Machine learning comprises computer systems developing new knowledge based on its past experiences. In the Statistics literature this is sometimes called "learning from data" and includes classification and clustering algorithms. Data mining uses machine learning and other techniques to discover previously unknown information in data "warehouses". We are developing new learning algorithms, mainly of the elegant support vector machine type, for dealing with exploding sample sizes, enhancing interpretability, and taking advantage of lower dimensional structure. We are also developing feature selection algorithms based on independent component analysis methods. While our methodology is generic and can potentially be used in several areas of application, it has been motivated by problems in medicine, immunology and marine ecology.