Academia.eduAcademia.edu

Figure 4 – uploaded by Mahesh Joshi

See full PDFdownloadDownload figure
Figure 3: Description of dataset syngen. There are 8 total attributes: four categorical attributes distinguish subclasses C3 and NC3 or target and non-target class, respectively. Four numerical attributes, over which the class distributions are shown, distinguish other subclasses. The distribution of a subclass over non- distinguishing attributes is random and uniform.

Figure 3 Description of dataset syngen. There are 8 total attributes: four categorical attributes distinguish subclasses C3 and NC3 or target and non-target class, respectively. Four numerical attributes, over which the class distributions are shown, distinguish other subclasses. The distribution of a subclass over non- distinguishing attributes is random and uniform.

Related Figures (8)

Table 2: Comparative Results on dataset nsyn5. Cte: C4.5-we, Re: RIPPER-we Celne! pate Mae eeeee, EER Peano: Seo EM oe nspa: Each subclass is distinguished by nspa number of disjoint signatures over a distinct pair of attributes. nwps: Each distinct pair of attributes has total nwps combinations of words that identify the corresponding signature. Each subclass of the target-class has same values of nspa and nwps, but different set of distinguishing words. Each subclass of the non-target-class has same values of nspa and nwps, but different set of distinguishing words.
Table 2: Comparative Results on dataset nsyn5. Cte: C4.5-we, Re: RIPPER-we Celne! pate Mae eeeee, EER Peano: Seo EM oe nspa: Each subclass is distinguished by nspa number of disjoint signatures over a distinct pair of attributes. nwps: Each distinct pair of attributes has total nwps combinations of words that identify the corresponding signature. Each subclass of the target-class has same values of nspa and nwps, but different set of distinguishing words. Each subclass of the non-target-class has same values of nspa and nwps, but different set of distinguishing words.
Table 5: Comparing effect of variation in target class size on syngen dataset. ntc-frac is the fraction of non-target-class examples randomly sampled from the dataset with 0.3% target class proportion.
Table 5: Comparing effect of variation in target class size on syngen dataset. ntc-frac is the fraction of non-target-class examples randomly sampled from the dataset with 0.3% target class proportion.
Table 4: Comparative Results on dataset syngen
Table 4: Comparative Results on dataset syngen
Related topics:
Data CompressionNetwork Intrusion Detection & PreventionFraud DetectionRule InductionSynthetic Data Generation
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved