Unsupervised Learning for Spam Email Filtering
Sambhangi Chandrahasa · International Journal of Advanced Trends in Computer Science and Engineering · 2020
Astounding number of features will have a deleterious effect on some learning classifier's performance, furthermore, the operational time frame during the training process for evaluating the content may be enhanced.A circa-processing period that mostly includes extraction of components and elimination of features in the sector of machine learning consequently plays a significant role in expediting or boosting processing precise classification.The concern addressed throughout this thesis is relevant to the integration of results, prior characterizing machine learning.Feature representation that restores class separability to less dimensionality to detect content.The key benefit with accordance to the envisaged feature representation has always been its rigidity, that mostly empowers data types like those of Random Forest, Assistance Vector Machines, and constraint solver C4.5 to characterize an inbound text as fraud or pseudo-spam at which the component dimension seems to be very limited under a solid truism unaware of the reference source.