Search Results (22,256)

This paper introduces AdamN, a nested-momentum adaptive optimizer that replaces the single Exponential Moving Average (EMA) numerator in Adam/AdamW with a compounded EMA of gradients plus an EMA of that EMA, paired with an exact double-EMA bias correction. This yields a smoother, curvature-aware search direction at essentially first-order cost, with longer, more faithful gradient-history memory and a stable, warmup-free start. Under comparable wall-clock time per epoch, AdamN matches AdamW’s final accuracy on ResNet-18/CIFAR-100, while reaching 80% and 90% training-accuracy milestones ~127 s and ~165 s earlier, respectively. On pre-benchmarking workloads (toy problems and CIFAR-10), AdamN shows the same pattern: faster early-phase convergence with similar or slightly better final accuracy. On language modeling with token-frequency imbalance—Wikitext-2-style data with training-only token corruption and a 10% low-resource variant—AdamN lowers rare-token perplexity versus AdamW without warmup while matching head and mid-frequency performance. In full fine-tuning of Llama 3.1–8B on a small dataset, AdamN reaches AdamW’s final perplexity in roughly half the steps (≈ 2.25 xfaster time-to-quality). Finally, on a ViT-Base/16 transferred to CIFAR-
100 (batch size 256), AdamN achieves 88.8% test accuracy vs. 84.2% for AdamW and reaches
40–80% validation-accuracy milestones in the first epoch (AdamW reaches 80% by epoch 59),
reducing epochs, energy use, and cost. Full article

This study systematically maps how artificial intelligence (AI) has been applied within finite-element (FE)-based structural engineering. A corpus of 5995 unique English-language publications was compiled and classified by discipline, with 3345 relevant papers further categorized by application group. A representative subset of 372 studies underwent detailed full-text classification across seven analytical dimensions covering AI methods, element formulations, materials, and structural objects. The analysis reveals rapid growth after 2015, including a pronounced expansion of surrogate modeling and data-driven prediction methods. The disciplinary composition of the literature has also evolved, with structural engineering studies becoming more prominent in recent years relative to earlier decades. Optimization & Design remains the largest application area across the full dataset, while Structural Performance Prediction and FEM Acceleration/Surrogate Modeling show the fastest growth, reflecting increasing emphasis on predictive, solver-efficient, and hybrid physics–data approaches. These findings indicate a maturing field in which AI is increasingly embedded across all stages of FE-based analysis and design. This study provides a structured overview of methodological patterns, identifies emerging hybrid strategies, and highlights opportunities for future research and industrial integration. Full article

Vehicle repositioning is essential for improving efficiency and service quality in ride-sharing platforms, yet existing approaches typically optimize proxy rewards that fail to reflect human-centered preferences such as wait time, service coverage, and unnecessary empty travel. We propose the first two-stage Bilevel Reinforcement Learning (RL) from Human Feedback (RLHF) framework for preference-aligned vehicle repositioning. In Stage 1, a value-based Deep Q-Network (DQN)-RLHF warm start learns an initial preference-aligned reward model and stable reference policy, mitigating the reward-model drift and cold-start instability that arise when applying on-policy RLHF directly. In Stage 2, a Kullback–Leibler (KL)-regularized Proximal Policy Optimization (PPO)-RLHF algorithm, equipped with action masking, behavioral-cloning anchoring, and alternating forward–reverse KL, fine-tunes the repositioning policy using either Large Language Model (LLM)-generated or rubric-based preference labels. We develop and compare two coordination schemes, pure alternating (PPO-Alternating) and k-step alternating (PPO-k-step), demonstrating that both yield consistent improvements across all tested arrival scales. Empirically, our framework reduces wait time and empty-mile ratio while improving served rate, without inducing trade-offs or reducing platform profit. These results show that human preference alignment can be stably and effectively incorporated into large-scale ride-sharing repositioning. Full article

