Rafael Oliveira

Rafael Oliveira
The University of Sydney · School of Computer Science

Doctor of Philosophy

About

28
Publications
1,628
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
94
Citations
Introduction
Rafael Oliveira is a research fellow at the School of Computer Science, the University of Sydney, Australia. His current research focuses on probabilistic methods for optimisation and inference with applications in robotics, health and geoscience.
Additional affiliations
October 2018 - present
The University of Sydney
Position
  • Research Associate
March 2015 - June 2018
The University of Sydney
Position
  • Tutor in Machine Learning and Data Mining (COMP5318)
Education
October 2014 - June 2019
The University of Sydney
Field of study
  • Engineering and Information Technologies
March 2009 - March 2014
Federal University of Rio de Janeiro
Field of study
  • Electronics and Computer Engineering

Publications

Publications (28)
Preprint
Full-text available
Bayesian optimisation (BO) algorithms have shown remarkable success in applications involving expensive black-box functions. Traditionally BO has been set as a sequential decision-making process which estimates the utility of query points via an acquisition function and a prior over functions, such as a Gaussian process. Recently, however, a reform...
Preprint
Full-text available
Stochastic model predictive control has been a successful and robust control framework for many robotics tasks where the system dynamics model is slightly inaccurate or in the presence of environment disturbances. Despite the successes, it is still unclear how to best adjust control parameters to the current task in the presence of model parameter...
Preprint
Full-text available
We propose an adaptive optimisation approach for tuning stochastic model predictive control (MPC) hyper-parameters while jointly estimating probability distributions of the transition model parameters based on performance rewards. In particular, we develop a Bayesian optimisation (BO) algorithm with a heteroscedastic noise model to deal with varyin...
Preprint
Full-text available
Bayesian optimisation (BO) has been a successful approach to optimise expensive functions whose prior knowledge can be specified by means of a probabilistic model. Due to their expressiveness and tractable closed-form predictive distributions, Gaussian process (GP) surrogate models have been the default go-to choice when deriving BO frameworks. How...
Article
Full-text available
Bayesian optimization (BO) has been a successful approach to optimize expensive functions whose prior knowledge can be specified by means of a probabilistic model. Due to their expressiveness and tractable closed-form predictive distributions, Gaussian process (GP) surrogate models have been the default go-to choice when deriving BO frameworks. How...
Poster
Full-text available
We derive a Bayesian optimisation algorithm for approximate inference problems with black-box likelihoods. Theoretical results provide approximation bounds and asymptotic convergence guarantees in terms of the Kullback-Leibler divergence. Experiments demonstrate the method in practice. Problem formulation Our goal is to estimate a posterior distrib...
Preprint
Full-text available
Model predictive control (MPC) schemes have a proven track record for delivering aggressive and robust performance in many challenging control tasks, coping with nonlinear system dynamics, constraints, and observational noise. Despite their success, these methods often rely on simple control distributions, which can limit their performance in highl...
Article
Model predictive control (MPC) has been successful in applications involving the control of complex physical systems. This class of controllers leverages the information provided by an approximate model of the system's dynamics to simulate the effect of control actions. MPC methods also present a few hyper-parameters which may require a relatively...
Preprint
Full-text available
We consider the regret minimisation problem in reinforcement learning (RL) in the episodic setting. In many real-world RL environments, the state and action spaces are continuous or very large. Existing approaches establish regret guarantees by either a low-dimensional representation of the probability transition model or a functional approximation...
Preprint
Full-text available
We establish a general form of explicit, input-dependent, measure-valued warpings for learning nonstationary kernels. While stationary kernels are ubiquitous and simple to use, they struggle to adapt to functions that vary in smoothness with respect to the input. The proposed learning algorithm warps inputs as conditional Gaussian measures that con...
Preprint
Full-text available
Model predictive control (MPC) has been successful in applications involving the control of complex physical systems. This class of controllers leverages the information provided by an approximate model of the system's dynamics to simulate the effect of control actions. MPC methods also present a few hyper-parameters which may require a relatively...
Conference Paper
