ARG USP

Autotuning Research Group at the University of São Paulo

View the Project on GitHub phrb/arg-usp

(Closed) Applications of Program Autotuning: Call for Proposals for Funding for Undergraduates, Masters, and PhD Students

We are glad to announce a call for masters and PhD scholarship proposals for a research project, done in the context of a collaboration between the University of São Paulo and the Hewlett Packard Enterprise company, starting on February 2020.

The work to be performed by candidates will require 10 hours per week, and will involve the application of techniques for statistical optimization and design of experiments to a problem presented by the candidate. We are looking for students interested in optimizing their applications, or applications used in their work, targeting different metrics, such as performance, memory or network usage.

Ideal problems for this project include problems where the search spaces involved are too large to be explored exhaustively, and have no trivial configuration. Examples of problems include the selection of hyperparameters of Machine Learning algorithms to increase prediction accuracy, selecting compiler flags to improve performance, and configuring user-developed applications to improve user-defined metrics.

We would like students who have already stated their research problems and have had some progress towards measuring and evaluating their objectives, but highly motivated students who are just starting are also welcome to apply.

Up to 5 accepted candidates will receive funding, comparable to scholarships from CAPES, payed by IME/USP for eight months, between February and September 2020. Students currently without funding are preferred.

We invite interested students to submit a CV and a cover letter, summarizing their problem and their interest in applying optimization techniques to it, by February 20th 2020. We assume your advisor is aware and consents with your participation in this project. Accepted candidates will be invited to a two-day workshop at IME/USP, where techniques and libraries for statistical optimization and design of experiments will be presented, and candidates will be asked to give short presentations about their work. Final acceptance notification will be sent by February 29th 2020.

Submit your proposal by Feb. 20th 2020 to phrb@ime.usp.br

Project Coordinator: Prof. Dr. Alfredo Goldman

Project Manager: Pedro Bruel