newsletter of the ACM Special Interest Group on Genetic and Evolutionary Computation

Ingo Rechenberg during his keynote talk at GECCO 2019: Evolution, Robotics and the somersaulting spider.


Welcome to the Winter 2021 issue of the SIGEvolution newsletter! This issue is dedicated to the memory of Ingo Rechenberg, who sadly passed away on September 25th, 2021, after a short serious illness in Berlin. Recognized as the inventor of one of the branches of evolutionary computation, Evolution Strategies, Ingo Rechenberg also played a major role worldwide in establishing and promoting Bionics as a scientific discipline.

Many of us had the wonderful opportunity to attend Ingo’s inspiring keynote talk at GECCO 2019 in Prague, where he shared his passion for evolution and biologically inspired robotics and showed us the biologic and robotic versions of the Moroccan flic-flac spider, Cebrennus rechenbergi, named in his honor, as he first collected specimens in the Moroccan desert. We pay tribute to Ingo by collecting anecdotes from one of his PhD students and other members of our community touched by his work and ideas.

We continue with a lively, visual report on GECCO 2021 going virtual, and conclude the issue by announcing recent events and forthcoming calls for submissions. Please get in touch if you would like to contribute an article for a future issue or have suggestions for the newsletter.

Wishing you a Happy New year and a prosperous 2022!

Gabriela Ochoa, Editor.

Tributes to Ingo Rechenberg (1934 – 2021)

Anne Auger, Nikolaus Hansen, Manuel López-Ibáñez, Günter Rudolph

Anne Auger

Ingo Rechenberg was one of the fathers of Evolution Strategies, my main research topic for almost two decades. I had the chance to meet him in 2005 at the GECCO conference in Washington where he presented a tutorial on “Bionik: Building on Biological Evolution”. I discovered a very passionate person who explained how biology, the flight of birds, and nature shaped and inspired his research. When discussing with him afterward, he told me that the biological motivation of the methods in our field is essential for him. He admitted that he never liked the CMA-ES algorithm (that started to be recognized at that time as a powerful method and was invented in his laboratory by Nikolaus Hansen and Andreas Ostermeier) because it is not bio-inspired enough!

Images of the Tabbot, the bionics rover inspired by the Moroccan flic-flac spider, Cebrennus rechenbergi, image taken from Ingo Rechenberg’s keynote talk at GECCO 2019: Evolution, Robotics and the somersaulting spider.

A few years later, I had the chance to visit his laboratory in Berlin which looked like a scientific exhibition. The lab was full of different experiments or prototypes: windturbines, windmills, 3D models to explain Evolution Strategies… We saw the latest topic he was working on related to geckos found in the desert for which he was studying how the sand is gliding on the skin. The geckos were in a small aquarium in a large office.

I saw Ingo Rechenberg for the last time in Prague in 2019 when we invited him to give the SIGEVO keynote at the closing ceremony of the GECCO conference. This keynote was memorable. He was proud to present his latest findings about the Cebrennus rechenbergi, a new spider species he discovered in the Sahara and named after him. He was happy and enthusiastic to present the robot prototype that was built inspired by the Cebrennus rechenbergi. He was fascinated by this robot.

I found it admirable that the years did not quench his thirst for discovery, his happiness to find things out. His memory will inspire me for many years to come.

Anne Auger

Nikolaus Hansen

I first met Ingo Rechenberg when I attended his lecture "Evolutionsstrategie I" in the late 1980s in Berlin. Rechenberg was an exciting lecturer. He was constantly marching back and forth from one end of the podium to the other, speaking with an energetic voice and visibly moved by the topic he was presenting. I learned about the 1/5th success rule and the surprisingly few assumptions we have to make such that it holds: in short, an undirected (random) change should be successful 20% of the time to be efficient. Changes with higher success rates are too conservative, changes with lower success rates are too ambitious.

Ingo Rechenberg as a young student. From DER SPIEGEL, 18th November 1964.

