With the implementation of the dual carbon policy, user-side energy management is an important carbon reduction initiative. Load identification is an important customer-side energy management method, and after obtaining the voltage and current information of the customer side, the hidden information is mined by machine learning and other methods. However, the sample entries on the customer side are small, and it is difficult to train a conventional network with better results. Therefore, this chapter proposes a small-sample load identification method that combines the squeeze-and-excitation (SE) attention mechanism. It first constructs colored V-I trajectory maps based on voltage and current signals of conventional users and then constructs a neural network model for trajectory map classification. The SE attention mechanism is added to the network so as to give more attention to the channels containing more information, thus improving the classification effect. Finally, the effectiveness of the proposed method is verified by classifying appliances using the WHITED dataset. The proposed method plays an important role in small sample load recognition information mining.