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 73 for “"Cross-Entropy Loss"”.

  1. Mathematical Guarantees of Near-Optimal Cross-Entropy Loss through Neural Collapse

    … neural network classifiers when the training cross-entropy loss is near-optimal. Theoretically, we show that, within a small neighborhood of the optimal cross-entropy loss, with $C$ target classes, neural network classifiers with batch normalization and weight decay achieve intra-class feature …

    gatech Repository record for Mathematical Guarantees of Near-Optimal Cross-Entropy Loss through Neural Collapse (opens in a new tab)

  2. LiDAR-Based Semantic Segmentation for Marine Surroundings: Optimization strategies for segmentation classification in a marine environment

    … done through examining the size of the network, loss functions, dataset and how it can be preprocessed in different ways. The investigation show that preprocessing the data do not improve the model and that cross entropy loss is the best loss function when the dataset is highly imbalanced. The …

    chalmers Repository record for LiDAR-Based Semantic Segmentation for Marine Surroundings: Optimization strategies for segmentation classification in a marine environment (opens in a new tab)

  3. A study of self-training variants for semi-supervised image classification

    … is achieved by applying several metric learning losses for the training of feedforward neural networks. Experimental results show that triplet loss – a metric learning loss – can achieve better results than cross-entropy loss with simple neural networks. For improving the performance of …

    waikato-masters Repository record for A study of self-training variants for semi-supervised image classification (opens in a new tab) Download file for A study of self-training variants for semi-supervised image classification (opens in a new tab)

  4. Closing the Gap Between Classification and Retrieval Models

    … networks due to multiple reasons: (1) The cross-entropy loss – used with classification networks – is stabler and converges faster compared to metric learning losses – used with retrieval networks. (2) The cross-entropy loss has a huge toolbox of utilities and extensions. For instance, both …

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

  5. Detekce jógových pozic v obraze

    … při použití aktivační funkce sigmoid a binary cross-entropy loss function. I přes slibně vypadající dosažené výsledky jsou hlavním přínosem této práce nástroje pro tvorbu datasetu a vytvořený Dataset samotný. Díky nim byla navrhovaná hypotéza úspěšně ověřena.

    brno-tech Repository record for Detekce jógových pozic v obraze (opens in a new tab)

  6. The Atomistic Reconstruction of Coarse-Grained Polymeric Systems via Machine Learning Techniques

    … CG site of a polyethylene system. A categorical cross-entropy loss function was used to allow for more flexibility in the model. The model’s training yielded consistent loss and validation loss demonstrating that the model did not overfit the data. Furthermore, the model was able to accurately …

    oregon Repository record for The Atomistic Reconstruction of Coarse-Grained Polymeric Systems via Machine Learning Techniques (opens in a new tab)

  7. Baltymų molekulinių struktūrų sąveikų regionų nusakymas naudojant giliojo mokymosi metodus /

    … and was converted to graphs. Model uses weighted cross entropy loss function, as well as Adam optimization algorithm. Matthews correlation coefficient (MCC) was employed as scoring algorithm. Optimized version of this model was based on graph convolution network with three layers which inputs’ …

    vilnius Repository record for Baltymų molekulinių struktūrų sąveikų regionų nusakymas naudojant giliojo mokymosi metodus / (opens in a new tab)

  8. Contrastive Learning, with Application to Forensic Identification of Source

    … between models trained by contrastive loss and models trained by cross-entropy loss. We also perform an ablation study to investigate the effects of different contrastive loss functions, metric functions, and margin values within contrastive learning. To test contrastive networks on …

    sdstate Repository record for Contrastive Learning, with Application to Forensic Identification of Source (opens in a new tab)

  9. Exploring Loss Functions in Machine Learning

    The loss function plays a critical role in machine learning. It is fundamental in training, evaluating, and optimizing machine learning models, directly impacting their effectiveness and efficiency in solving specific tasks. We explore three new loss functions and their applications. Softmax …

    claremont Repository record for Exploring Loss Functions in Machine Learning (opens in a new tab)

  10. Symmetric Variational Inference with High Mutual Information

    … data and latent variables; and 3) low marginal entropy, or compression, which tends to encourage clustered latent representations. Taken together, these objectives yield a cross entropy loss for learning latent variable models. The resulting form of amortized, symmetric variational inference …

    toronto-retro Repository record for Symmetric Variational Inference with High Mutual Information (opens in a new tab)

  11. Detection and Monitoring of Tailings Dam Surface Erosion Using UAV and Machine Learning

    … segmentation with UNet architecture and weighted cross entropy loss function for the detection of rills on tailings dams. The final model showed a promising result with precision, recall, and F1-score of 83.3%, 72.0%, 77.2% respectively.

