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 125 for “"Perplexity"”.

  1. An Underlying Perplexity

    An underlying perplexity is a celebration of human beings’ inability to comprehend reality, a celebratory gesture for our confusion. Whether in this text or the two parts of my exhibition, I have humbly tried to position myself where I need to be. If I were to seek an affect within the pieces of my …

    sask Repository record for An Underlying Perplexity (opens in a new tab)

  2. "The perplexity of desire: Representation and poetic thinking in Shakespeare's ""Venus and Adonis"" and ""Love's Labor's Lost"""

    While examining and challenging the current thought that desire only exists in the alienated movement between anticipation and insufficiency, this study arranges itself around three or four sentences used in discussions about desire: the proposition, a statement judged either true or false; the …

    uiuc Repository record for "The perplexity of desire: Representation and poetic thinking in Shakespeare's ""Venus and Adonis"" and ""Love's Labor's Lost""" (opens in a new tab)

  3. A Matter of Increasing Perplexity: Public Perception, Treatment, and Military Influence of Refugees in the Shenandoah Valley During the American Civil War

    This thesis examines the ways in which definitions and perceptions of refugees in the Shenandoah Valley of Virginia evolved over the course of the American Civil War. It investigates perspectives from individuals in both the United States and Confederate States to illustrate how misconceptions …

    vt Repository record for A Matter of Increasing Perplexity: Public Perception, Treatment, and Military Influence of Refugees in the Shenandoah Valley During the American Civil War (opens in a new tab)

  4. An incremental syntactic language model for statistical phrase-based translation.

    … that demonstrate substantial improvements in perplexity for our syntactic language model over traditional n-gram language models; we also present empirical results on a constrained Urdu-English translation task that demonstrate the use of our syntactic LM.A standard measure of language model …

    umn Repository record for An incremental syntactic language model for statistical phrase-based translation. (opens in a new tab)

  5. Conditional Replicated Softmax for Topic Modelling with Metadata

    … of the new model was associated with higher test perplexity, which scores their ability to predict unseen documents from a test set, and higher penalized perplexity which penalizes perplexity for model complexity.

    guelph Repository record for Conditional Replicated Softmax for Topic Modelling with Metadata (opens in a new tab)

  6. The Suitability of Generative Adversarial Training for BERT Natural Language Generation

    … Representations from Transformers (BERT). Perplexity and BLEU scores were used as metrics for evaluation on 1000 samples of generated text. Results indicated that perplexity decreased and BLEU scores comparing the original data distributions increased; thus, there was evidence that quality …

    washington Repository record for The Suitability of Generative Adversarial Training for BERT Natural Language Generation (opens in a new tab)

  7. Data analysis applications in software engineering

    … focus on understanding model confidence through perplexity analysis. This work provides the first large-scale, language-specific analysis of code perplexity measured through LLMs, offering insights into how model confidence varies across programming languages and how this relates to code quality …

    greece Repository record for Data analysis applications in software engineering (opens in a new tab)

  8. Comprehensive assessment and quantification of incoherent speech using natural language processing

    … is limited exploration on using language model perplexity to assess coherence, especially given the recent advancement of large language models (LLM). This work bridges the gaps through the following contributions: 1) Two new global coherence assessment methods were developed based on centroids …

    washington Repository record for Comprehensive assessment and quantification of incoherent speech using natural language processing (opens in a new tab)

  9. Language Models can Generalize from Indirect Evidence: Evidence from Filtered Corpus Training (FICT)

    … transformers are better qua LMs (as measured by perplexity), both models perform equally and surprisingly well on linguistic generalization measures, suggesting that they are capable of generalizing from indirect evidence. This adds to a growing body of evidence on the limitations of perplexity …

    washington Repository record for Language Models can Generalize from Indirect Evidence: Evidence from Filtered Corpus Training (FICT) (opens in a new tab)

  10. Dataset Augmentation for LLM Reasoning

    … This is reflected in higher likelihood/ perplexity scores prior to fine-tuning, indicating that the model is more accustomed to generating LLM-style responses. To leverage this insight, we propose a minimum-change strategy, where we minimally edit the target model’s own predictions to …

    adelaide Repository record for Dataset Augmentation for LLM Reasoning (opens in a new tab)