This study aimed to identify young school students’ language learning strategies (LLSs) within their diverse socio-linguistic reality. The study was conducted in one elementary school in a peripheral city characterized by a heterogeneous population (Arabs and Jews) and immigrants from various countries who speak multiple languages. The principal of this school opted to introduce young children (Grades 1 and 2) to four languages: Hebrew, as a socially dominant language; Russian and Arabic, as the children’s home languages; and English, as a global language. We used photo elicitation and dialogical conversation to obtain reflections of 11 Arab and Jewish students (Grade 2). Each student was asked to describe the strategies they used to learn a novel language in the classroom and at home. Findings support the appropriateness of Oxford’s taxonomy to young language learners: all LLSs’ categories were reported. This study contributes to our understanding of children’s ability to use LLSs in early primary school. It highlights the leading role of language teachers who seem to mediate by modelling LLSs. Furthermore, it enriches the understanding of how 7–8-year-old learners can use diverse metacognitive LLSs and transfer them across languages. We also found one “child-specific” characteristics of the strategy related to parental involvement. Full article

Background: Rapid and accurate fracture detection is critical in emergency departments (EDs), where high patient volume and time pressure increase the risk of diagnostic error, particularly in radiographic interpretation. Multimodal large language models (LLMs) with image-recognition capability have recently emerged as general-purpose tools for clinical decision support, but their diagnostic performance within routine emergency department imaging workflows in orthopedic trauma remains unclear. Methods: In this retrospective diagnostic accuracy study, we included 1136 consecutive patients referred from the ED to orthopedics between 1 January and 1 June 2025 at a single tertiary center. Given the single-center, retrospective design, the findings should be interpreted as hypothesis-generating and may not be fully generalizable to other institutions. Emergency radiographs and clinical data were processed by a multimodal LLM (2025 version) via an official API using a standardized, deterministic prompt. The model’s outputs (“Fracture present”, “No fracture”, or “Uncertain”) were compared with final diagnoses established by blinded orthopedic specialists, which served as the reference standard. Diagnostic agreement was analyzed using Cohen’s kappa (κ), sensitivity, specificity, accuracy, and 95% confidence intervals (CIs). False-negative (FN) cases were defined as instances where the LLM reported “no acute fracture” but the specialist identified a fracture. The evaluated system is a general-purpose multimodal LLM and was not trained specifically on orthopedic radiographs. Results: Overall, the LLM showed good diagnostic agreement with orthopedic specialists, with concordant results in 808 of 1136 patients (71.1%; κ = 0.634; 95% CI: 68.4–73.7). The model achieved balanced performance with sensitivity of 76.9% and specificity of 66.8%. The highest agreement was observed in knee trauma (91.7%), followed by wrist (78.8%) and hand (69.6%). False-negative cases accounted for 184 patients (16.2% of the total cohort), representing 32.4% of all LLM-negative assessments. Most FN fractures were non-displaced (82.6%), and 17.4% of FN cases required surgical treatment. Ankle and foot regions showed the highest FN rates (30.4% and 17.4%, respectively), reflecting the anatomical and radiographic complexity of these areas. Positive predictive value (PPV) and negative predictive value (NPV) were 69.4% and 74.5%, respectively, with likelihood ratios indicating moderate shifts in post-test probability. Conclusions: In an emergency department-to-orthopedics consultation cohort reflecting routine clinical workflow, a multimodal LLM demonstrated moderate-to-good diagnostic agreement with orthopedic specialists, broadly within the range reported in prior fracture-detection AI studies; however, these comparisons are indirect because model architectures, training strategies, datasets, and endpoints differ across studies. However, its limited ability to detect non-displaced fractures—especially in anatomically complex regions like the ankle and foot—carries direct patient safety implications and confirms that specialist review remains indispensable. At present, such models may be explored as hypothesis-generating triage or decision-support tools, with mandatory specialist confirmation, rather than as standalone diagnostic systems. Prospective, multi-center studies using high-resolution imaging and anatomically optimized algorithms are needed before routine clinical adoption in emergency care. Full article

