Intrusion Detection System with H2O Deep Learning Approach

With advances in technology, future events can be predicted or categorized based on knowledge gathered in the past so that errors can be minimized. These strategies play an essential role in avoiding attacks on large-scale data network systems in more specialized technological fields, especially computer network systems. IDS (Instruction Detection System) can mitigate large-scale network system attacks by detecting suspicious behavior in network systems. The Deep Learning approach, which uses artificial neural networks, is used to help maintain network systems using IDS. In this study, researchers used the RapidMiner application program to help analyze the data set. Results obtained by using the Deep Learning H2O algorithm method with a dataset of 25192 data training and 42 records produced the highest level of accuracy and a recall value of 99.36% for ratios (number division) of 0.6 and 0.4 for epoch values of 40 activations using the rectifier itself. For wave rectifiers, or what is commonly known as input and output, these results are obtained from the model created as a system used to classify network attacks.