In this talk, we are going to see the basics of kernels methods. After a brief presentation of a very simple kernel classifier, we'll give the definition of a postive definite kernel and explain Support vector machine learning. Then, a few kernels for structured data, namely sequences and graphs, will be described. The representer theorem is presented, which explains the rationale for the usual kernel expansion encountered when working with kernel methods. Finally, a few elements from statistical learning theory are given.