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 26 for “"Convolution kernels"”.
-
EXTENDING DIOSPYROS AUTO-VECTORIZATION TO LLVM
… Diospyros LLVM pass translating linear algebraic kernels to SIMD vector architectures. The pass uses the idea from the Diospyros project to apply a search-based equality saturation method to find vector instructions. Results show that the Diospyros LLVM pass generates vector code with good runtime …
-
On the Performance of Line Integral Convolution in a Distributed-Memory Parallel Setting
Line integral convolution (LIC) is a powerful tool for visualizing vector fields by combining particle advection with image convolution. Practical usage of LIC is limited by its computational expense, requiring many calculations for every cell in the mesh. Fortunately, computation of LIC can be …
-
Perturbation theory in extended systems: Sharp nonlocal center manifolds and weakly quenched planar fronts
… nonlocal systems with exponentially localized convolution kernels. Technically, the use of C0-based spaces allows for optimal regularity of the center manifold in a certain set of coordinates, as well as allowing for a simpler cutoff argument for existence of the manifold. As an application, we …
-
Properties of convolution operators on Lp(0,1)
Convolution operators on Lp(0,1) have many similarities with the classical Volterra operator V, but it is not known in general for which convolution kernels the resulting operator behaves like V. It is shown that many convolution operators are cyclic, and the cyclic property is related to the …
-
Two -dimensional cubic convolution for image interpolation, restoration, and super -resolution
Cubic convolution has been used for image interpolation since the 1970's and provides a good compromise between computational complexity and interpolation accuracy. Cubic convolution can be parameterized and then optimized either for general performance characteristics or for optimal fidelity over …
-
Near-to-Far Field Signal Propagation for the Wave and Maxwell Equations
… We achieve such a conversion via application of convolution kernels in the time-domain, although the derivation of the appropriate kernels relies on Laplace transform arguments. Decomposing the wave and Maxwell equations using scalar and vector spherical harmonics respectively, we have solved the …
-
A method of optimal image subtraction: development of the mathematics and software for general use in astronomical research
… of proofs of theorems involving two-dimensional convolutions. The OIS method uses sub-images, called “stamps”, that are small sections of images, each containing one star, used to define the kernel. An automated stamp selection procedure was designed utilizing utilities from the Goddard Space …
-
Opportunistic Breast Cancer Detection with Photon-Counting CT: A Quantitative Pilot Study of Reconstruction-Dependent Image Quality
… reconstructions were performed with varying convolution kernels, field-of-view sizes, and virtual monoenergetic spectra. Image quality was assessed using signal-to-noise ratio (SNR), dose-normalized contrast-to-noise ratio (CNRD), and spatial resolution quantified via full width at …
-
Experimental Study of Highly-Viscous Vertical Pipe Flow Using a Non-Intrusive Multi-View Measurement Technique
… multi-view measurement system. Combinations of convolution kernels and thresholding methods perform boundary detection, light reflection corrections and geometrical parameter measurements in the software ImageJ (Fiji). Receiver Operating Characteristics (ROC) plots based on a True Positive Rate …
-
Topics in Harmonic Analysis, Sparse Representations, and Data Analysis
… transforms, which are particular realizations of convolutional neural networks. While the latter uses trained convolution kernels, scattering transforms use fixed ones, and this simplification allows mathematicians to develop a model of deep learning. Mallat originally introduced a wavelet …
-
Probabilistic Modeling of Rumour Stance and Popularity in Social Media
… the potential to be used in other applications: convolution kernels for streams of text over continuous time and a sequence classification algorithm based on point processes.
-
Methods of fast Fourier transform in diffraction problems of elastic and acoustic waves with applications to crack mechanics
… of fast methods for integral equations with convolution kernels arising in these fields of application. Such integral equations, after discretization, may be reduced to linear algebraic systems with matrix of Toepliz or circulant form. For both the types there can be applied fast iteration …
-
Pruning Convolution Neural Network (SqueezeNet) for Efficient Hardware Deployment
… focuses on reducing the model size of the Convolution Neural Network (CNN) by various compression techniques like Architectural compression, Pruning, Quantization, and Encoding (e.g., Huffman encoding). Network pruning is one of the promising technique to solve these problems. This thesis …
-
Information Theoretic Graph Kernels
… on developing novel information theoretic kernels for graphs. To this end, we commence in Chapter 3 by defining a family of Jensen-Shannon diffusion kernels, i.e., the information theoretic kernels, for (un)attributed graphs. We show that our kernels overcome the shortcomings of …
-
Evolutionary Optimisation in Convolutional Neural Networks
… is proposed to reduce the size and number of convolution kernels in each convolutional layer and generate trade-offs between computational cost and model's classification accuracy. For data representation optimisation, an evolutionary-based adaptive integer quantisation methodology is …
-
Filtered polynomial approximation on the sphere
… on the sphere. Using filtered polynomial kernels and a special numerical integration (quadrature) rule we construct a fully discrete needlet approximation. The localisation of the filtered approximation can be seen from the localisation properties of its convolution kernel. We investigate …
-
Modelling spatio-angular patterns in cell biology
… of partial differential equations containing convolution terms. Such models appear to contain a rich diversity of possible behaviour and dynamics, depending on the details of the convolution kernels involved. The analysis of the equations, and predictions in several test cases are presented. …
-
Deep Learning for 3D Information Extraction from Indoor and Outdoor Point Clouds
… correlations between local points when defining convolution kernels, (2) the hierarchical local and global feature learning within an end-to-end trainable framework, (3) the relation feature learning from nearby objects, and (4) 2D image leveraging for 3D object detection from point clouds. …
-
Hydrodynamic modelling of the oscillating wave surge converter
… models the device as an ordered expansion of convolution integrals. The convolution kernels are extracted from tank test data using a system identification technique.<br/><br/>The accuracy of the hydrodynamic and the 1st order Volterra model is found to be comparable. The accurate prediction …
Page 1 of 2