Carlos R. B. Azevedo

Postdoctoral Research Fellow at Universidade Estadual de Campinas

Location
Campinas Area, Brazil
Industry
Research
Current
  1. Universidade Estadual de Campinas,
  2. IEEE Computational Intelligence Society
Previous
  1. Laboratory of Bioinformatics and Bio-inspired Computing (LBiC),
  2. IEEE Computational Intelligence Society Student Chapter Unicamp,
  3. IEEE Computational Intelligence Society
Education
  1. Universidade Estadual de Campinas
Recommendations3 people have recommended Carlos R. B.
Websites
500+connections

Join LinkedIn and access Carlos R. B.'s full profile. It's free!

As a LinkedIn member, you'll join 300 million other professionals who are sharing connections, ideas, and opportunities.

  • See who you know in common
  • Get introduced
  • Contact Carlos R. B. directly

Background

Summary

Experienced Computational Intelligence/Machine Learning specialist in multi-objective optimization and decision-making. I'm constantly looking for opportunities to apply my research expertise for solving grand challenges and making the world we live a smarter and better place.

The research I've developed has focused on developing efficient neural and evolutionary-inspired computation algorithms for a range of relevant problems such as financial portfolio optimization, time series forecasting, inventory routing, data clustering and classification and data compression.

My long-term goals are to advance scientific research among developing countries, to engage students into research, and to foster innovation by strengthening the ties between the industry and the academy.

Besides, I'm deeply interested on the future of scientific research, particularly on the developing countries, as well on poverty alleviation and technological sustentability issues.

Skills

  • Computational...
  • Evolutionary Algorithms
  • Statistical Learning
  • Stochastic Optimization
  • Data Classification
  • Portfolio Optimization
  • Neural Networks
  • Information Theory
  • Cluster Analysis
  • Soft Computing
  • SVM
  • Bayesian networks
  • ANOVA
  • Pattern Matching
  • Matlab
  • C/C++ STL
  • Combinatorial...
  • Game Theory
  • Machine Learning
  • Fuzzy Logic
  • Operations Research
  • Artificial Intelligence
  • Computer Science
  • Time Series Analysis
  • Optimization
  • Multi-objective...
  • Optimizations
  • Algorithms
  • Science
  • Text Mining
  • Data Mining
  • Bioinformatics
  • Recommender Systems
  • Pattern Recognition
  • Evolutionary Computation
  • Python
  • Natural Language...
  • LaTeX
  • Software Development
  • Computer Vision
  • Mathematical Modeling
  • Image Processing
  • Signal Processing
  • Distributed Systems
  • Algorithm Design
  • R
  • Applied Mathematics
  • See 32+  See less

Experience

Postdoctoral Research Fellow

Universidade Estadual de Campinas
– Present (1 month)Campinas Area, Brazil

Title: Convex reductions for online multi-objective optimization via Hypervolume maximization
CNPq postocdoral research program grant.

Vice-Chair of the Students Activities Subcommitee

IEEE Computational Intelligence Society
– Present (1 year 8 months)

The CIS Student Activities Subcommittee strives to promote the student participation in various CIS activities. It has the following major tasks: (i) administer student travel grants for attending CIS-sponsored conferences; (ii) organize student activities in CIS-sponsored conferences; and (iii) cooperate in organizing activities in CIS Student Chapters.

Member of Webnars Subcommittee

IEEE Computational Intelligence Society
– Present (8 months)

The goal of Webinars Subcommittee (WS) is to organize webinars to the CIS members on topics related to computational intelligence, especially the state-of-the-arts.

Doctoral Researcher

Laboratory of Bioinformatics and Bio-inspired Computing (LBiC)
(3 years 7 months)Campinas Area, Brazil

Thesis: Anticipation in Multiple Criteria Decision-Making Under Uncertainty.

Research Proposal: Anticipatory optimization for decision making systems subject to multiple and time-varying criteria (FAPESP Scholarship).

Chair

IEEE Computational Intelligence Society Student Chapter Unicamp
(1 year 1 month)Campinas Area, Brazil

I was responsible for coordinating the efforts to bring together the CI research community at the University of Campinas, engaging students and researchers toward pro-active learning and research environments under the support of a world-class network of leading associate members of IEEE CIS.

Member of the Students Activities Subcommitee

IEEE Computational Intelligence Society
(1 year 1 month)

The CIS Student Activities Subcommittee strives to promote the student participation in various CIS activities. It has the following major tasks: (i) administer student travel grants for attending CIS-sponsored conferences; (ii) organize student activities in CIS-sponsored conferences; and (iii) cooperate in organizing activities in CIS Student Chapters.