In his research, Ingo Rechenberg was an engineering visionary. He envisioned a vortex wind concentrator (based on the spread wingtips of eagles) in order to generate more electricity with a smaller rotor. He envisioned producing hydrogen with bacteria on a large scale (based on bacteria he had found in the desert). He envisioned optimizing technical systems the same way biology optimizes organisms, inventing Evolution Strategies. He pursued his visions with scrutiny and passion for years or even decades.

In the mid 1990s, I joined Rechenberg's research lab and started to work on Evolution Strategies. Despite his already undoubted fame, or the Porsche in his garage, or his apparent contentment to be called "Herr Professor", he considered doctoral students as equals. We were independent, we had to be visionary, we had to invent and figure it out (as he did). He often asked for our viewpoint but he never shied away from scientifically challenging and even confronting discussions and disputes. The open and spirited atmosphere in his lab shaped and sharpened my scientific senses and resonates as of today. A remarkable insight he was able to formalize then, in our now common field of research, explained the importance of recombination in (artificial) evolution: in short, recombination allows for much larger variations without destroying as much information. Even in his latest presentation in July 2019, at the age of 84, he revealed a specific result around this idea which was new to me. Ingo Rechenberg never stopped searching and investigating and this spirit and passion shall remain with us.

Nikolaus Hansen

Manuel López-Ibáñez

Prof. Dr Ingo Rechenberg is widely regarded as one of the inventors of Evolution Strategies (ES), one of the variants of what today are more generally known as Evolutionary Algorithms (Rechenberg, 1973). As Prof. Rechenberg recounted (Rechenberg, 2000), the motivation for developing ES was to simply answer the question of: how long would it take simulated evolution to find the optimal design of an airfoil? What is notable about the early experiments carried out by Prof. Rechenberg is that they did not use a mathematical optimization model, nor a simulation implemented in a computer. These were physical experiments carried out in an actual laboratory, some of them using a wind tunnel. The evaluation of a new "solution" required manually adjusting the physical properties of the experiment and measuring new observations. Even the selection and variation steps were sometimes performed using dice rolls and a calculator (Rechenberg, 1965; Knowles, 2009). When I first learned about these experiments, I was impressed by how much of what we do today in terms of practical applications of Evolutionary Computation, Evolutionary Robotics and Bayesian Optimisation was already envisioned in these experiments by Rechenberg, decades before I was born.

I had the pleasure of meeting Prof. Rechenberg in person when he graciously accepted our invitation to give a keynote at GECCO 2019 (Rechenger, 2019). He told us about his annual travels to the Sahara desert to study the locomotion of various animals, including the spider that carries his name. He demonstrated to us a robot prototype capable of mimicking the movements of this spider. Listening to his energetic presentation, it was obvious to me that his research was not motivated by citations, rankings, impact factors or prizes, but by the pure interest in understanding how the natural world works and the joy of demonstrating this understanding by building new things out of this knowledge.


Rechenberg, I. (1965) Cybernetic solution path of an experimental problem. Royal Aircraft Establishment, Library Translation 1122, Farnborough, Reprinted in: D.B. Fogel (Ed.), Evolutionary Computation. The Fossil Record. Selected Readings on the History of Evolutionary Computation. IEEE Press, 301–309, 1998.

Rechenberg, I. (1973) Evolutionsstrategie–Optimisierung technischer Systeme nach Prinzipien der biologischen Evolution, Frommann-Holzboog, Stuttgart.

Rechenberg, I. (2000) Case studies in evolutionary experimentation and computation. Comput. Methods Appl. Mech. Engrg. 186:125–140. doi:10.1016/S0045-7825(99)00381-3

Knowles, J.D. (2009) Closed-loop evolutionary multiobjective optimization. IEEE Computational Intelligence Magazine, 4:77–91. doi: 10.1109/MCI.2009.933095

Rechenberg, I. (2019) Evolution, Robotics and the Somersaulting Spider. Keynote, Genetic and Evolutionary Computation Conference, Prague.

