International Workshop on Optimization and Learning (OLA 2019)
Co-Organized by El-Ghazali Talbi
Bangkok, 29–31 Feb 2019
The workshop OLA 2019 will provide an opportunity to the international research community in optimization and learning to discuss recent research results and to develop new ideas and collaborations in a friendly and relaxed atmosphere. OLA 2019 welcomes presentations that cover any aspects of optimization and learning research such as new high-impact applications, parameter tuning, 4th industrial revolution, new research challenges, hybridization issues, optimization-simulation, meta-modeling, high-performance and exascale computing, surrogate modeling, multi-objective optimization, optimization for machine learning, machine learning for optimization, optimization and learning under uncertainty.
Workshop on Synergy between Parallel Computing, Optimization and Simulation (PaCOS 2018)
Organized by Nouredine Melab, Imen Chakroun, Jan Gmys, and Peter Korošec
International Conference on High Performance Computing & Simulation (HPCS 2018)
Orleans, 16–20 July 2018
Most real-world problems are complex and stochastic and, thus, hard to approach. Such problems originate from a wide range of areas including manufacturing and production, logistics and supply chain management, healthcare, and many more. Simulation and optimization have traditionally been considered separately as alternative approaches to deal with such problems. The recent advances in computational power have promoted the proliferation of hybrid techniques that combine both approaches. The challenge is to design efficient and effective hybridization mechanisms taking advantage of great details provided by simulation as well as the ability of optimization methods to provide (near-)optimal solutions.
Simulation-optimization (SO) approaches include, but are not limited to:
- statistical selection methods such as ranking and selection, and multiple comparison procedures;
- black-box search methods that directly make use of the simulation estimates of the objective function like random search algorithms, simulation-based single- and multi-objective metaheuristics (e.g. local search or evolutionary algorithms) (also called simheuristics), probability distribution model–based methods;
- meta-model based methods such as first- and second-order regression models and neural networks;
- gradient-based methods like stochastic approximation, etc.
On the other hand, SO applications can/might become increasingly large (parameter, variable, and objective spaces) and complex (cross-disciplinary and mixed discrete-continuous optimization), requiring the use of parallel computing techniques for efficient and effective execution. However, the design and implementation of parallel SO methods raise several issues such as problem decomposition, parallel sampling, synchronization between simulation and optimization processes, load balancing, scalability, etc.
This workshop provided an opportunity for researchers to present their original contributions on the joint use of advanced single- and multi-objective optimization methods, simulation and distributed and/or parallel multi/many- core computing, and any related issues.
Topics related to synergy between Simulation-Optimization and parallel computing include, but not limited to:
- Parallelization techniques and advanced data structures for SO methods.
- Parallel mechanisms for the hybridization of optimization and simulation.
- Implementation issues of parallel SO on multi-core processors, accelerators, clusters, grids/clouds, etc.
- Energy- and thermal-aware implementation of parallel SO methods.
- Software frameworks for the design and implementation of parallel and/or distributed SO techniques.
- SO applications including healthcare, manufacturing, logistics, biological applications, advanced big data analytics, engineering design, etc.
- Computational/theoretical studies reporting results on solving complex problems using SO techniques.
Game-Benchmark for Evolutionary Algorithms
Organized by Vanessa Volz, Boris Naujoks, Tea Tušar, and Pascal Kerschke
Genetic and Evolutionary Computation Conference (GECCO 2018)
Kyoto, 15–19 July 2018
Games are a very interesting topic that motivates a lot of research.
Key features of games are controllability, safety and repeatability, but also the ability to simulate properties of real-world problems such as measurement noise, uncertainty and the existence of multiple objectives. They have therefore been repeatedly suggested as testbeds for AI algorithms. However, until now, there has not been any concerted effort to implement such a benchmark.
The proposed workshop is intended to fill this gap by (1) motivating and coordinating the development of game-based problems for EAs and (2) encouraging a discussion about what type of problems and function properties are of interest. As a result of the workshop, we aim to obtain a first game-based testsuite for the COCO (COmparing Continuous Optimisers) platform.