Owner, Editor-in-Chief.

vocenofuturo.com
(2 years 3 months)

Owner and editor-in-chief of "youinthefuture", a Brazilian science and emerging tech news portal in portuguese.

Lecturer on Software Development

Brazilian Union of Technological Institutes
(8 months)

+ Courses taught:
Logics for Web Development; Tools for Web Development

Programmer and Instructor on Programming Languages

Especializa Treinamentos
(1 year 7 months)

+ Instructor of PHP Programmer course series;
+ Programmer.

Volunteer Experience & Causes

Opportunities Carlos R. B. is looking for:

  • Joining a nonprofit board

Causes Carlos R. B. cares about:

  • Economic Empowerment
  • Education
  • Poverty Alleviation
  • Science and Technology

Organizations Carlos R. B. supports:

Projects

A methodology for multicriteria stochastic anticipatory optimization(Link)

This research project aims to design new sequential decision-making systems, operating in uncertain environments under multiple conflicting optimization criteria. It is assumed that the dynamics of the system under control (in discrete time and over a finite horizon) is linear and that the exogenous uncertainty can be estimated by parametric probabilistic models. In this context, four challenges are covered, namely: (1) the learning of probabilistic models capable of measuring the influence of the decisions implemented on the future operating costs; (2) the determination of stochastic policies capable of modeling the decision maker; (3) the determination of the risks of violating the problem constraints; and (4) the incorporation of partial preferences in the decision making process. It is emphasized that research activity on the treatment of multiple conflicting criteria and the incorporation of chance-constraints is scarce, considering the literature of anticipatory meta-heuristics and approximate dynamic programming. As its main contribution, this project proposes a new methodology as well as tools to allow for the effective synthesis of anticipatory multicriteria decision-making systems. The methodology will be investigated over a broad class of problems, ranging from vendor managed inventory-routing problems; optimization of financial and product portfolios, and the control of public transport systems operating in real time.

Contextualização semântica para a antecipação de demandas por conteúdo: um experimento social em redes de microblogs(Link)

A proposta deste projeto motiva-se pela oportunidade de aprimorar os processos decisórios de impressão de anúncios, aumentando as chances de engajamento do usuário e promovendo maior retorno para o anunciante. A pesquisa a ser desenvolvida enquadra-se na linha de publicidade contextualizada (do inglês contextual advertising), a qual, tradicionalmente, parte da extração de palavras-chave das páginas Web onde os anúncios serão exibidos, associando-as a palavras-chave vinculadas previamente aos anúncios pelos gerentes de campanha.

Education

Universidade Estadual de Campinas

PhD, Electrical and Computer Engineering

CNPq and FAPESP Scholarships.

Research fields: evolutionary anticipatory systems; natural and biologically-inspired computing with emphasis on dynamic, combinatorial and multiobjective optimization problems; dynamical Bayesian models; reasoning under uncertainty; conflicts resolution; axiomatic decision-theory; preference for flexibility; hypervolume maximization.

Activities and Societies: IEEE Computational Intelligence Society Student Chapter, Association of Electrical Engineering Graduate Students at Unicamp (APOGEEU)

Universidade Federal de Pernambuco

Master, Computer Science

Dissertation titled “Diversity Generation in Dynamic Multiobjective Optimization by Non-Dominance Landscapes”, CNPq MSc Scholarship

Singularity University

Graduate Studies Program 2010, Artificial Intelligence and Robotics, Future Studies and Forecasting, Biotechnology

Recipient of a SU partial scholarship to attend GSP10.

Was activelly engaged on the Space Team Project, more especifically on the subgroup which researched how to augment robotic autonomy for space exploration through Artificial General Intelligence.

Performed an Ignite Presentation on "The Future of Scientific Research". Video available through the following link: http://t.co/iHCGzzX

Activities and Societies: Space Team Project

Universidade Católica de Pernambuco

Bachelor, Computer Science, 88.4

+ Laureate student
+ Active in two scientific initiation programs on Time Series Forecasting with Quantum Neural Networks and Evolutionary Algorithms for Signal Compression
+ CNPq/PIBIC Scholarship (National Council of Technological and Scientific Development, Institutional Scientific Initiation Program)
+ Courses taught (teaching assistant): Stochastic Simulation; Algorithms and Data Structure

Activities and Societies: Institutional Scientific Initiation Program, Undergraduate Teaching Assistant

Courses

Universidade Federal de Pernambuco

  • Machine Learning (IN110)
  • Statistical Analysis and Experiment Design (IN111)
  • Neural Networks (IN0997)
  • Biometrics (IN112)
  • Evolutionary Computation (IN113)
  • Research Project on Computational Intelligence (IN118)