Manuel López-Ibáñez

Günter Rudolph

Not surprisingly, being a PhD student of Hans-Paul Schwefel in the 1990s, I knew the early work of Ingo Rechenberg in the field of evolution strategies, and I have also seen him occasionally at scientific meetings. But the first true personal contact was established only in May 1994 when he invited me to the Technical University of Berlin to present a talk about my research in the regular seminar of his institute.

Prior to the talk, I enjoyed a guided tour through his lab, where I could admire all the exhibits of his pioneering experiments demonstrating the power of evolutionary optimization. But this huge experimental hall held yet another surprise: it also served as a garage for 4WD vehicles with which he made frequent expeditions into the desert in Morocco. During these expeditions, he discovered, among other things, hydrogen-producing bacteria and a species of spider that achieved a faster speed by a kind of rolling locomotion than by running. This biomimetic research of Ingo was unknown to me until then.

After the talk, we gathered in the library of his institute where we had a lively discussion about all kinds of aspects of evolution strategies, which he had also written down in his book "Evolutionary Strategy '94" (in German) published at that time. The hypotheses presented in it were cause and incentive for me (and others) to derive mathematically founded confirmations for them.

Finally, he chauffeured me with his white Porsche to the city airport of Berlin where I caught the evening flight back to Dortmund. This was undoubtedly a successful day and the beginning of a lasting scientific relationship.

The name "Ingo Rechenberg" will be inextricably linked with the emergence of evolutionary strategies and the concepts like the "1/5 success rule" as well as the "evolution window". Our community has lost a special person. We will miss him.

GECCO 2021: Virtualization Report

Arnaud Liefooghe (Virtualization Chair), Nadarajen Veerapen (Electronic Media Chair)

The 2021 Genetic and Evolutionary Computation Conference (GECCO 2021 @ Lille, was organized as an online-only conference on July 10-14. After 2020, this is the second time that GECCO is going online. This report describes the virtual environments used for the conference and provides statistics about the GECCO virtualization.

Let us start by acknowledging all the organizers, in particular Krzysztof Krawiec (General Chair), Francisco Chicano (Editor-in-Chief), Bilel Derbel (Local Chair) and all the other organizers and event chairs ( as well as the business committee and SIGEVO officers.

Virtual Platforms

GECCO 2021 was hosted in two main virtual platforms: Whova and Gather.

  • Whova: All talks, including sessions, keynotes, workshops, and other events involving oral presentations, were streamed live on Whova using the Zoom video teleconferencing service. Pre-recorded videos were hosted on the ACM YouTube channel, and were made available from Whova shortly after the corresponding live session has ended.

  • Gather: Gather was mostly meant for socializing and was used for coffee breaks, poster sessions, and other interactive events. The Gather space also featured virtual lecture rooms from which it was possible to directly join a session or a talk.

Both platforms run in-browser and were cross-linked to facilitate moving between them. While Whova provides a static view of the event and structured access to schedule and content, Gather is a game-like environment where each participant controls a mobile customizable avatar that can interact with other participants face-to-face via video chats and in other ways. Our Gather map emulates a virtual conference center and its surroundings, featuring a spacious foyer, lecture rooms, info desks, exhibitions, drawing tables for technical discussions, and other facilities.

Virtual venue in Gather

To connect better with other participants and so improve the experience, attendees were strongly encouraged to spend most of their time between sessions in Gather, interacting with other participants, visiting exhibitions and exploring the environment. The platform was operating 24/7 for the duration of the conference, so it could be enjoyed also from the less conveniently located time zones. See our promotional video, and the guide for attendees. A number of Zoom backgrounds were also made available to the participants in order to provide a personalized experience of GECCO 2021.

Group photo souvenir

Poster Room 1

The two poster room layouts in Gather were custom designs inspired by previous successful poster sessions, notably at EvoStar 2021 and NeurIPS 2020. However, the GECCO 2021 poster rooms were much larger, with almost 80 posters in each room. This meant providing additional navigation features to help attendees find their way around the space, such as subway-style lines on the floor to direct user traffic and teleportation platforms to move around more easily. Furthermore, a tutorial video was available to help attendees feel more at ease with the Gather platform.

