Features we need: A refinement of IDS dataset

To analyze any network for good or bad traffic, we must have a refined dataset that will help us to train the Intrusion Detection systems in a more efficient manner. The larger the dimension of the data in hand, the longer it will take for any algorithm to make itself aware about the type of data it is dealing with. During our experimental process of designing efficient algorithms, we tried to filter the data in both dimensions. The result was a pruned dataset which contains only those attributes whose values are primarily differentiated during intrusions and can therefore easily lead to identifying an attack.