If you have been working on a game-related problem that could potentially be solved with evolutionary algorithms, submit it to our workshop. The more diverse problems we have, the better will the resulting benchmark be. Plus, you'll receive solutions to your problem from state-of-the-art optimisation algorithms.
Workshop on Real-Parameter Black-Box Optimization Benchmarking (BBOB 2018)
Organized by Anne Auger, Julien Bect, CentraleSupélec, Dimo Brockhoff, Nikolaus Hansen, Rodolphe Le Riche, Victor Picheny, and Tea Tušar
Genetic and Evolutionary Computation Conference (GECCO 2018)
Kyoto, 15–19 July 2018
The Black-Box-Optimization Benchmarking (BBOB) methodology associated to the BBOB GECCO workshops has become a well-established standard for benchmarking stochastic and deterministic continuous optimization algorithms in recent years https://github.com/numbbo/coco. So far, the BBOB GECCO workshops have covered benchmarking of blackbox optimization algorithms for single- and bi-objective, unconstrained problems in exact and noisy, as well as expensive and non-expensive scenarios. A substantial portion of the success can be attributed to the Comparing Continuous Optimization benchmarking platform (COCO) that builds the basis for all BBOB GECCO workshops and that automatically allows algorithms to be benchmarked and performance data to be visualized effortlessly.
Like for the previous editions of the workshop, we provided source code in various languages (C/C++, Matlab/Octave, Java, and Python) to benchmark algorithms on three different test suites (single-objective with and without noise a well as a noiseless bi-objective suite). Postprocessing data and comparing algorithm performance was equally automatized with COCO (up to already prepared LaTeX templates for writing papers). As a new feature for the 2018 edition, we provided significantly easier access to the already benchmarked data sets such that the analysis of already available COCO data became simple(r).
Analyzing the vast amount of available benchmarking data (from 200+ experiments collected throughout the years) was therefore a special focus of BBOB-2018. Given that the field of (multiobjective) Bayesian optimization received renewed interest in the recent past, we would also like to re-focus our efforts towards benchmarking algorithms for expensive problems (aka surrogate-assisted algorithms developed for limited budgets). Moreover, several classical multiobjective optimization algorithms have not yet been benchmarked on the bbob-biobj test suite, provided since 2016, such that we encouraged contributions on these three following topics in particular:
- expensive/Bayesian/surrogate-assisted optimization
- multiobjective optimization
- analysis of existing benchmarking data
Interested participants of the workshop were invited to submit a paper (not limited to the above topics) which might or might not use the provided LaTeX templates to visualize the performance of unconstrained single- or multiobjective black-box optimization algorithms of their choice on any of the provided testbeds. We encouraged particularly submissions about algorithms from outside the evolutionary computation community as well as any papers related to topics around optimization algorithm benchmarking.
International Workshop on Optimization and Learning (OLA 2018)
Organized by El-Ghazali Talbi and Peter Korošec
Alicante, 26–28 Feb 2018
The workshop OLA 2018 will provide an opportunity to the international research community in optimization and learning to discuss recent research results and to develop new ideas and collaborations in a friendly and relaxed atmosphere. OLA 2018 welcomes presentations that cover any aspects of optimization and learning research such as new high-impact applications, parameter tuning, 4th industrial revolution, new research challenges, hybridization issues, optimization-simulation, meta-modeling, high-performance and exascale computing, surrogate modeling, multi-objective optimization, optimization for machine learning, machine learning for optimization.
Student Workshop
Organized by Vanessa Volz and Boris Naujoks
Genetic and Evolutionary Computation Conference (GECCO 2017)
Berlin, 15–19 July 2017
At the workshop, students receive valuable feedback on the quality of their work and their presentation style. This is assured by constructive discussions after each talk led by a mentor panel of established researchers. Students were encouraged to use this opportunity for guidance on future research directions.
In addition, we provide opportunities to:
- present the work to the whole conference audience at the poster session,
- receive a best paper award, and,
- network and discuss ideas during and after the workshop.
Variety of covered topics (all topics at GECCO) caught the attention of a wide range of conference attendees, who learned about fresh research ideas and meet young researchers with related interests. Even when not submitting or presenting, students were encouraged to attend the workshop to learn of and from the work of their colleagues and broaden their (scientific) horizons.
