PDF Publication Title:
Text from PDF Page: 143
which is not a bad result considering the large number of values of this feature. It would be necessary to obtain more training data to achieve a better result in the recovery of features with many values, due to the greater uncertainty associated with a label with more values. The remaining metrics (F1, precision, and recall) follow a similar pattern (Figure 8). Figure 8. Performance metrics (aggregated) for predicting missing features of NSL- KDD test dataset. Figures 9–11 provide detailed results on reconstruction performance for each of the three recovered features (protocol, flag, and service). Each figure provides the name and frequency distribution for feature’s values, together with one vs. rest reconstruction metrics. The reconstruction metrics are given for each value of the reconstructed feature (one vs. rest). We can see that in all cases the frequency distribution of values is remarkably unbalanced (column “Frequency” in Figures 9–11). The unbalanced distribution creates additional problems to the recovery task. There are cases where this unbalanced scenario is so strong that the algorithm cannot recover certain values of the reconstructed feature. This is the case in Figure 10, where several rare values of the flag feature (S1, S2, and RSTOS0) have an F1 equal or close to zero, implying that we cannot predict any positive occurrence of these values. This happens for values with extremely low frequency (less than 0.09%). Figure 9. Performance metrics (One vs. Rest) for reconstruction of all features values when feature: ‘protocol’ is missing. Doctoral Thesis: Novel applications of Machine Learning to NTAP - 141PDF Image | Novel applications of Machine Learning to Network Traffic Analysis
PDF Search Title:
Novel applications of Machine Learning to Network Traffic AnalysisOriginal File Name Searched:
456453_1175348.pdfDIY PDF Search: Google It | Yahoo | Bing
Cruise Ship Reviews | Luxury Resort | Jet | Yacht | and Travel Tech More Info
Cruising Review Topics and Articles More Info
Software based on Filemaker for the travel industry More Info
The Burgenstock Resort: Reviews on CruisingReview website... More Info
Resort Reviews: World Class resorts... More Info
The Riffelalp Resort: Reviews on CruisingReview website... More Info
| CONTACT TEL: 608-238-6001 Email: greg@cruisingreview.com | RSS | AMP |