We discuss here the mean-field theory for a cellular automata model of meta-learning.Meta-learning is the process of combining outcomes of individual learning procedures inorder to determine the final decision with higher accuracy than any single learning method.Our method is constructed from an ensemble of interacting, learning agents that acquireand process incoming information using various types, or different versions, of machinelearning algorithms. The abstract learning space, where all agents are located,is constructed here using a fully connected model that couples all agents withrandom strength values. The cellular automata network simulates the higher levelintegration of information acquired from the independent learning trials. The finalclassification of incoming input data is therefore defined as the stationary state of themeta-learning system using simple majority rule, yet the minority clusters that share theopposite classification outcome can be observed in the system. Therefore, theprobability of selecting a proper class for a given input data, can be estimatedeven without the prior knowledge of its affiliation. The fuzzy logic can be easilyintroduced into the system, even if learning agents are built from simple binaryclassification machine learning algorithms by calculating the percentage of agreeingagents.