Full-text available
We consider the problem of sequentially optimising the conditional expectation of an objective function, with both the conditional distribution and the objective function assumed to be fixed but unknown. Assuming that the objective function belongs to a reproducing kernel Hilbert space (RKHS), we provide a novel upper confidence bound (UCB) based a...
Preprint
Full-text available
Accurate simulation of complex physical systems enables the development, testing, and certification of control strategies before they are deployed into the real systems. As simulators become more advanced, the analytical tractability of the differential equations and associated numerical solvers incorporated in the simulations diminishes, making th...
Chapter
In outdoor environments, mobile robots are required to navigate through terrain with varying characteristics, some of which might significantly affect the integrity of the platform. Ideally, the robot should be able to identify areas that are safe for navigation based on its own percepts about the environment while avoiding damage to itself. Bayesi...
Conference Paper
Full-text available
Inverse problems are ubiquitous in natural sciences and refer to the challenging task of inferring complex and potentially multi-modal posterior distributions over hidden parameters given a set of observations. Typically, a model of the physical process in the form of differential equations is available but leads to intractable inference over its p...
Conference Paper
Full-text available
Bayesian optimisation (BO) has been a successful approach to optimise functions which are expensive to evaluate and whose observations are noisy. Classical BO algorithms, however, do not account for errors about the location where observations are taken, which is a common issue in problems with physical components. In these cases, the estimation of...
Preprint
Full-text available
Bayesian optimisation (BO) has been a successful approach to optimise functions which are expensive to evaluate and whose observations are noisy. Classical BO algorithms, however, do not account for errors about the location where observations are taken, which is a common issue in problems with physical components. In these cases, the estimation of...
Presentation
Full-text available
Bayesian optimisation (BO) algorithms have been successfully applied to a wide range of problems where the objective function is expensive to evaluate and the observed values are corrupted by noise. In areas such as robotics, however, it is common to find problems where the querying of the objective function is noisy as well, for example due to loc...
Conference Paper
Full-text available
In this paper, we propose a Bayesian optimi-sation (BO) method to actively learn a model of a robot's power consumption and use it to find energy-efficient paths between two fixed locations over an uneven terrain. Most of the prior work in this area relies on models of the vehicle's dynamics or on accurate information about the terrain and its phys...
Conference Paper
In the automation of many kinds of processes, the observable outcome can often be described as the combined effect of an entire sequence of actions, or controls, applied throughout its execution. In these cases, strategies to optimise control policies for individual stages of the process might not be applicable, and instead the whole policy might h...
Article
Full-text available
In the automation of many kinds of processes, the observable outcome can often be described as the combined effect of an entire sequence of actions, or controls, applied throughout its execution. In these cases, strategies to optimise control policies for individual stages of the process might not be applicable, and instead the whole policy might h...
Article
Full-text available
In outdoor environments, mobile robots are required to navigate through terrain with varying characteristics, some of which might significantly affect the integrity of the platform. Ideally, the robot should be able to identify areas that are safe for navigation based on its own percepts about the environment while avoiding damage to itself. Bayesi...

Network

Cited By

Projects

Projects (3)
Project
This study aims to develop probabilistic models of functional trajectories among young people presenting for youth mental health care. Specifically, this work will combine demographic and clinical information to predict functional trajectories and determine the most likely trajectory for an individual. This will inform the development of a prognostic decision support tool that could be used to predict future functional outcomes and used by clinicians to make personalised care allocation decisions.
Project
To develop Bayesian optimisation techniques to optimise control policies in robotics problems with black-box cost functions.
Project
We investigate optimisation problems where queries to an objective function are affected by noise. Our goal is to design data-efficent Bayesian optimisation techniques to solve these problems and to develop a theoretical framwork to analyse optimisation algorithms in these settings.