Kevin V. Mack

Kevin V. Mack

PhD Student and graduate research assistant at Clarkson University

Posts by Tags

DCT

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

DMD

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

Equation-free modeling of the n-body problem (using MATLAB)

32 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on the n-body problem in celestial mechanics. In this work, we not only generate a working simulation of the n-body problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling techniques to this system. Details are given in ee520-hw7.pdf.

Delaunay Triangulation

Euler

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

Euler beam

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

Fisher's linear discriminant

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

Fourier basis

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

Fourier transform

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

LDA

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

Rossler Attractor

SVD

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

acoustic processing

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

basis function

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

basis selection

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

beam

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

bpsk

chaos

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

chaotic attractor

chaotic beam

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

classification

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

communications

detection

discrete cosine transform

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

dynamic mode decomposition

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

Equation-free modeling of the n-body problem (using MATLAB)

32 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on the n-body problem in celestial mechanics. In this work, we not only generate a working simulation of the n-body problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling techniques to this system. Details are given in ee520-hw7.pdf.

dynamical systems

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

dynamical systems modeling

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

equation-free modeling

Equation-free modeling of the n-body problem (using MATLAB)

32 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on the n-body problem in celestial mechanics. In this work, we not only generate a working simulation of the n-body problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling techniques to this system. Details are given in ee520-hw7.pdf.

finite element methods

image processing

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

k-means

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

linear discriminant analysis

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

machine learning

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

matlab

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

Equation-free modeling of the n-body problem (using MATLAB)

32 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on the n-body problem in celestial mechanics. In this work, we not only generate a working simulation of the n-body problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling techniques to this system. Details are given in ee520-hw7.pdf.

monte-carlo

n-body

Equation-free modeling of the n-body problem (using MATLAB)

32 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on the n-body problem in celestial mechanics. In this work, we not only generate a working simulation of the n-body problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling techniques to this system. Details are given in ee520-hw7.pdf.

ook

qpsk

singular value decomposition

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

Equation-free modeling of a chaotic (Euler) beam (using MATLAB)

18 minute read

Published:

This blog post will be an old homework assignment from the Data Analysis of Complex Systems course. The assignment focuses on modeling the dynamics of a perturbed beam. In this work, we not only generate a working simulation of the perturbed beam problem using MATLAB’s ordinary differential equation (ode) solvers , we also apply equation-free modeling. Details are given in ee520-hw5.pdf.

supervised learning

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

wavelet basis

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

wavelet decomposition

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.

wavelet transform

Linear Discriminant Analysis on images of cats and dogs and types of music (using MATLAB)

less than 1 minute read

Published:

This post focuses on using an appropriate basis to decompose certain types of data (images and audio) as a preprocessing step to feature learning and classification. We select a reasonable basis based on prior knowledge of images and audio data and intuition, then transform the original datasets into this new space. After multiplying each data sample by the chosen basis, we attempt to find the features of the transformed dataset that are “most useful” for cluster analysis (in this case, linear discriminant analysis (LDA)). The basis selection is done to facilitate dimensionality reduction in the decomposed space. In this way, a single image or audio sample can be represented as a point in $N$-dimensional space, with the $N$ modes corresponding to the features with the most energy.
Details are given in ee520-hw3-A.
The answer to problem 2 is given separately, here.