Abstract: In order to reduce the deception rate of small sample databases and improve the attack defense effectiveness of small sample databases,a small sample database adversarial attack defense algorithm based on latent data mining was designed.With the improved Apriori algorithm,accurate strong association rule advantages are obtained through the working process of frequent attribute value sets,and potential data is mined from small sample databases to resist attacks,with the process of finding frequent sets from candidate sets optimized.On this basis,adversarial attacks are detected through association analysis,and the access rate is controlled through credibility scheduling to defend against malicious sessions,achieving defense against small sample database adversarial attacks.The experimental results show that defense algorithms for potential data mining can effectively defend against various types of attacks on small sample databases,reduce the database spoofing rate caused by attacks,and thus ensure the stability of server utilization in small sample databases.

Key words: data mining, association rules, strong association rules, small sample database, attack detection, Apriori algorithm