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 21 for “"Random weights"”.

  1. Performance Analysis of Fixed-Random Weights in Artificial Neural Networks

    … To alleviate these concerns, the idea of fixed-random weights in deep neural networks is explored. More critically the goal is to maintain performance akin to fully trained models. Metrics such as floating point operations per second and memory size are compared and contrasted for fixed-random …

    rit Repository record for Performance Analysis of Fixed-Random Weights in Artificial Neural Networks (opens in a new tab)

  2. Tail asymptotics of a sum of heavy-tailed distributions with application to risk theory /

    … both tail probability ant tail expectation of a randomly weighted sum with application to risk theory. In this sum primary random variables are real- valued and heavy-tailed, and random weights are nonnegative and nondegenerated at zero. The asymptotic analysis is carried out for various …

    vilnius Repository record for Tail asymptotics of a sum of heavy-tailed distributions with application to risk theory / (opens in a new tab)

  3. Sunkiauodegių skirstinių sumos asimptotika rizikos teorijoje /

    … both tail probability ant tail expectation of a randomly weighted sum with application to risk theory. In this sum primary random variables are real- valued and heavy-tailed, and random weights are nonnegative and nondegenerated at zero. The asymptotic analysis is carried out for various …

    vilnius Repository record for Sunkiauodegių skirstinių sumos asimptotika rizikos teorijoje / (opens in a new tab)

  4. Tiesinis neuroninių tinklų kombinavimas /

    … function. Various types of theoretically optimal weights calculation (TOWC) functions are analyzed. Finished experiments. The aim of this work is to investigate whether it is possible to improve the quality of MCS using the initial TOWC as initialization of weights of combiner-perceptron. The …

    vilnius Repository record for Tiesinis neuroninių tinklų kombinavimas / (opens in a new tab)

  5. Bayesian Gmm

    … A mixture of parametric distributions with random weights is used to flexibly model an unknown data generating process. The random mixture weights are defined by the exponential tilting projection method to ensure that the joint distribution of the data distribution and the structural …

    penn Repository record for Bayesian Gmm (opens in a new tab)

  6. Asymptotic analysis of the sums of heavy-tailed random variables /

    … nonidentically distributed heavy-tailed random variables are investigated. With some conditions for the (heavy-tailed) distribution of maximal element, the weak max-sum equivalence is proved. Some copula-based examples of dependence structures are given. The sums with dependent …

    vilnius Repository record for Asymptotic analysis of the sums of heavy-tailed random variables / (opens in a new tab)

  7. Qualitative and Quantitative Analysis of Weighted Ergodic Theorems

    … analysis of weighted operators, where the weights are obtained by almost everywhere sampling in a stationary stochastic process. The major theme of our investigation is whether one can break the duality, in other words, if one still gets convergence if the duality restriction is removed. …

    uiuc Repository record for Qualitative and Quantitative Analysis of Weighted Ergodic Theorems (opens in a new tab)

  8. Sunkiauodegių atsitiktinių dydžių sumų asimptotinė analizė /

    … nonidentically distributed heavy-tailed random variables are investigated. With some conditions for the (heavy-tailed) distribution of maximal element, the weak max-sum equivalence is proved. Some copula-based examples of dependence structures are given. The sums with dependent …

    vilnius Repository record for Sunkiauodegių atsitiktinių dydžių sumų asimptotinė analizė / (opens in a new tab)

  9. Approximate universal artificial intelligence and self-play learning for games

    … introduction of a new technique for training the weights of a heuristic evaluation function from data collected by classical game tree search algorithms. This method is shown to outperform previous self-play training routines based on Temporal Difference learning when applied to the game of Chess. …

    unsw Repository record for Approximate universal artificial intelligence and self-play learning for games (opens in a new tab)

  10. Understanding 1D Convolutional Neural Networks Using Multiclass Time-Varying Signals

    … used to train, test and to analyze the learned weights. The field of digital signal processing (DSP) gives a lot of insight into understanding the seemingly random weights learned by CNN. In particular, the concepts of Fourier transform, Savitzky-Golay filters, Guassian filters and FIR filter design …

    clemson Repository record for Understanding 1D Convolutional Neural Networks Using Multiclass Time-Varying Signals (opens in a new tab)

  11. Design Considerations for Training Memristor Crossbars Used in Spiking Neural Networks

    … computing, the training time is curtailed due to random weight initialization in the hidden layer, which will remain constant during training. The CMOS/memristor variability can be exploited to generate these random weights and reduce the area overhead. Recent studies have shown that the …

    rit Repository record for Design Considerations for Training Memristor Crossbars Used in Spiking Neural Networks (opens in a new tab)

