this is for holding javascript data
Timothy O'Donnell edited section_Results_We_evaluated_the__.tex
almost 8 years ago
Commit id: f8f5b3b47c6787035ce109674d274d7927ef1dc1
deletions | additions
diff --git a/section_Results_We_evaluated_the__.tex b/section_Results_We_evaluated_the__.tex
index bf35b8a..2bcd682 100644
--- a/section_Results_We_evaluated_the__.tex
+++ b/section_Results_We_evaluated_the__.tex
...
\end{table}
\section{Discussion}
Imputing training data shows promise in cross-validation as a way to improve performance on alleles with few observations, but only seems to help for very small training sizes ($\leq 100$). Unfortunately, none of the alleles included in the BLIND dataset had fewer than 100 samples in
BD2009 (in fact, BD2009, and only one
allele had fewer than
200). 200. Thus, additional work is required to assess the accuracy of MHCflurry and other predictors on alleles with scarce training data. Additionally, we need to further investigate the interaction between imputation parameters, the decay schedule for the weights of imputed samples, and stopping criteria for training individual allele-specific predictors.
% These are generated in the 'paper plots' notebook; do not edit by hand.