Global ETD Search

Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.

Results

Showing 1 to 20 of 57 for “"Logits"”.

  1. A comparison of item exposure control procedures with the generalized partial credit model

    … for efficacy were the modified within .10 logits, restricted modified within .10 logits, randomesque, restricted randomesque, and progressive restricted procedures. The modified within .10 logits, restricted modified within .10 logits, and randomesque, and restricted randomesque procedures …

    texas Repository record for A comparison of item exposure control procedures with the generalized partial credit model (opens in a new tab)

  2. Development of Reading Fluency During an Extensive Reading Course Incorporating Reading Fluency Tasks

    … texts was significantly greater for IR (1.00 logits) than ERFT (.17 logits) (p = .010); and vocabulary size increase was significantly greater for IR (.36 logits) than ERFT (-.19 logits) (p = .001). For the ERFT-before-IR group, reading comprehension increase for authentic texts was …

    temple Repository record for Development of Reading Fluency During an Extensive Reading Course Incorporating Reading Fluency Tasks (opens in a new tab)

  3. Analysis of algorithms to reduce hallucinations in vision language models /

    … performance peaks at an ultra-narrow window of 2 logits (AUROC of 97.67%), improves substantially as the sequence is expanded to 20 logits. Simultaneous cross-dataset and cross-prompt transfer causes complete classifier collapse. High decoding temperatures introduce arithmetic overflow that is …

    vilnius Repository record for Analysis of algorithms to reduce hallucinations in vision language models / (opens in a new tab)

  4. Strategies for controlling item exposure in computerized adaptive testing with polytomously scored items

    … baseline condition, the randomesque, within .10 logits, Sympson-Hetter, conditional Sympson-Hetter, aStratified, and enhanced a-Stratified procedures were implemented to control item exposure rates. The a-Stratified and enhanced a-Stratified procedures were not evaluated with the partial credit …

    texas Repository record for Strategies for controlling item exposure in computerized adaptive testing with polytomously scored items (opens in a new tab)

  5. A randomised controlled trial and systematic review comparing two methods of constraint induced movement therapy to improve upper limb function in pre-school children with hemiplegic cerebral palsy

    … similar study and a treatment effect of 0.92 AHA logits was demonstrated. This is compatible with the smallest detectable difference (0.97 logits) indicating actual change in bimanual performance. The short-term efficacy, excellent recruitment and retention rates and acceptability of the trial …

    birmingham Download PDF for A randomised controlled trial and systematic review comparing two methods of constraint induced movement therapy to improve upper limb function in pre-school children with hemiplegic cerebral palsy (opens in a new tab)

  6. Towards optimal trade-offs between accuracy and robustness in adversarial training

    … which minimizes the discrepancies between the logits of clean and adversarial examples. This dual approach effectively strengthens the model's robustness against adversarial perturbations by fostering stable decision boundaries that are less prone to manipulation. In contrast, PMHR-AT is …

    iastate Repository record for Towards optimal trade-offs between accuracy and robustness in adversarial training (opens in a new tab)

  7. In the Shadow of Prompts: Adversarial Attacks and Model Cloning in Large Language Models

    … (APIs) that still expose full or top-k logits and lack mature safeguards, they present a serious, often overlooked attack surface. Earlier work has shown how to rebuild the output projection layer or distill surface behavior, but no attack has produced a deployable clone within a tight …

    embry-riddle Repository record for In the Shadow of Prompts: Adversarial Attacks and Model Cloning in Large Language Models (opens in a new tab)

  8. A multilevel analysis of elders’ cognitive performance using data from the longitudinal study of generations

    … their cube drawing performance was .034 logits lower than those who did not have any functional impairment at age 70. However, having at least some functional impairment was associated with a .004 logits increment to the growth rate of cube drawing performance. Possible reasons for both …

    ttu Repository record for A multilevel analysis of elders’ cognitive performance using data from the longitudinal study of generations (opens in a new tab)

  9. Utilizing the assessment of motor and process skills ability measures to predict level of community dependence

    … demarcating independence were 1.50 and 1.10 logits respectively, and proposed ADL motor and ADL process cutoff measures demarcating the need for moderate to maximal assistance were 1.00 and 0.70 logits respectively. The ability measures between the proposed cutoff measures demarcate a risk …

    colostate Repository record for Utilizing the assessment of motor and process skills ability measures to predict level of community dependence (opens in a new tab)

  10. Exploring Teachers' L2 Listening Assessment Literacy

    … knowledge test and the questionnaire to provide logits values for items and participants. The Wright maps from the three Rasch analyses provide insightful information about teachers' current L2 listening assessment literacy. The goal here is to provide the overall pattern of the teachers L2 …

    iastate Repository record for Exploring Teachers' L2 Listening Assessment Literacy (opens in a new tab)

