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 16 of 16 for “"Model Distillation"”.

  1. Dynamic Knowledge Elicitation: Leveraging Student Feedback For Improved Language Model Distillation

    Large Language Models (LLMs) have significantly advanced the field of natural language processing but remain resource-intensive and impractical for many organizations. Specialist models offer a viable alternative, often developed through Knowledge Distillation (KD) techniques. However, traditional …

    kennesaw Repository record for Dynamic Knowledge Elicitation: Leveraging Student Feedback For Improved Language Model Distillation (opens in a new tab)

  2. Knowledge Distillation as a Solution for Domain Transfer

    … in fact verification tasks. We present a data distillation technique for delexicalization, which we then combine with a model distillation method to prevent aggressive data distillation. We show that by using our solution, not only does the performance of an ex- isting state-of-the-art model …

    arizona-thes Repository record for Knowledge Distillation as a Solution for Domain Transfer (opens in a new tab)

  3. Refining the sample complexity of comparative learning

    … the benchmark class. This setting represents model distillation tasks, where a predictor with specific requirements (e.g., interpretability) is trained on the labels from another model.

    york Repository record for Refining the sample complexity of comparative learning (opens in a new tab)

  4. Modeling and Learning Prosumer Preferences in Peer-to-Peer Energy Sharing Using Federated Techniques

    … introduces three customizable preference models that enable prosumers to express trade-offs across energy-sharing attributes. A three-layer federated learning (FL) architecture is then employed to implement these models in a privacy-preserving and scalable manner, avoiding centralization …

    carleton Repository record for Modeling and Learning Prosumer Preferences in Peer-to-Peer Energy Sharing Using Federated Techniques (opens in a new tab)

  5. Reducing Training Cost and Improving Inference Speed Through Neural Network Compression

    As AI models have become integral to many software applications used in everyday life, the need for ways to run these computationally intensive applications on mobile and edge devices has grown. To help solve these problems, a new research area of neural network compression has emerged. Techniques …

    texas-state Repository record for Reducing Training Cost and Improving Inference Speed Through Neural Network Compression (opens in a new tab)

  6. Enhancing the Performance of the MtCNN for the Classification of Cancer Pathology Reports: From Data Annotation to Model Deployment

    … overfitting and improve the performance of ML models for the classification of cancer pathology reports so that we can reduce the dependency on human expert annotators. We (1) show that using active learning can mitigate extreme class imbalance by increasing the representation of documents …

    utk-theses Repository record for Enhancing the Performance of the MtCNN for the Classification of Cancer Pathology Reports: From Data Annotation to Model Deployment (opens in a new tab)

  7. Exploring video diffusion models in echocardiogram generation : a novel approach to data augmentation in cardiac imaging

    … explores the application of video diffusion models to generate synthetic echocardiograms as a means to augment limited datasets, thereby enhancing ejection fraction estimation models. By leveraging synthetic data generation, we aim to address data scarcity, enhance model performance, and …

    ubc Repository record for Exploring video diffusion models in echocardiogram generation : a novel approach to data augmentation in cardiac imaging (opens in a new tab) Download PDF for Exploring video diffusion models in echocardiogram generation : a novel approach to data augmentation in cardiac imaging (opens in a new tab)

  8. On the Effect of Data on Image Classification Tasks

    … its reliance on large-scale datasets and complex models raises concerns about sustainability and efficiency. This thesis aims to leverage data potential in image classification tasks, focusing on two critical aspects: efficient learning and multi-label representation learning. While existing …

    toronto-retro Repository record for On the Effect of Data on Image Classification Tasks (opens in a new tab)

  9. Statistical Inference for Machine Learning: Feature Importance, Uncertainty Quantification and Interpretation Stability

    … and clinical decisions. However, many models are complex in nature and act as ``black boxes", providing predictions but little insight as to how they were arrived at. In this thesis, we present our work from three different perspectives towards a better understanding of several types of …

    cornell Repository record for Statistical Inference for Machine Learning: Feature Importance, Uncertainty Quantification and Interpretation Stability (opens in a new tab)

  10. ASYMPTOTICS AND INTERPRETABILITY OF DECISION TREES AND DECISION TREE ENSEMBLES

    … are widely used nonparametric statistical models. A decision tree is a binary tree that recursively segments the covariate space along the coordinate directions to create hyper rectangles as basic prediction units for fitting constant values within each of them. A decision tree ensemble …

    cornell Repository record for ASYMPTOTICS AND INTERPRETABILITY OF DECISION TREES AND DECISION TREE ENSEMBLES (opens in a new tab)

  11. Transfer Learning Methods for Object and Action Recognition

    … objective of Domain Adaptation (DA) is to learn models that are invariant to distributional differences between train and test data (i.e., source and target domain data). Zero-Shot Learning (ZSL) is a framework for recognizing instances from classes never seen during training (i.e., missing …

    anu Repository record for Transfer Learning Methods for Object and Action Recognition (opens in a new tab)

  12. Towards Human-Embodied Visual Intelligence

    … I. Low Task Generalizability. Current visual models are often specifically designed for targeted tasks, constraining their generalizability with varying adaptation scenarios; II. Weak Model Interpretability. Within the paradigm of connectionism, most of these models are frequently regarded as …

    rit Repository record for Towards Human-Embodied Visual Intelligence (opens in a new tab)

  13. Robust Submodular Partitioning and Linear Models of Deep ReLU Networks

    Machine learning models, especially deep neural networks, have achieved great success in numerous real-world tasks. As we achieve better performance with larger models, one major challenge emerges that the costs of training machine learning systems become expensive and even prohibitive. Also, the …

    washington Repository record for Robust Submodular Partitioning and Linear Models of Deep ReLU Networks (opens in a new tab)

  14. Neuronové sítě, předtrénování komplexity

    Tato práce se zabývá technikou obrácené destilace modelu neuronových sítí, při které menší model (učitel) vede počáteční fázi tréninku většího modelu (studenta). Cílem je ověřit, zda tento přístup může sloužit jako efektivní způsob inicializace vah a zlepšit průběh tréninku. Experimenty na …

    brno-tech Repository record for Neuronové sítě, předtrénování komplexity (opens in a new tab)

  15. Optimization-Based Solutions for Planning and Control

    … while meeting process constraints. The planning model is typically formulated as a mixed integer nonlinear programming (MINLP), which is hard to solve to global optimality due to nonconvexity and large dimensionality attributes. Uncertainty in component qualities in gasoline blending due to …

    mcmaster Repository record for Optimization-Based Solutions for Planning and Control (opens in a new tab)