Large language models (LLMs) are increasingly used to predict human behavior from plain-text descriptions of experimental tasks that range from judging disease severity to consequential medical decisions. While these methods promise quick insights without complex psychological theories, we reveal a critical flaw: they often latch onto accidental patterns in the data that seem predictive but collapse when faced with novel experimental conditions. Testing across multiple behavioral studies, we show these models can generate wildly inaccurate predictions, sometimes even reversing true relationships, when applied beyond their training context. Standard validation techniques miss this flaw, creating false confidence in their reliability. We introduce a simple diagnostic tool to spot these failures and urge researchers to prioritize theoretical grounding over statistical convenience. Without this, LLM-driven behavioral predictions risk being scientifically meaningless, despite impressive initial results. Full article

Objectives: The industrial production of lanolin alcohol currently employs batch saponification, which suffers from high energy consumption, prolonged processing time, and excessive solid waste generation, rendering it incompatible with green chemistry principles. This study aimed to develop an efficient, sustainable saponification process by addressing these limitations through integrating large language models (LLMs) with microfluidic technology. Methods: An LLM-based intelligent agent called SapoMind (version 1.0) was constructed. SapoMind employs LLMs as its software core and a continuous-flow microreactor as the experimental platform. Its performance was enhanced through supervised fine-tuning. The system enables automated recommendation of saponification process parameters, autonomous experimental design, and automatic execution of experiments. Saponification conditions were automatically optimized considering product quality, energy consumption, material consumption, and time consumption. Results: The optimal continuous-flow saponification conditions were determined as 70 °C reaction temperature and 9 min residence time, producing lanolin alcohol complying with European Pharmacopoeia standards. Compared to batch processing, the optimized process reduced carbon emissions by 53% and solid waste by 37%, with the greenness score increasing from 82 to 93. Conclusions: This study demonstrates the effectiveness of LLM-driven intelligent agents in optimizing green chemical processes. SapoMind offers significant environmental and operational benefits for lanolin alcohol production. Full article

Background/Objectives: Prior studies suggest that rates of Helicobacter pylori colonization are higher among Hispanic immigrants compared to U.S.-born Hispanics. It is unknown whether differences in H. pylori colonization rates among Hispanics are related to immigration status or to cultural factors such as diet. Methods: This was a survey study, conducted among self-identified Hispanics who had an endoscopy for symptoms of gastroesophageal reflux disease (GERD). Qualifying patients completed a telephone survey which included questions about immigration status and the 12-item Short Acculturation Scale for Hispanics (SASH), a validated instrument which measures cultural factors such as language preference and diet. We examined the relationship between SASH factors and H. pylori status, classified based on endoscopic biopsy results. Results: We called 400 patients and 186 completed the survey. Median age was 65 (interquartile range 21 to 82) and 81% were female. Thirty of 186 (16%) respondents were born in the U.S. while 156/186 (84%) were immigrants, primarily from the Dominican Republic. Among immigrants, 69% had immigrated before 1990. Rates of H. pylori were 8/30 (27%) among U.S. born Hispanics compared to 51/156 (33%) among Hispanic immigrants (p = 0.67). Rates of H. pylori were 51/147 (35%) among those with a mostly Latino diet vs. 8/39 (21%) among those with a U.S or mixed diet (p = 0.05). In a multivariable model predicting H. pylori status, a mostly Latino diet was the only cultural predictor which approached statistical significance (p = 0.05) (aOR 2.61, 95% CI 0.94–7.20). Conclusions: Rates of H. pylori colonization were modestly higher among Hispanic immigrants compared to U.S.-born Hispanics. A novel preliminary finding was that higher rates of H. pylori colonization were observed among those who ate a predominantly Latino diet. Full article