Women@GECCO Workshop
Organized by Amarda Shehu and Tea Tušar
Genetic and Evolutionary Computation Conference (GECCO 2017)
Berlin, 15–19 July 2017
The workshop series started in 2013 as a venue in which successful women researchers welcome and support other women in evolutionary computation (EC). Over the years, the workshop organically became a venue where students and junior researchers from different under-represented cohorts in EC interacted in an informal setting with more established women researchers on various issues related to fostering and balancing one’s professional and social life, as well as on inserting oneself in the EC community. To acknowledge the growing body of EC researchers and the need for newcomers to integrate themselves in the community, as well as glean effective ways to support growth from informative experiences of other researchers, the 2017 Women@GECCO workshop expanded its focus to “GECCO women welcome EC newcomers”.
Workshop on Real-Parameter Black-Box Optimization Benchmarking (BBOB 2017)
Organized by Anne Auger, Dimo Brockhoff, Nikolaus Hansen, Dejan Tušar, and Tea Tušar
Genetic and Evolutionary Computation Conference (GECCO 2017)
Berlin, 15–19 July 2017
Quantifying and comparing the performance of optimization algorithms is a difficult and tedious task to achieve---but ubiquitous when designing and applying numerical optimization algorithms.
The Black-Box-Optimization Benchmarking (BBOB) methodology associated to the BBOB-GECCO workshops has become a well-established standard for benchmarking stochastic and deterministic continuous optimization algorithms in recent years. A substantial portion of its success can be attributed to the Comparing Continuous Optimization benchmarking platform (COCO) that automatically allows algorithms to be benchmarked and performance data to be visualized effortlessly.
Workshop on Visualisation in Genetic and Evolutionary Computation (VizGEC 2017)
Organized by David Walker, Richard Everson, Jonathan Fieldsend, Bogdan Filipič, and Tea Tušar
Genetic and Evolutionary Computation Conference (GECCO 2017)
Berlin, 15–19 July 2017
Building on workshops held annually since 2010, the eighth annual workshop on Visualisation Methods in Genetic and Evolutionary Computation (VizGEC), to be held at GECCO 2017 in Berlin, is intended to explore, evaluate and promote current visualisation developments in the area of genetic and evolutionary computation (GEC). Visualisation is a crucial tool in this area, providing vital insight and understanding into algorithm operation and problem landscapes as well as enabling the use of GEC methods on data mining tasks. Particular topics of interest are:
- visualisation of the evolution of a synthetic genetic population
- visualisation of algorithm operation
- visualisation of problem landscapes
- visualisation of multi-objective trade-off surfaces
- the use of genetic and evolutionary techniques for visualising data
- novel technologies for visualisation within genetic and evolutionary computation
- visual steering of algorithms
- visualisation in real-world applications
As well as allowing us to observe how individuals interact, visualising the evolution of a synthetic genetic population over time facilitates the analysis of how individuals change during evolution, allowing the observation of undesirable traits such as premature convergence and stagnation within the population. In addition to this, by visualising the problem landscape we can explore the distribution of solutions generated with a GEC method to ensure that the landscape has been fully explored. In the case of multi- and many-objective optimisation problems this is enhanced by the visualisation of the trade-off between objectives, a non-trivial task for problems comprising four or more objectives, where it is necessary to provide an intuitive visualisation of the Pareto front to a decision maker. All of these areas are drawn together in the field of interactive evolutionary computation, where decision makers need to be provided with as much information as possible since they are required to interact with the GEC method in an efficient manner, in order to generate and understand good solutions quickly.
In addition to visualising the solutions generated by a GEC process, we can also visualise the processes themselves. It can be useful, for example, to investigate which evolutionary operators are most commonly applied by an algorithm, as well as how they are applied, in order to gain an understanding of how the process can be most effectively tuned to solve the problem at hand. Advances in animation and the prevalence of digital display, rather than relying on the paper-based presentation of a visualisation, mean that it is possible to use visualisation methods so that aspects of an algorithm's performance can be evaluated online.
GEC methods have also recently been applied to the visualisation of data. As the amount of data available in areas such as bioinformatics increases rapidly, it is necessary to develop methods that can visualise large quantities of data; evolutionary methods can, and have, been used for this. Work on visualising the results of evolutionary data mining is also now appearing.
