In recent times, it is critical to give abnormal state security to guarantee protected and successful correspondence of data through the web. Nonetheless, secured information correspondence over the internet or some other system is dependably a tested undertaking because of the risk of interruptions and assaults. Along these lines, intrusion detection systems (IDS) have turned into a key segment in system security. Previously, different methodologies were used for creating interruption in location frameworks. In any case, sadly, any of these frameworks are not totally faultless because of the vulnerability of system activity made by ordinary clients and assailants. Henceforth, the requirement for the advancement of productive IDS has expanded consistently. This work proposes a versatile IDS taking into account fuzzy rough sets for characteristic determination. Also, another fluffy unpleasant set-based nearest neighbourhood grouping is proposed for powerful arrangement of the KDD container dataset. This model uses a biased dataset that has 50:50 normal and attack information rather than the ordinary datasets that have 80:20 normal and attack information. The effectiveness of the proposed IDS is upgraded because of the utilisation of one-sided information. The blend of highlight determination and characterisation utilising biased information set diminishes the false alert rate and builds the identification precision. © 2019 Inderscience Enterprises Ltd.