Recent technical advancements in molecular, genetic, and optogenetic methods for model animals, as well as whole brain activity recordings, have allowed investigators to elucidate key brain regions regulating learning and memory. Neurons are proposed to work together to encode and regulate memories; therefore, the next crucial step is to develop methods to identify the exact neural network structure underlying memory encoding and investigate how these neural populations process information. Here, we describe an established methodological pipeline for analyzing longitudinal two-photon imaging data from the prefrontal cortex using a regularized regression method and graphical modeling, named elastic net and conditional random field modeling, respectively. This pipeline allowed the detection of learning-dependent changes in neural network topologies and information processing in the specific neural population encoding associative fear memory.