This study introduces an adaptive dropout artificial neural network-based approach, ADAN2_VN, for the detection of VN attacks in RPL-based IoT environments. The proposed framework is structured into four phases: (1) extraction of novel features using statistical analysis of IoT traffic data; (2) data preprocessing encompassing cleansing, balancing, and normalizati on; (3) ensemble feature selection to isolate the most significant attributes; and (4) implementation of an adaptive dropout strategy within an artificial neural network to enhance detection performance.