Universidade Estadual de Campinas

  • Information Theory (IE660)
  • Fuzzy Systems (IA861)
  • Evolutionary Computing (IA707)
  • Game Theory (IA891)
  • Scalable Data Mining and Machine Learning for Complex Data (IA900)

Independent Coursework

  • Introduction to Quantum Mechanics
  • Quantum Computation
  • Quantum Information
  • Introduction to RSA Encryption
  • Modern Cryptography Techniques
  • Parallel Algorithms for Sequence Processing
  • Game Theory on Computer Networks
  • Transportation and Road Traffic Intelligent Systems
  • Digital Signal Processing
  • Java2SE 5.0
  • PHP, MySQL, PostgreSQL

Languages

  1. English

    Full professional proficiency
  2. Portuguese

    Native or bilingual proficiency
  3. Spanish

    Elementary proficiency
  4. French

    Elementary proficiency

Certifications

Certificate of Proficiency in English

Universitiy of Michigan
– Present

Publications

Anticipatory Stochastic Multi-Objective Optimization for uncertainty handling in portfolio selection(Link)

IEEE Congress on Evolutionary Computation (CEC)
July 2013

An anticipatory stochastic multi-objective model based on S-Metric maximization is proposed. The environment is assumed to be noisy and time-varying. This raises the question of how to incorporate anticipation in metaheuristics such that the Pareto optimal solutions can reflect the uncertainty about the subsequent environments. A principled anticipatory learning method for tracking the dynamics of the objective vectors is then proposed so that the estimated S-Metric contributions of each solution can integrate the underlying stochastic uncertainty. The proposal is assessed for minimum holding, cardinality constrained portfolio selection, using real-world stock data. Preliminary results suggest that, by taking into account the underlying uncertainty in the predictive knowledge provided by a Kalman filter, we were able to reduce the sum of squared errors prediction of the portfolios ex-post return and risk estimation in out-of-sample investment environments.

Authors:

Regularized hypervolume selection for robust portfolio optimization in dynamic environments(Link)

IEEE Congress on Evolutionary Computation (CEC)
July 2013

This paper proposes a regularized hypervolume (SMetric) selection algorithm. The proposal is used for incorporating stability and diversification in financial portfolios obtained by solving a temporal sequence of multi-objective Mean Variance Problems (MVP) on real-world stock data, for short to longterm rebalancing periods. We also propose the usage of robust statistics for estimating the parameters of the assets returns distribution so that we are able to test two variants (with and without regularization) on dynamic environments under different levels of instability. The results suggest that the maximum attaining Sharpe Ratio portfolios obtained for the original MVP without regularization are unstable, yielding high turnover rates, whereas solving the robust MVP with regularization mitigated turnover, providing more stable solutions for unseen, dynamic environments. Finally, we report an apparent conflict between stability in the objective space and in the decision space.

Authors:

The influence of supervised clustering for RBFNN centers definition: A comparative study(Link)

International Conference on Artificial Neural Networks (ICANN), Lecture Notes in Computer Science
September 2012

Several clustering algorithms have been considered to determine the centers and dispersions of the hidden layer neurons of Radial Basis Function Neural Networks (RBFNNs) when applied both to regression and classification tasks. Most of the proposed approaches use unsupervised clustering techniques. However, for data classification, by performing supervised clustering it is expected that the obtained clusters represent meaningful aspects of the dataset. We therefore compared the original versions of k-means, Neural-Gas (NG) and Adaptive Radius Immune Algorithm (ARIA) along with their variants that use labeled information. The first two had already supervised versions in the literature, and we extended ARIA toward a supervised version. Artificial and real-world datasets were considered in our experiments and the results showed that supervised clustering is better indicated in problems with unbalanced and overlapping classes, and also when the number of input features is high.

Authors:

Money in trees: How memes, trees, and isolation can optimize financial portfolios(Link)

Information Sciences
July 2011

In this paper, we propose the hybrid application of two nature inspired approaches to the problem of Portfolio Optimization. This problem consists of the selection and weighting of financial assets. Its goal is to form an investment strategy which maximizes a return measure and minimizes a risk measure. We perform a series of simulation experiments with historical data in the NASDAQ and S&P500 markets between 2006 and 2008. The results show that adding a terrain strategy to a previously successful Memetic Algorithm promoted niching and speciation of the population, which led to a significant improvement in the performance when compared to previous evolutionary methods. We also show that the use of Memetic Algorithms gives the evolved solutions a degree of adaptability to changes in a dynamic market.