  12. MixtureGrowth: growing neural networks by recombining learned parameters

    … one can grow from a small network by adding random weights over time to gradually achieve the target network size. However, this naive approach falls short in practice as it brings too much noise to the growing process. Prior work tackled this issue by leveraging the already learned weights …

    bu Repository record for MixtureGrowth: growing neural networks by recombining learned parameters (opens in a new tab)

  13. Statistical analysis of networks with community structure and bootstrap methods for big data

    … We develop a resampling method called Dependent Random Weighting (DRW) for irregularly spaced dependent data, where instead of using blocks we use random weights to resample the data. By allowing the random weights to be dependent, the dependency structure of the data can be preserved in the …

    uiuc Repository record for Statistical analysis of networks with community structure and bootstrap methods for big data (opens in a new tab)

  14. Topics in percolation and sequence analysis

    … Bernoulli last passage percolation with i.i.d.~weights on vertices over a $n\times n$ grid and for $n$ large enough, the geodesics are shown to be concentrated in a cylinder, centered on the main diagonal and of width of order $n^{(2\kappa+2)/(2\kappa+3)}\sqrt{\ln n}$, where $1\le\kappa<\infty$ …

    gatech Repository record for Topics in percolation and sequence analysis (opens in a new tab)

  15. A sparsity-free compressed sensing theory with applications in generative model recovery

    … A∈ℝᵐˣⁿ; therefore allowing for a family of random measurement matrices which may have heavy tails, dependent rows and columns, and a large dynamic range for the singular values. The structure is either given as a non-convex cone T⊂ℝⁿ, or is induced via minimizing a given convex function …

    ubc Repository record for A sparsity-free compressed sensing theory with applications in generative model recovery (opens in a new tab) Download PDF for A sparsity-free compressed sensing theory with applications in generative model recovery (opens in a new tab)

  16. Short-term hydro-meteorological forecasting with extreme learning machines

    … a feedforward neural network model which assigns random weights in the single hidden layer and optimizes only the weights in the output layer, has the fully nonlinear modelling capability of the traditional artificial neural network (ANN) model but is solved via linear least squares, as in …

    ubc Repository record for Short-term hydro-meteorological forecasting with extreme learning machines (opens in a new tab) Download PDF for Short-term hydro-meteorological forecasting with extreme learning machines (opens in a new tab)

  17. Application of Bio-inspired Algorithms to Selected Real-World Problems

    … high-quality feasible solutions difficult. Randomised methods such as bio-inspired algorithms can solve intractable combinatorial problems in a reasonable timeframe. Bio-inspired algorithms can solve a variety of problems by using operators inspired by nature. The thesis focuses on the …

    adelaide Repository record for Application of Bio-inspired Algorithms to Selected Real-World Problems (opens in a new tab)

  18. Asymmetric particle systems and last-passage percolation in one and two dimensions

    … is directed last-passage percolation on the random graph <em>G</em> = (<em>V,E</em>) where <em>V</em> = ℤ and each edge (<em>i,j</em>), for <em>i</em> <: <em>j</em> ∈ , is present in <em>E</em> independently with some probability <em>p</em> ∈ (0,1]. To every (<em>i,j</em>) ∈ <em>E</em> we …

    oxford Repository record for Asymmetric particle systems and last-passage percolation in one and two dimensions (opens in a new tab)

  19. Artificial intelligence in dentistry: Scoping review and bridging observed knowledge gaps via a methodological study and a clinical trial

    … on diagnostic accuracy and decision-making in a randomized controlled trial. Methods: The scoping review included studies using ML in dentistry published between 1st January 2015 and 31st May 2021 on MEDLINE, IEEE Xplore, and arXiv. The risk of bias and reporting quality were assessed with the …

    fu-berlin Repository record for Artificial intelligence in dentistry: Scoping review and bridging observed knowledge gaps via a methodological study and a clinical trial (opens in a new tab)

  20. Analyzing robustness of models of chaotic dynamical systems learned from data with Echo state networks

    … and in training parameters (initialization of random weights, reservoir size, spectral radius). Others are open theoretical questions such as: why do ESNs work at all, in particular, which aspects of the underlying dynamical systems are captured by the learned reservoirs, and which factors …

    rice Repository record for Analyzing robustness of models of chaotic dynamical systems learned from data with Echo state networks (opens in a new tab)

Page 1 of 2