All of these methods benefit greatly from developments in high-powered graphics cards and work on 3D visualisation, largely driven by the computer games community. A workshop provides a good environment for the demonstration of such methods. As well as presenting the results of a GEC process in a traditional visual way, we are also keen to solicit work on other forms of presentation.
International Workshop on Parallel Optimization using/for Multi and Many-Core High Performance Computing (POMCO 2016)
Organized by Albert Y. Zomaya, Nouredine Melab, and Imen Chakroun
International Conference on High Performance Computing & Simulation (HPCS 2016)
Innsbruck, 18–22 July 2016
On the road to exascale, multi-core processors and many-core accelerators/coprocessors are increasingly becoming key-building blocks of many computing platforms including laptops, high performance workstations, clusters, grids, and clouds. On the other hand, plenty of hard problems in a wide range of areas including engineering design, telecommunications, logistics and transportation, biology, energy, etc., are often modeled and tackled using optimization approaches. These approaches include greedy algorithms, exact methods (dynamic programming, Branch-and-X, constraint programming, A*, etc.) and meta-heuristics (evolutionary algorithms, particle swarm, ant or bee colonies, simulated annealing, Tabu search, etc.). In many research works, optimization techniques are used to address high performance computing (HPC) issues including HPC hardware design, compiling, scheduling, auto-tuning, etc. On the other hand, optimization problems become increasingly large and complex, forcing the use of parallel computing for their efficient and effective resolution. The design and implementation of parallel optimization methods raise several issues such as load balancing, data locality and placement, fault tolerance, scalability, thread divergence, etc.
This workshop seeks to provide an opportunity for the researchers to present their original contributions on the joint use of advanced (discrete or continuous, single or multi-objective, static or dynamic, deterministic or stochastic, hybrid) optimization methods and distributed and/or parallel multi/many-core computing, and any related issues.
The POMCO Workshop topics include (but are not limited to) the following:
- Parallel models (island, master-worker, multi-start, etc.) for optimization methods revisited for multi-core and/or many-core (MMC) environments.
- Parallelization techniques and advanced data structures for exact (e.g. tree-based) optimization methods.
- Parallel mechanisms for hybridization of optimization algorithms on MMC environments
- Parallel strategies for handling uncertainty, robustness and dynamic nature of optimization methods.
- Implementation issues of parallel optimization methods on MMC workstations, MMC clusters, MMC grids/clouds, etc.
- Software frameworks for the design and implementation of parallel and/or distributed MMC optimization algorithms
- Computational/theoretical studies reporting results on solving challenging problems using MMC computing
- Energy-aware optimization for/with MMC parallel and/or distributed optimization methods
- Optimization techniques for efficient compiling, scheduling, etc. for MMC environments
- Optimization techniques for scheduling, compiling, auto-tuning for MMC clusters, MMC grids/clouds, etc.
Workshop on Nature Inspired Distributed Computing (IEEE NIDISC 2016)
Organized by Pascal Bouvry, Franciszek Seredynski, El-Ghazali Talbi, and Grégoire Danoy
The 30th IEEE/ACM International Parallel and Distributed Processing (IPDPS 2016)
Chicago, 23–27 May 2016
Techniques based on metaheuristics and nature-inspired paradigms can provide efficient solutions to a wide variety of problems. Moreover, parallel and distributed metaheuristics can be used to provide more powerful problem solving environments in a variety of fields, ranging, for example, from finance to bio- and health-informatics.
This workshop seeks to provide an opportunity for researchers to explore the connection between metaheuristics and the development of solutions to problems that arise in operations research, parallel computing, telecommunications, and many others. Topics of interest include, but are not limited to:
- Nature-inspired methods (e.g. ant colonies, GAs, cellular automata, DNA and molecular computing, local search, etc) for problem solving environments.
- Parallel and distributed metaheuristics techniques (algorithms, technologies and tools).
- Applications combining traditional parallel and distributed computing and optimization techniques as well as theoretical issues (convergence, complexity, etc).
- Other algorithms and applications relating the above mentioned research areas.