The Visible Light Communication (VLC) is a promising technology to assure the wireless data transmission for optical wireless communication system. Some modulation methods used in VLC systems require to apply a threshold value on the receiver side. In this paper, therefore, it has been investigated the classifier based demodulator architectures to solve the data detecting problem for threshold-based transmission methods such as On-Off Keying (OOK) scheme, under the condition of the unknown threshold level value. The presented receiver schemes use the K-Nearest Neighbors (KNN algorithm to classify the logical level of received OOK signal. These classification method requires a few training data to detect data bits for OOK receiver system. It can be observed from the simulation results that the receiver system is affected from the number of the neighbor employed in the KNN algorithm. Addition to this, Observation of the result clearly shows that the training set size is crucial for the Bit Error Rate (BER) Performance. According to simulation results, the transmission distance can be increased 2.451 m to 2.530 m at the BER of approximately 10-3 by using KNN method at the receiver side.