This article focuses on the French translation of Fo Guo Ji (佛国记), entitled Foě-kouě-ki, ou Relation des royaumes bouddhiques, co-translated by four scholars, Jean-Pierre Abel-Rémusat, Julius Klaproth, Ernest Clerc de Landresse, and Eugène Burnouf. This study examines how the translators interpreted Buddhist doctrines and history by analyzing the translation’s paratexts, including an extensive introduction, detailed in-text annotations as well as the translators’ related studies of Buddhism and other fields. As renowned Sinologists whose expertise spanned religion, linguistics, geography, and other fields, the translators infused their interpretive work with multifaceted academic perspectives. These paratexts collectively serve as pivotal sources for understanding how the French translation constructed and interpreted Buddhist history. The study finds that Abel-Rémusat profoundly elaborated on the core of Buddhist doctrines, dispelling the prevalent mystification of Buddhism in European academia and uncovering its inherent rational logic. All four translators, endowed with profound philological and linguistic expertise, analyzed Buddhist history through a distinctive approach. Moreover, Abel-Rémusat and his successors focused more on the history of Buddhist transmission across regions and languages, positioning Buddhist history as well as Asian studies into the framework of world history. Full article

This study investigates the influence of toxic language in hate speech targeting immigrants, particularly through narrative formats like first-person X (Twitter) threads. Hate speech, defined as promotion of hatred based on personal or group characteristics, increasingly escalates on social media, impacting public attitudes and behaviors. While previous research has primarily focused on measuring the scope of hate speech through content analysis and computational methods, there has been limited attention to its effects on audiences. This study presents the results of an online experiment (N = 339) with a 2 × 2 between-subjects design that manipulates the presence of toxic language and message popularity. Results indicate that hate messages lacking toxic language promote greater identity fusion with the author of the message, which in turn increases the intention to share the message, reinforces negative attitudes toward immigrants, and increases support for harsh policies against irregular immigration. Moreover, non-toxic hate messages significantly enhance narrative transportation exclusively for individuals with conservative political views, thereby further increasing their intention to share the message. These findings highlight that subtler forms of hate speech can create strong audience connections with hostile perspectives, emphasizing the need for anti-hate campaigns to address both overt and subtle hate content. Full article

When a physical theory is in its early stages of development, it presents many concepts and constructs that may not be necessary for its interpretation, or may simply be equivocal. As the theory is developed by the physics community, it hopefully passes through a depuration process that washes away many of these constructs and introduces others not initially devised. This process can be viewed, and modeled, as a semiotic process by which the suggested interpretations of the theory, initially dispersed in the space of concepts, taper off in a way that leaves only a small number of possibilities, ideally only one. However, to qualify this process and impose semiotic and epistemological constraints on the depuration process, it seems natural to consider a physical theory as an excerpt of a language and its suggested interpretations as texts, endowed with syntactics and semantics. In this paper we present this framing of general physical theories and apply the resulting semiotic and epistemological constraints we uphold to the special case of quantum mechanics, which shows particular resistance to interpretation tapering. We then show that the findings of this paper are especially important for allowing one to form a hierarchy of interpretations of the same formal structure of a physical theory, even in the case of an experimental underdetermination of these interpretations, which is precisely the case for quantum mechanics. This result is particularly important for more modern physical theories, which are becoming increasingly more abstract and difficult to interpret. Full article

The Thailand 4.0 agenda elevates entrepreneurship education (EE) as a lever to escape the middle-income, inequality, and imbalance traps, yet EE remains weakly embedded in basic education—especially in Thai language. We designed and piloted a community-economy context-based learning model integrating EE (CEC-EE) for Grade 12 Thai via a two-cycle R&D process: needs analysis (surveys and focus groups with teachers and students) and prototype development. The model operationalizes six instructional steps (6Cs: connect, comprehend, clarify, construct, carry over, and conclude) anchored in Mae Chan’s community economy and targets entrepreneurial language skills (ELSs) consisting of analytical reading and creative writing. In a one-group pretest–posttest with Grade 12 students (n = 32), academic achievement and ELSs—analytical reading and creative writing—improved markedly. Posttest means exceeded pretests with very large effect. Experts rated the model appropriate, feasible, and useful; teachers and students reported high perceived value alongside concerns about implementation cost, support capacity, and student readiness. The CEC-EE model offers a context-responsive pathway for embedding EE in Thai-language instruction; future work should employ comparative designs, multi-site samples, and cost-effectiveness analyses to assess scalability and sustained impact. Full article