  11. Odhad věku ze snímků sítnice

    … ResNet-34 a algoritmu Consistent Rank Logits predikující věk jako ordinální proměnnou. Nejlepší model dosáhl průměrné absolutní chyby 3.42 roku, čímž překonal stávající modely odhadující věk ze snímků sítnice. Veškeré algoritmy jsou implementovány v programovacím jazyce Python s …

    brno-tech Repository record for Odhad věku ze snímků sítnice (opens in a new tab)

  12. Strategies for controlling testlet exposure rates in computerized adaptive testing systems

    … randomesque procedure, the modified within .10 logits procedure, two levels of the progressive restricted procedure, and two levels of the Sympson-Hetter procedure. Each of these was compared to a baseline no exposure control procedure, maximum information. The testlets were reading passages …

    texas Repository record for Strategies for controlling testlet exposure rates in computerized adaptive testing systems (opens in a new tab) Restricted access

  13. ADVANCING TEST-TIME COMPUTE SCALING IN LANGUAGE MODELS: FROM TRAINING TO INFERENCE

    … parallel reasoning trajectories whose next-token logits are aggregated during shared decoding. Together, these methods form a unified framework for making additional test-time computation more useful and reliable without increasing the size of the base model.

    nus Repository record for ADVANCING TEST-TIME COMPUTE SCALING IN LANGUAGE MODELS: FROM TRAINING TO INFERENCE (opens in a new tab) Download file for ADVANCING TEST-TIME COMPUTE SCALING IN LANGUAGE MODELS: FROM TRAINING TO INFERENCE (opens in a new tab)

  14. Dual knowledge distillation with adaptive class-wise prototype margins for heterogeneous federated learning

    … knowledge aggregation by combining client logits with prototype feature representations, while mitigating the margin-shrinking problem through a contrastive- learning-based trainable server prototype equipped with a class-wise adaptive prototype margin. Additionally, the method evaluates …

    iastate Repository record for Dual knowledge distillation with adaptive class-wise prototype margins for heterogeneous federated learning (opens in a new tab)

  15. Closing the Gap Between Classification and Retrieval Models

    … convolutional feature. The network output can be logits trained with cross-entropy or a space embedding trained with a ranking loss. Furthermore, this approach neither changes the original network weights nor requires fine-tuning. Thus, network performance remains intact. The visualization filter …

    maryland Repository record for Closing the Gap Between Classification and Retrieval Models (opens in a new tab)

  16. A New Multilevel Bayesian Nonparametric Algorithm and Its Application in Causal Inference

    … compared to S-BART methods and PSM using logits regression models and show great capacities in dealing with nonlinearity, cluster effects and treatment effect heterogeneity.

    tamu Repository record for A New Multilevel Bayesian Nonparametric Algorithm and Its Application in Causal Inference (opens in a new tab)

  17. Automatic Handwriting Analysis for Classifying Multi-Label Personality Traits using Transformer OCR

    … class imbalance and Binary Cross-Entropy with Logits for accurate classification. The dataset includes 873 French and 181 English handwriting samples from CENPARMI, originally labeled for Extraversion and Conscientiousness. It has been expanded to cover all five BFFM traits: Extraversion, …

    concordia Download PDF for Automatic Handwriting Analysis for Classifying Multi-Label Personality Traits using Transformer OCR (opens in a new tab)

  18. Dimensionality analysis of students’ performance in 2013 BGCSE agricultural examination: implications for differential item functioning

    … done according to gender and location type using logits test for t-test significance (p < .05). The findings revealed that examination was not unidimensional. None of the items fit the 1PL. Only one fit the 2PL and 8 items fitted the 3PL. The findings from this study on gender based DIF indicated …

    botswana Repository record for Dimensionality analysis of students’ performance in 2013 BGCSE agricultural examination: implications for differential item functioning (opens in a new tab)

  19. Research on eeg classification algorithms by sleep stage /

    … Hidden Markov Model (HMM) to the per-epoch logits as temporal post-processing, and enable the same HMM smoothing in the model serving side to maintain parity with evaluation, for the sequence-aware CNN+BiLSTM, which already models temporal context, we serve raw predictions without HMM. …

    vilnius Repository record for Research on eeg classification algorithms by sleep stage / (opens in a new tab)

  20. An investigation of machine learning systems security with a forensics perspective

    … noise is injected into the surrogate model's logits before adversarial example generation, with the aim of disrupting the attribution signals relied upon by the tracing mechanism. Experiments are conducted across multiple architectures, including ResNet18, ResNet50, and VGG16, and evaluated on …

    qu-belfast Repository record for An investigation of machine learning systems security with a forensics perspective (opens in a new tab) Download PDF for An investigation of machine learning systems security with a forensics perspective (opens in a new tab)

Page 1 of 3