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 143 for “"Softmax"”.

  1. A Responsible Softmax Layer in Deep Learning

    … present a new algorithm which I call responsible softmax for doing classification with neural networks. This algorithm is intended to handle imbalanced training sets and is accomplished via multiplication by per class weights. These weights may be interpreted as class probabilities for a …

    arizona-thes Repository record for A Responsible Softmax Layer in Deep Learning (opens in a new tab)

  2. Conditional Replicated Softmax for Topic Modelling with Metadata

    … thesis introduces the conditional replicated softmax model, which is an undirected graphical model that models document word counts and document specific metadata using restricted Boltzmann machines. An additional input layer that is associated with the metadata is added to the replicated …

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

  3. On the Softmax Bottleneck of Word-Level Recurrent Language Models

    … over all the words in the vocabulary using a softmax function. When the log of probability outputs for all such contexts are stacked together, the resulting matrix is a log probability matrix which can be denoted as Q_theta, where theta denotes the model parameters. When language modeling is …

    ottawa-retro Repository record for On the Softmax Bottleneck of Word-Level Recurrent Language Models (opens in a new tab)

  4. RTL design and analysis of Softmax Layer in Deep Neural Networks

    … on the nature of training and inference tasks. Softmax Layer is a critical layer in DNNs and is usually used as the output layer in multi-category classification tasks. Softmax layer involves exponentiation and division, thereby resulting in high computational complexity and long critical paths. …

    texas Repository record for RTL design and analysis of Softmax Layer in Deep Neural Networks (opens in a new tab)

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

    … addresses 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)

  6. Approximation Schemes to Simplify Posterior Computation

    … approximate the max of affine functions with the softmax of affine functions. Convex regression methods that use the max of affine functions appear to do well in traditional frequentist settings, but does not scale well to large data in Bayesian settings. We propose the softmax-affine convex (SMA) …

    tamu Repository record for Approximation Schemes to Simplify Posterior Computation (opens in a new tab)

  7. Differentiable Hebbian Consolidation for Continual Lifelong Learning

    … model which replaces the traditional softmax layer with a Differentiable Hebbian Plasticity (DHP) Softmax that adds a fast learning plastic component to the fixed (slowly changing) parameters of the softmax output layer. Similar to the hippocampal system in Complementary Learning …

    guelph Repository record for Differentiable Hebbian Consolidation for Continual Lifelong Learning (opens in a new tab)

  8. Biological Plausibility in Modern Hopfield Networks

    … these implementations often still utilize the softmax function. Computing the softmax for a single node in a layer requires the knowledge of all other nodes in that layer, resulting in non-local computations. The softmax function also causes each target memory stored in the network to …

    waterloo Repository record for Biological Plausibility in Modern Hopfield Networks (opens in a new tab)

  9. CONTROLLED EVALUATION OF ATTENTION MECHANISMS FOR THE WIDEST PATH PROBLEM

    … architecture and attention variants (softmax, adaptive, and tropical) to a new bottleneck-based problem, the WPP. Models are trained in a supervised setting using exact DP solutions and evaluated under both in-distribution (ID) and out-of-distribution (OOD) conditions, including …

    nps Repository record for CONTROLLED EVALUATION OF ATTENTION MECHANISMS FOR THE WIDEST PATH PROBLEM (opens in a new tab)

  10. Data parallel hardware accelerators for machine learning algorithms

    … step. Also, we propose a SIMD-based softmax hardware unit that flexibly supports both full-vector and segmented softmax computations. This design enables the unit to compute a complete softmax over an input vector or multiple smaller softmax operations over its sub-vectors. The …

    greece Repository record for Data parallel hardware accelerators for machine learning algorithms (opens in a new tab)

  11. OUT-OF-DISTRIBUTION RELIABILITY ESTIMATION FOR DRONE OBJECT DETECTION UNDER UNSEEN-DRONE SIMILARITY SHIFT USING SYNTHETIC DATA

