Sentence examples for malicious training from inspiring English sources

Malicious training samples are kept indefinitely.

Malicious training samples older than four time periods are discarded Fig. 14 SWF-KeepMal dataset.

As shown in the script below, the labels are first overwritten with ones to indicate benign samples for training and zeros for malicious training samples.

Access to more malicious training data also increased the false positive rate, but the increase for the variant with numerical features remained within bounds, except for the last time period.

Applied on SWF-KeepMal, where malicious training samples are accumulated over time, Hidost has an overall much higher AUROC than on SWF-Normal, where malware is discarded after four periods.

In order to quantify the effect of high class imbalance on detection performance, we generated another data partitioning just for SWF data, labeled SWF-KeepMal and illustrated in Fig. 14, in which malicious training samples older than four periods are not discarded.

These findings clearly show Hidost's potential for further improvement of detection performance, given a greater availability of malicious SWF training data.

Using 10% malicious files in the training set showed a clear trend in which the performance improves when the training set is updated on a yearly basis.

Then, we repeated the experiments with 70% malicious nodes in the training set.

To test this hypothesis, we conducted further experiments, this time using different percentages of malicious nodes in the training sets.

For the second set of experiments, we used 30% malicious nodes in the training set and repeated the experiments with the same test sets as before.