Knowledge discovery is the task of extracting new information from large databases, such as recurrent patterns or structural cues. In this framework, cluster analysis refers to the sub-domain dealing with partitioning a given data space such that two samples in the same cluster are similar, while those in different ones are not. Clustering algorithms exploit an input similarity measure on the samples, which should be fine-tuned with the data format and the application athand. However, manually dening a suitable similarity measure is a difficult task in case of limited prior knowledge or complex data structures for example.The purpose of this internship is to investigate an approach for automatically building such a measure by taking advantage of the discriminative abilities of state-of-the-art classication techniques. While classication systems usually require a set of samples annotated with their ground-truth classes, recent work has shown it is possible to exploit classiers trained on an articial annotation of the data in order to induce a similarity measure. In this report, after introducing related scientific background, we propose a unified framework, SIC (Similarity by Iterative Classications), which explores the idea of diverting supervised learning for automatic similarity inference. We study several of its theoretical and practical aspects. We also implement and evaluate SIC on three tasks of knowledge discovery on multimedia content. Results show that in most situations the proposed approach indeed benefits from the underlying classier's properties and outperforms usual similarity measures for clustering applications.