    unr Repository record for Detection and Monitoring of Tailings Dam Surface Erosion Using UAV and Machine Learning (opens in a new tab)

  12. Trajectory-User Linking Using Higher-Order Mobility Flow Representations

    … embeddings, then fine-tuned using a balanced cross-entropy loss to make predictions while handling imbalanced data. Experiments on real-life mobility data show TULHOR's effectiveness as compared to current state-of-the-art models.

    york Repository record for Trajectory-User Linking Using Higher-Order Mobility Flow Representations (opens in a new tab)

  13. Training Reject-Classifiers for Out-of-distribution Detection via Explicit Boundary Sample Generation

    … by minimizing the standard cross-entropy loss on in-distribution samples and minimizing the KL divergence between the predictive distribution of OOD samples in the low-density “boundary” of in-distribution and the uniform distribution (maximizing the entropy of the outputs). …

    waterloo Repository record for Training Reject-Classifiers for Out-of-distribution Detection via Explicit Boundary Sample Generation (opens in a new tab)

  14. Disease Diagnosis Using Natural Language Processing and Deep Learning

    … are trained and evaluated with categorical cross-entropy loss and standard multi-class metrics, including accuracy, F1-score, and ROC-AUC, while hyperparameters such as learning rate, optimizer configuration, and training epochs are systematically tuned. Experimental results show that …

    eiu Repository record for Disease Diagnosis Using Natural Language Processing and Deep Learning (opens in a new tab)

  15. ONBOARD SHIP CLASSIFICATION FOR MARITIME SURVEILLANCE: BENCHMARKING CONVOLUTIONAL NEURAL NETWORK-BASED TRANSFER LEARNING ON THE NVIDIA JETSON ORIN NANO 8GB

    … are fine-tuned on the ShipRSImageNet dataset across five ship classes: warship, cargo, tanker, fishing, and other. Both models are compared under baseline and improved training configurations, across 13 image degradation conditions, on an out-of-distribution (OOD) validation set from GEGD Pro …

    nps Repository record for ONBOARD SHIP CLASSIFICATION FOR MARITIME SURVEILLANCE: BENCHMARKING CONVOLUTIONAL NEURAL NETWORK-BASED TRANSFER LEARNING ON THE NVIDIA JETSON ORIN NANO 8GB (opens in a new tab)

  16. Development of a Multi-Scale Model for Enhanced Performance by Using a Modified Soft-Margin SoftMax Loss and its Application for Facial Expression Recognition

    … the limitations of the traditional SoftMax Loss function, which tends to overfit by wrongly classifying due to shortage of discriminability through integrating a regularization technique to the conventional Cross-Entropy loss and introducing an adaptive-margin to the standard SoftMax …

    concordia Download PDF for Development of a Multi-Scale Model for Enhanced Performance by Using a Modified Soft-Margin SoftMax Loss and its Application for Facial Expression Recognition (opens in a new tab)

  17. Constraint-based calibration for reliable deep learning models

    … The standard practices of minimizing the cross-entropy loss during training promote the predicted softmax probabilities to match the one-hot label assignments. Nevertheless, this yields a pre-softmax activation of the correct class that is significantly larger than the remaining …

    ets-quebec Download PDF for Constraint-based calibration for reliable deep learning models (opens in a new tab)

  18. Improving the Computational Efficiency of Training and Application of Neural Language Models for Automatic Speech Recognition

    … language modeling, we propose a new loss function that modifies the standard cross-entropy loss that trains the neural network to self-normalize, which we call a linear loss. The linear loss significantly reduces inference-time computation and allows us to use an importance-sampling …

    jhu Repository record for Improving the Computational Efficiency of Training and Application of Neural Language Models for Automatic Speech Recognition (opens in a new tab)

  19. Truncation maps and geometry of deep learning

    … polyhedral cones, and to construct explicit zero-loss classifiers for clustered, sequentially linearly separable, and certain non-separable data sets. Second, the dissertation analyzes training dynamics by comparing standard gradient flow in parameter space with a modified flow adapted to the …

    texas Repository record for Truncation maps and geometry of deep learning (opens in a new tab)

  20. Text-based fashion search system for the lithuanian language using computer vision /

    … were fine-tuned using the symmetric cross entropy loss function on the FashionGen dataset. The fine-tuned models were assessed on retrieval metrics (Recall@K, Precision@K, mean and median rank) and embedding alignment measures (diagonal dominance, accuracy of alignment, contrast …

    vilnius Repository record for Text-based fashion search system for the lithuanian language using computer vision / (opens in a new tab)

Page 1 of 4