The invention relates to a network intrusion detection method and system, a medium and a device. The method comprises the following steps: acquiring sample features of samples in a sample set, calculating the similarity between the samples according to the sample features, and constructing a sample similarity matrix among the samples according to the similarity, wherein the sample set comprises known label samples and unknown label samples; determining a transition probability matrix containing transition probabilities among the samples according to the sample similarity matrix; and endowing the unknown label samples with the label of the known label sample having the maximal transition probability of being transferred to the unknown label samples to obtain a label matrix. According to thenetwork intrusion detection method and system provided by the invention, on the basis of the transition probability matrix, the unknown label information is initialized based on the maximum membership degree principle of the propagation probability, the convergence speed is faster than the random initialization of the unknown label, and meanwhile, the probability of convergence to the global optimal solution is higher.