Authors:

Non-Dominance Landscapes for Guiding Diversity Generation in Dynamic Environments(Link)

VIII Best MSc Dissertation/PhD Thesis Contest in Artificial Intelligence (CTDIA)
October 2012

This master thesis reports a novel framework to increase the effectiveness of raising diversity levels when coping with environment changes in Evolutionary Multi-Objective Optimization. The framework employs multivariate order statistics to estimate the probability of a solution being non-dominated when drawn from a given distribution of objective vectors. The manifolds wherein such probabilities vary along the objective space over time are called Non-Dominance Landscapes (NDLs). We then proposed a NDL-guided selection criterion. A case study on 24 scenarios revealed that the proposed framework contributed for improving the quality of the evolved solutions in 79% of those scenarios, when compared with commonly used diversity generation schemes. By intensifying the selective pressure in regions for which higher non-dominance probabilities were estimated and by inserting solutions in the vicinity of the non-dominated set, we obtained higher offine hypervolume values when compared to keeping high diversity without any guidance.

Authors:

Generalized immigration schemes for dynamic evolutionary multiobjective optimization(Link)

2011 IEEE Congress on Evolutionary Computation (CEC)
June 2011

The insertion of atypical solutions (immigrants) in Evolutionary Algorithms populations is a well studied and successful strategy to cope with the difficulties of tracking optima in dynamic environments in single-objective optimization. This paper studies a probabilistic model, suggesting that centroid-based diversity measures can mislead the search towards optima, and presents an extended taxonomy of immigration schemes, from which three immigrants strategies are generalized and integrated into NSGA2 for Dynamic Multiobjective Optimization (DMO). The correlation between two diversity indicators and hypervolume is analyzed in order to assess the influence of the diversity generated by the immigration schemes in the evolution of non-dominated solutions sets on distinct continuous DMO problems under different levels of severity and periodicity of change. Furthermore, the proposed immigration schemes are ranked in terms of the observed offline hypervolume indicator.

Authors:

Correlation between diversity and hypervolume in evolutionary multiobjective optimization(Link)

2011 IEEE Congress on Evolutionary Computation (CEC)
June 2011

This paper reports a study of the influence of diversity in the convergence dynamics of Multiobjective Evolutionary Algorithms (MOEAs) towards the Pareto Front (PF). By varying mutation and crossover parameters, several scenarios of exploration and exploitation are considered, in which each of them is analysed in order to assess the role of diversity levels on the evolution of high quality sets of non-dominated solutions, in terms of the Hypervolume indicator. For this task, the application of the NSGA2 algorithm for approximating the PF in five ZDT benchmark problems is considered. The results not only indicate that there are significant statistical correlations between several diversity metrics and the observed maximum Hypervolume levels on the evolved populations, but also suggest that there are predictable temporal patterns of correlation when the evolutionary process is portrayed generation wise.

Authors:

Adaptive terrain-based memetic algorithms(Link)

Proceedings of the 11th Annual conference on Genetic and evolutionary computation
July 2009

The Terrain-Based Memetic Algorithm (TBMA) is a diffusion MA in which the local search (LS) behavior depends on the topological distribution of memetic material over a grid (terrain). In TBMA, the spreading of meme values -- such as LS step sizes -- emulates cultural differences which often arise in sparse populations. In this paper, adaptive capabilities of TBMAs are investigated by meme diffusion: individuals are allowed to move in the terrain and/or to affect their environment, by either following more effective memes or by transmitting successful meme values to nearby cells. In this regard, four TBMA versions are proposed and evaluated on three image vector quantizer design instances. The TBMAs are compared with K-Means and a Cellular MA. The results strongly indicate that utilizing dynamically adaptive meme evolution produces the best solutions using fewer fitness evaluations for this application.

Authors:

Terrain-Based Memetic Algorithms for Vector Quantizer Design(Link)

Nature-Inspired Cooperative Strategies for Optimization (NICSO 2008)
2009

Recently, a Genetic Accelerated K-Means Algorithm (GAKM) was proposed as an approach for optimizing Vector Quantization (VQ) codebooks, relying on an accelerated version of K-Means algorithm as a new local learning module. This approach requires the determination of a scale factor parameter (η), which affects the local search performed by GAKM. The problem of auto-adapting the local search in GAKM, by adjusting the η parameter, is addressed in this work by the proposal of a Terrain-Based Memetic Algorithm (TBMA), derived from existing spatially distributed evolutionary models. Simulation results regarding image VQ show that this new approach is able to adjust the scale factor (η) for different images at distinct coding rates, leading to better Peak Signal-to-Noise Ratio values for the reconstructed images when compared to both K-Means and Cellular Genetic Algorithm + K-Means. The TBMA also demonstrates capability of tuning the mutation rate throughout the genetic search.

