Item preknowledge refers to the phenomenon in which some examinees have access to live items before taking a test. It is one of the most common and significant concerns within the testing industry. Thus, various statistical methods have been proposed to detect item preknowledge in computerized linear or adaptive testing. However, the success of those approaches has been less than desirable.Machine learning (ML) techniques have drawn considerable attention across diverse scientific fields. Studies from various industries have used ML to detect anomalies. As responses with preknowledge can be treated as anomalies, the successful applications of ML in anomaly detection in other contexts suggest that preknowledge detection might be improved with ML techniques.Since only a few machine-learning studies can be found in the field of item preknowledge detection, the purpose of this dissertation is to develop ML algorithms to detect item preknowledge in high-stakes testing. This dissertation proposed (1) an unsupervised approach based on deep clustering to detect preknowledge in linear testing; (2) a semi-supervised approach using a support vector machine to detect preknowledge in computerized adaptive testing; (3) a confidence score capitalizing on an autoencoder model to measure the confidence that the detection result is indeed real, and not merely the most unusual among a set of entirely null data. Throughout simulation studies, results show that under the conditions typically observed in practice, the proposed item preknowledge detection algorithms perform well to identify the examinees with preknowledge and compromised items. It is also shown that the confidence score can provide helpful information for practitioners. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com.bibliotheek.ehb.be/en-US/products/dissertations/individuals.shtml.]