About
7
Publications
2,886
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
119
Citations
Citations since 2017
Introduction
Currently CTO at UpStride (upstride.io). I was a PostDoc at Thales Research & Technology, supported by a Marie Curie Research Fellowship (INFIERI project), where I designed statistical-learning algorithms for High Performance Computing. In 2016, I earned a Ph.D. in Electronics Engineering from the University of Sao Paulo (with year-long research period at the Technical University Munich), where I carried out research on adaptive filtering. My interests include signal processing, artificial intelligence, machine learning, computer vision, and HPC.
Additional affiliations
April 2016 - January 2017
Thales Research and Technology
Position
- Researcher
Description
- Research on algorithms for improving the workload partitioning in heterogeneous high performance computing systems.
April 2014 - May 2015
Position
- Visiting Ph.D. Researcher
Description
- Year-long research stay at the Institute for Media Technology (LMT) of TU Munich. I developed adaptive filters, based on Geometric Algebra, for 3D registration of Point Clouds.
February 2012 - July 2016
Position
- PhD Student
Description
- Research on adaptive filtering theory, one of the building blocks of artificial intelligence. My thesis uses Geometric Algebra (GA) to formulate a new class of adaptive filters (AFs), the so-called Geometric-Algebra Adaptive Filters.
Education
July 2012 - August 2012
Field of study
- Aerospace Engineering - CanSat Leader Training Program (CLTP 3)
February 2012 - July 2016
March 2010 - February 2012
Publications
Publications (7)
This paper presents a systematic study of the scan skewing problem. Scan skewing is the non-rigid deformation of point clouds acquired by LiDAR's and is the result of their sequential scanning nature. We theoretically analyze the impact of skewing on scan matching and subsequently quantify the impact using synthetic LiDAR data with controlled skew...
This document introduces a new class of adaptive filters, namely Geometric-Algebra Adaptive Filters (GAAFs). Those are generated by formulating the underlying minimization problem (a least-squares cost function) from the perspective of Geometric Algebra (GA), a comprehensive mathematical language well-suited for the description of geometric transfo...
In this paper we show that a Geometric Algebra-basedleast-mean-squares adaptive filter (GA-LMS) can be usedto recover the 6-degree-of-freedom alignment of two pointclouds related by a set of point correspondences. We presenta series of techniques that endow the GA-LMS with out-lier (false correspondence) resilience to outperform stan-dard least squ...
This paper exploits Geometric (Clifford) Algebra (GA) theory in order to devise and introduce a new adaptive filtering strategy. From a least-squares cost function, the gradient is calculated following results from Geometric Calculus (GC), the extension of GA to handle differential and integral calculus. The novel GA least-mean-squares (GA-LMS) ada...
The incremental combination of adaptive filters (AFs), recently introduced in the literature, presents intrinsic features capable of improving the overall filtering performance. In this work, the incremental combination is extended to account for AFs with different adaptive rules; when Recursive Least-Squares (RLS) and the Least-Mean-Squares (LMS)...
In parallel combinations of adaptive filters, the component filters are usually run independently to be later on combined, leading to a stagnation phase before reaching a lower error. Conditional transfers of coefficients between the filters have been introduced in an attempt to address this issue. The present work proposes a more natural way of ac...
A new topology for combination of adaptive filters is proposed. Based on incremental strategies, the standard convexly combined parallel-independent filters are rearranged into a series-cooperative configuration without changing the computational complexity. Two new algorithms are derived from the new topology. Simulations in a stationary system id...