  11. Novel Methods for Natural Language Modeling and Pretraining

    … modeling language sequences to achieve lower perplexity, better generation, and benefit downstream language tasks; specifically, this thesis addresses the importance of natural language features including the segmentation feature, lexical feature, and alignment feature. We present three new …

    waterloo Repository record for Novel Methods for Natural Language Modeling and Pretraining (opens in a new tab)

  12. Leveraging Training Data from High-Resource Languages to Improve Dependency Parsing for Low-Resource Languages

    … attempted into instance selection based on the perplexity of part-of-speech tag sequences, using the Google Universal Dependency Treebank, which spans ten languages. Additionally, I leverage another instance selection technique based on cross-entropy difference, which has shown superior results …

    washington Repository record for Leveraging Training Data from High-Resource Languages to Improve Dependency Parsing for Low-Resource Languages (opens in a new tab)

  13. Expertní systém pro kardiologii

    … výstupy s jazykovými modely ChatGPT, Gemini a Perplexity AI. Výsledky ukázaly, že přestože chatboti někdy určují správnou diagnózu, postrádají odborné hodnocení relevance hypotéz, které expertní systém nabízí. Závěr potvrzuje, že expertní systémy zůstávají nenahraditelné v odborné diagnostice.

    brno-tech Repository record for Expertní systém pro kardiologii (opens in a new tab)

  14. Speaker dynamics as a source of pronunciation variability for continuous speech recognition models

    … rate of 2.3% relative and similar reductions in perplexity over a baseline model using only phonetic context.Incorporating higher-level information (such as hypothesis-dependent word context or word-level FO values) into ASR systems requires a rescoring approach. A framework for this is …

    washington Repository record for Speaker dynamics as a source of pronunciation variability for continuous speech recognition models (opens in a new tab)

  15. An evaluation of latent Dirichlet allocation in the context of plant-pollinator networks

    … a new model selection score called ``penalized perplexity". The performance of the model selection criteria, and of LDA in general, are evaluated through a comprehensive simulation study that consider networks of various size along with varying levels of nesting and numbers of interaction …

    guelph Repository record for An evaluation of latent Dirichlet allocation in the context of plant-pollinator networks (opens in a new tab)

  16. Advances in Fully-Automatic and Interactive Phrase-Based Statistical Machine Translation

    … technique obtains phrase models with lower perplexity than those obtained by means of the standard estimation technique. However, the reduction in the perplexity of the model did not allow us to obtain improvements in the translation quality. To deal with the search problem, we propose a …

    upv Repository record for Advances in Fully-Automatic and Interactive Phrase-Based Statistical Machine Translation (opens in a new tab)

  17. Adaptation and Augmentation: Towards Better Rescoring Strategies for Automatic Speech Recognition and Spoken Term Detection

    … corpus show a 30.2% relative reduction in perplexity at the word-cluster level and a 2.3% relative reduction in WER in a state-of-the-art Italian ASR system. (2) Incorporating source application information associated with speech queries. By exploring a range of adaptation model …

    cuny-grad Repository record for Adaptation and Augmentation: Towards Better Rescoring Strategies for Automatic Speech Recognition and Spoken Term Detection (opens in a new tab)

  18. Low-Rank RNN Adaptation for Context-Aware Language Modeling

    … layer is always helpful, as measured by perplexity, the standard for evaluating language models. We also demonstrate impact on two applications: personalized query completion and context-specific text generation, finding that the enhanced adaptation benefits both. We also show that the …

    washington Repository record for Low-Rank RNN Adaptation for Context-Aware Language Modeling (opens in a new tab)

  19. Modelování jazyka v rozpoznávání češtiny

    … mluveném korpusu (cca 1 170 000 slov) - redukce perplexity o zhruba 20%. Výsledky dosažené při reskórování N-best listů ukazují zlepšení při rozpoznávání spontánní řeči o více než 1%. V závěru práce jsou uvedeny možnosti využití práce, její možná rozšíření a také jsou uvedeny hlavní nevýhody …

    brno-tech Repository record for Modelování jazyka v rozpoznávání češtiny (opens in a new tab)

  20. Syllables and Concepts in Large Vocabulary Speech Recognition

    … language model alone. The study reports that the perplexity of a trigram syllable language model drops by half when compared to a trigram word language model using the same training transcript. The drop in perplexity carries over to error rate. The error rate of a recognizer equipped with syllable …

    unm Repository record for Syllables and Concepts in Large Vocabulary Speech Recognition (opens in a new tab)

Page 1 of 7