    … included native detector confidence, maximum softmax probability, maximum softmax probability with temperature scaling, energy-based scoring, and Mahalanobis scoring based on internal visual features. Performance was measured using AUROC and FPR95, with class-score and feature analyses used to …

    nps Repository record for OUT-OF-DISTRIBUTION RELIABILITY ESTIMATION FOR DRONE OBJECT DETECTION UNDER UNSEEN-DRONE SIMILARITY SHIFT USING SYNTHETIC DATA (opens in a new tab)

  12. Constellation Design for Multi-user Communications with Deep Learning

    … We propose a method based on optimization of softmax and introduce the L1/2 regularization in MSE loss function for SISO case and MIMO case, separately. The simulation shows that both our optimized softmax function method and L1/2 regularization loss function have a better performance than the …

    mcmaster Repository record for Constellation Design for Multi-user Communications with Deep Learning (opens in a new tab)

  13. Performance Primitives for Artificial Neural Networks

    … operators and two optimized building blocks for softmax operators. High-intensity convolutional operators with large kernel sizes and unit stride benefit from asymptotically fast convolution algorithms based on Winograd transform and Fast Fourier transform. I jointly consider Fourier or Winograd …

    gatech Repository record for Performance Primitives for Artificial Neural Networks (opens in a new tab)

  14. Enhancing Anomaly Detection with Flexible Distribution Models

    … data. Then, we propose an approximation of the softmax output vector of a classifier with generalized Dirichlet (GD), scaled Dirichlet (SD), shifted scaled Dirichlet (SSD), and Beta-Liouville (BL) distributions. We use a technique based on likelihood to determine its parameters. Motivated by the …

    concordia Download PDF for Enhancing Anomaly Detection with Flexible Distribution Models (opens in a new tab)

  15. Enhancing intrusion detection in vehicular networks through deep learning approaches

    … Gated Recurrent Units (BiGRU), and a softmax classifier. The LSTMVAE is employed as a statistical feature extraction technique capable of learning time series and multivariate data from the IoV network. The extracted features are then fed into the BiGRU and softmax classifier for the …

    ets-quebec Download PDF for Enhancing intrusion detection in vehicular networks through deep learning approaches (opens in a new tab)

  16. Adaptive learning approaches for smart home environments with a simulator implementation

    … approaches, namely K-Nearest Neighbor and Softmax Regression with batch learning and online learning algorithms, are evaluated with different datasets representing different scenarios, and the results show that all three methods are able to perfectly capture the usage pattern when the …

    texas Repository record for Adaptive learning approaches for smart home environments with a simulator implementation (opens in a new tab)

  17. Improving Training of Deep Neural Network Sequence Models

    … bottleneck of neural training which is the softmax operation. Among different softmax approximation techniques, Noise Contrastive Estimation (NCE) is seen as a method that often does not work well with deep neural models for language modelling. A thorough investigation was done to find out …

    kent Download PDF for Improving Training of Deep Neural Network Sequence Models (opens in a new tab)

  18. Uncertainty-aware weather-related road surface condition classification

    … The methods used to detect samples were Maximum Softmax Probability, Energy-Based OOD detection, Outlier Exposure, Virtual Outlier Synthesis, Rectified Activations, Deep Nearest Neighbour, and Virtual Logit Matching. Evaluation of the methods involved three datasets - Cars, Slush and Glare - …

    chalmers Repository record for Uncertainty-aware weather-related road surface condition classification (opens in a new tab)

  19. Modelling variations in human learning in probabilistic decision-making tasks

    … different combinations of parameters. Replacing softmax with epsilon greedy, as the way of calculating the action choice probabilities, was found to improve parameter recovery in simulated data. Datasets from three existing unpublished probabilistic decision-making tasks were examined. These …

    goldsmiths Repository record for Modelling variations in human learning in probabilistic decision-making tasks (opens in a new tab) Download PDF for Modelling variations in human learning in probabilistic decision-making tasks (opens in a new tab)

  20. Investicinio portfelio sudarymo ir optimizavimo metodų palyginimas /

    … The improved versions are generated using the softmax function, which transforms signals into regime probabilities, enabling the formation of a regime aware portfolio structure. Results show that incorporating market signals into investment strate gies can improve performance, especially when …

    vilnius Repository record for Investicinio portfelio sudarymo ir optimizavimo metodų palyginimas / (opens in a new tab)

Page 1 of 8