Specifically, we study the challenging problems of i) learning to detect video events from solely a textual description of the event, without using any positive video examples, and ii) additionally exploiting very few positive training samples together with a small number of "related" videos.
To investigate the substrate specificity of acetylated lysine sites, WebLogo [ 26] was applied to generate the sequence logo of positive training dataset presenting with 15-mer flanking sequences.
We used these peaks as positive training cases, together with a set of randomly drawn control (anti-IgG) ChIP-seq peaks as negative training cases, to train supervised classifiers.
We experimentally validated 18 SRC-1 peak by quantitative PCR (qPCR) experiments (data not shown), which were used as positive training cases, together with a set of randomly drawn control (anti-IgG) ChIP-seq peaks as negative training cases, to train supervised classifiers.
The SVM with RBF kernel is able to model the class of elephants well, although the positive training samples partly overlap with the samples of the background class (see Section 3.2).
While some studies have shown positive training effects from occupations with high physical demands [ 27, 28], other studies on seniors with a history of manual labor have associated a longer lifetime of physically demanding work with higher rates of disability, lowered physical function and reduction in muscular strength [ 29– 31].
In order to avoid a biased prediction performance for a binary classification between positive and negative data, the negative training data was balanced with the positive training data.
Our data, combined with the positive training effect for toes observed by Yoshimura (1960) and Savourey (1996), seem to suggest that toe blood flow is trainable, albeit modestly.
With an average over 1,000 randomizations, we observed that whether it was all set, predicted set, or positive training set, HPV16 preferred to interact with hub proteins (proteins interacting with a large number of partners) in the integrated human PPI network).
Interestingly, the maximum similarity with examples in positive training datasets is 40% for ISYGRK 50) KRRQR, 50% for SYGRKK (51) RRQRR, 70% for YGRKKR 53) RQRRR and 70% for GRKKRR 54) QRRRP, respectively.