To establish the best approach for conserving a species, it is necessary to understand the biology of that species. To better understand the behavior of American black bears (Ursus americanus), we observed 246 black bears for 7950 h in nature over a 24-year period to quantify how the bears communicated. Black bears communicated using several different behaviors. These included thirteen types of vocalizations, eight olfactory behaviors, eight marking behaviors, sixteen different body postures and gestures constituting their body language, and various emotional expressions. Some behaviors appeared to be automatic, including facial expression, ear movements, some forms of body language, the intensity of various vocalizations, and various moans. Other behaviors appeared to be intentional, including mechanically generated sounds and actions that could be used to bluff or deceive, such as the chomping of teeth, huffing, swatting, false charging, and various vocalizations. The conservation of black bears can be improved by establishing management strategies that take into account the vocal and non-vocal communication of the bears. Conflicts and negative encounters between humans and bears can be reduced through behavioral modifications by humans based on our new understanding of the communication system of bears. Knowledge of the communication system of the black bear provides a basis for improved conservation through the non-lethal management of bears involved in bear–human conflicts. Full article

Autologous serum (AS) tears are an effective therapeutic option for advanced DED, mimicking the biochemical composition of natural tears. However, the absence of universally accepted guidelines has resulted in variability in AS tear concentration, diluents, processing of collected blood, and storage conditions, raising questions regarding the optimal parameters for AS tear use. This perspective provides a framework to inform clinical implementation and to guide future research on AS tear therapy optimization. PubMed, Scopus, and the Cochrane Library were searched for English-language articles from January 2022 through September 2025 using the terms “autologous serum,” “dry eye disease,” “dry eye syndrome,” “dry eye,” and “DED.” Evidence suggests that AS tears diluted to 20% are widely used for moderate DED, whereas higher concentrations may provide faster, more pronounced and more durable improvements, particularly in severe cases. Levofloxacin-containing eye drops, artificial tears without emphasis on a specific component, sodium hyaluronate (SH)-containing eye drops, cyclosporine A (CsA)-containing ultra-nano emulsions, and methylcellulose have been investigated as alternatives to conventional diluents. Standardization of clotting, centrifugation and storage parameters is expected to enhance efficacy of AS tears and ensure stability of growth factors. Combination with estrogen replacement therapy in perimenopausal women or with topical insulin eye drops, as well as perioperative prophylactic use in patients with graft-versus-host disease (GVHD)-associated dry eye undergoing cataract surgery, represent emerging applications of AS tears that demonstrate potential to improve therapeutic outcomes. Overall, this perspective highlights the need for consensus protocols, supports severity-based concentration tailoring, and notes that diluents and processing methods require further refinement. Full article

Laser Doppler vibrometer (LDV) has the characteristics of long-distance, non-contact, and high sensitivity, and plays an increasingly important role in industrial, military, and security fields. Remote speech acquisition technology based on LDV has progressed significantly in recent years. However, unlike microphone receivers, LDV-captured signals have severe signal distortion, which affects the quality of the LDV-captured speech. This paper proposes a nested U-network with gated temporal convolution (TCNUNet) to enhance monaural speech based on LDV. Specifically, the network is based on an encoder-decoder structure with skip connections and introduces nested U-Net (NUNet) in the encoder to better reconstruct speech signals. In addition, a temporal convolutional network with a gating mechanism is inserted between the encoder and decoder. The gating mechanism helps to control the information flow, while temporal convolution helps to model the long-range temporal dependencies. In a real-world environment, we designed an LDV monitoring system to collect and enhance voice signals remotely. Different datasets were collected from various target objects to fully validate the performance of the proposed network. Compared with baseline models, the proposed model achieves state-of-the-art performance. Finally, the results of the generalization experiment also indicate that the proposed model has a certain degree of generalization ability for different languages. Full article