Authors:

Improving Image Vector Quantization with a Genetic Accelerated K-Means Algorithm(Link)

Avanced Concepts for Intelligent Vision Systems
July 2008

In this paper, vector quantizer optimization is accomplished by a hybrid evolutionary method, which consists of a modified genetic algorithm (GA) with a local optimization module given by an accelerated version of the K-means algorithm. Simulation results regarding image compression based on VQ show that the codebooks optimized by the proposed method lead to reconstructed images with higher peak signal-to-noise ratio (PSNR) values and that the proposed method requires fewer GA generations (up to 40%) to achieve the best PSNR results produced by the conventional GA + standard K-means approach. The effect of increasing the number of iterations performed by the local optimization module within the proposed method is discussed.

Authors:

An Evolutionary Approach for Vector Quantization Codebook Optimization(Link)

Advances in Neural Networks - ISNN 2008
2008

This paper proposes a hybrid evolutionary algorithm based on an accelerated version of K-means integrated with a modified genetic algorithm (GA) for vector quantization (VQ) codebook optimization. From simulation results involving image compression based on VQ, it is observed that the proposed method leads to better codebooks when compared with the conventional one (GA + standard K-means), in the sense that the former leads to higher peak signal-to-noise ratio (PSNR) results for the reconstructed images. Additionally, it is observed that the proposed method requires fewer GA generations (up to 40%) to achieve the best PSNR results produced by the conventional method.

Authors:

Time Series Forecasting with Qubit Neural Networks(Link)

Artificial Intelligence and Soft Computing - 2007
October 2007

This paper proposes a quantum learning scheme approach for time series forecasting, through the application of the new non-standard Qubit Neural Network (QNN) model. The QNN description was adapted in this work in order to resemble classical Artificial Neural Networks (ANNs). Three stock market series were predicted. The results are discussed over several statistics and are compared with ANNs experiments with equivalent degrees of freedom.

Authors:

Methods to Accelerate a Competitive Learning Algorithm Applied to VQ Codebook Desing(Link)

Tendências em Matemática Aplicada e Computacional
2010

Codebook design plays a crucial role in the performance of signal pro-cessing systems based on vector quantization (VQ). This paper is concerned with methods for reducing the processing time spent by a competitive learning (CL) algorithm applied to VQ codebook design. Using analytical expressions for the number of operations (multiplications, additions, subtractions and comparisons) performed by the CL algorithm, it is shown that almost all the operations are due to the nearest neighbor search (NNS). Simulation results regarding image VQ show that simple modifications introduced in CL lead to considerable number clock cyclessavings.

Authors:

Additional Info

  1. Interests

    • computer science,
    • innovation,
    • futurism,
    • mathematics,
    • playing tennis,
    • singularity,
    • blogging
  2. Personal Details

  3. Advice for Contacting Carlos R. B.

Organizations

Honors & Awards

October 2012 - 2nd Best Master Thesis on Artificial Intelligence, Special Commission on AI of The Brazilian Computer Society.

June 2011 - IEEE CIS Outstanding Student-Paper Travel Grant for CEC'11, IEEE Computational Inteligence Society.

July 2010 - Awarded a US $8k partial scholarship to attend GSP10, Singularity University.

May 2009 - Awarded a Student Travel Grant by the Association for Computing Machinery (AMC) to attend GECCO'09.

January 2009 - Laureate Student in Computer Science by Catholic University of Pernambuco.

September 2008 - Best 2007/2008 Scientific Initiation Program Student in Exacts and Earth Sciences Award, Catholic University of Pernambuco.

December 2007 - Top 20 Essay on "How to Overcome Poverty and Inequality" Award by UNESCO out of a wide pool of 41,329 Brazilian undergrad applicants.

September 2007 - Second Best 2006/2007 Scientific Initiation Program Student in Exacts and Earth Sciences Award, Catholic University of Pernambuco.

View Carlos R. B.'s full profile to...

  • See who you know in common
  • Get introduced
  • Contact Carlos R. B. directly

Not the Carlos R. B. Azevedo you're looking for? View more

Insights


People Also Viewed

LinkedIn member directory - Browse members by country

  1. a
  2. b
  3. c
  4. d
  5. e
  6. f
  7. g
  8. h
  9. i
  10. j
  11. k
  12. l
  13. m
  14. n
  15. o
  16. p
  17. q
  18. r
  19. s
  20. t
  21. u
  22. v
  23. w
  24. x
  25. y
  26. z
  27. more