840a of Rhamphomyia Meigen, 1822, n. syn. [Empididae]; Rhymosia Loew, 1872a of Rymosia Winnertz, 1864, n. syn. [Mycetophilidae]; Ryphus Loew, 1844a of Silvicola Harris, 1780, n. syn. [Anisopodidae]; Sitaria Loew, 1862b of Orellia Robineau-Desvoidy, 1830, n. syn. [Tephritidae]; Sympicnus Bigot, 1890 of Sympycnus Loew, 1857c, n. syn. [Dolichopodidae]; Tanyrrhina Bezzi, 1913 of Hammatorrhina Loew, 1869i, n. syn. [Blephariceridae]; Tephrochlamys Loew, 1862e of Heteromyza Fallén, 1820a, n. syn. [Heleomyzidae]; Toxoneura Loew, 1873e of Palloptera Fallén, 1810, n. syn. [Pallopteridae]; Xiphocerus Loew, 1847f of Xiphocera Macquart, 1834a, n. syn. [Asilidae]. Reversal of precedence is invoked for two cases of objective synonymy to promote stability in nomenclature: Aphrosylus Haliday in Walker, 1851b, nomenprotectum and Aphrozylus Loew, 1850b, nomenoblitum [in Dolichopodidae]; Lynchiella Lahille, 1904, nomen protectum and Megarhina Loew, 1862b, nomen oblitum [in Culicidae].