this is for holding javascript data
Dylan Freedman edited InitialResults.tex
about 9 years ago
Commit id: a1e1a7f7750b986a5d09c7d12eb7f07c4c2cd033
deletions | additions
diff --git a/InitialResults.tex b/InitialResults.tex
index e0e652a..b510e81 100644
--- a/InitialResults.tex
+++ b/InitialResults.tex
...
\chapter{Initial Results} \chapter{Results}
\section{Variables Used and Notation}
This section details a number of experiments run on the data. To compare chord progressions, only the Smith-Waterman algorithm ($SW$) is used, with normalization measures \textit{raw score} and ${SW}_{norm}$ being tested. Gap costs (${gap}_{open}$ and ${gap}_{ext}$) are allowed to vary from 0 through 128 (maximum range of an 8-bit integer). The Harte distance metric ($Harte$) and Tonal Pitch Space ($TPS$) are used to evaluate chord distances. Lastly, multiplication and subtraction factors $m_x$ and $m_s$ are used to round the chord distance metrics to integers with an expected value below 0. A full summary of the variables and their tested ranges is as follows:
\begin{align*}
\textbf{Variables} & \hspace{1cm} & \textbf{Values} \\
\text{Normalization } (norm) && \{\textit{raw score},{SW}_{norm}\} \\
\text{Gap open cost } ({gap}_{open}) && [0-128] \\
\text{Gap extension cost } ({gap}_{ext}) && [0-128] \\
\text{Chord Distance Metric } (C_d) && \{Harte, TPS\} \\
\text{Chord Distance Multiplier } (m_x) && \{1, 30\} \\
\text{Chord Distance Subtraction Factor } (m_s) && \{0, 30\} \\
\end{align*}
\subsection{Smith-Waterman Normalization}
\subsection{Smith-Waterman Gap Costs}
\subsection{Chord Distance Function}
\subsection{Pairwise distance computation}
\subsection{Clustering}
\subsection{Visualization}
\subsection{Ranking Fully Connected Pairwise Comparisons}
\subsection{Ranking Random N-Gram Search}
\subsection{Key-Finding Accuracy}
\section{Smith-Waterman Results}