At the avant-garde of scientic and industrial computing, there are problems involving very large-scale streams and collections of complex data. Multimedia, multimodal, spatial-temporal or involving social networks, these data are beyond the models created for handling structured and textual information. Thus, two dimensions appear as immediate challenges: the volume of data and their complexity. At the front end of large-scale and high complexity, it is insu cient to work in isolation with algorithmic solutions or with execution environment improvements. Rather, it is necessary to conceive and synergistically combine those two lines of work. New massively parallel architectures, derived from the GPUs (graphics processing units) and APUs (advanced processing units), promise to disrupt the high-performance computing market. However, due to the large di erence between those architectures and those based on conventional CPUs, the concretization of their potential is far from trivial, involving intricate compromises ranging from algorithm design to online dynamic adjustments made at runtime. Advancing the state of the art at this vanguard is the goal of this ambitious project, which brings together a team with complementary skills: on one hand, in the elds of machine learning, multimedia search and classi cation, and processing of other complex data and streams; on the other hand, in the elds of parallel computing, high performance computing and new computing architectures. The versatility and relevance of the techniques studied will be demonstrated in bold applications of high impact: Information Retrieval, Multimedia Classi cation, and Visualization of Complex Data. The project encompasses data analysis and data mining, which have become a key activity for many organizations.

Integrantes: William Robson Schwartz – Coordenador / Hélio Pedrini – Integrante / Eduardo Alves do Valle Júnior – Integrante / Anderson Rocha – Integrante / Ricardo da Silva Torres – Integrante / Siome Klein Goldenstein – Integrante / Rodolfo Jardim de Azevedo – Integrante / Jacques Wainer – Integrante / George Teodoro – Integrante / Edson Borin – Integrante.