Fortunately, Gather provides an API to interact with the objects in the different spaces. This allowed us to schedule time-dependent links from each Gather session room to the relevant Whova page. Scripts were also used for populating poster data such as titles, descriptions, and images.

Online Schedule

GECCO had a record number of 949 participants in 2021, coming from 62 countries and covering six continents. We believe that the reduced registration fees for non-presenters attracted participants who do not regularly attend GECCO, which has increased the visibility of the conference.

Number of participants per country

However, in the case of online events, this makes it particularly challenging to accommodate the different time zones. Although the largest attendance was centered on Central European Summer Time (UTC+02:00, Lille's time zone), the participants were spanning time zones from UTC-10:00 to UTC+12:00.

Number of participants per time zone (UTC+2 is Lille’s time zone)

Compared to the typical program, we applied a number of changes to the GECCO 2021 schedule in an effort to provide a better online experience. First, given that the attention span of the audience tends to be shorter in online events, oral presentation slots have been reduced to 20 minutes (16 minutes for presentation followed by 4 minutes for discussion). Second, since the participants were in different time zones, we made no distinction between coffee and lunch breaks.

All breaks were made longer than usual coffee breaks, leaving time for lunch in appropriate time zones. Longer breaks were also beneficial to allow for one-on-one interactions in Gather. There were also two slots for poster sessions this year, with each poster being presented in both sessions, once again allowing a better convenience for the different time zones. Due to the aforementioned changes, the duration of the conference has been slightly extended with the use of Wednesday afternoon.



We had a particularly active community on Whova, which was a mandatory entry point to access most sessions. Most participants used the platform to set-up their personal agenda and to communicate with other attendees, either through the public board or using private messages. The publicly-available agenda provided by Whova was accessed more than 4,000 times.

Usage statistics provided by Whova

Average attendance for each type of event, according to Whova

Whova also provides a number of unique viewers for each session, based on both live streams and recorded videos. It should be noted that the statistics provided by Whova seem to underestimate the number of participants, compared to our manual monitoring. The three keynotes and the opening and closing ceremonies were the most attended sessions, while introductory tutorials were the most attended parallel sessions.


For live sessions, 5 Zoom licenses were provided by ACM, and 10 were subscribed for GECCO 2021. One Zoom license was used for the SIGEVO Summer School on July 5-9, and all of them were used (or available as backups) during the first two days or the three main days of the conference.

Usage statistics provided by Zoom: number of meetings/webinars, average number of participants per meetings/webinars, and total duration of meetings/webinars for each conference day

Since there were more parallel sessions in the first two days of the conference, we had 37 meetings per day and accumulated over 2,500 hours of Zoom stream. For the last three days, there were no more than 18 meetings or webinars per day, and we therefore observe a better attendance of about 100 participants per meeting/webinar on average.


GECCO 2021 attendees in Gather probably noticed that moving from the main space to the poster rooms, as well as between poster rooms, was not seamless. Due to technical constraints, including the limit on the number of objects and concurrent users per space, the main space and poster rooms are actually in different spaces (individual silos). Monitoring attendance was therefore important to check whether the simultaneous users constraint was not being reached. Unfortunately, at the time of GECCO'21, Gather did not provide a dashboard to track multiple spaces nor historical attendance data, we therefore recorded the data manually.

Attendance for each Gather space

On the first day of the main conference, the first poster session lasted 2 hours, with 200 or more attendees throughout most of that period. The second 1-hour poster session on the second day received a bit less foot-traffic but was well attended nevertheless. Outside of the poster sessions, the main Gather space was attended by a core group of users, with small noticeable attendance spikes for each break between regular sessions on Whova. On the last day, the massive attendance spike corresponds to the end of the closing ceremony and the invitation of our General Chair, Krzysztof Krawiec, to congregate to Gather for a souvenir group photo.


A total of 441 pre-recorded videos were uploaded to the ACM YouTube channel. They were accessible as backups in case of technical issues, and they were also made available via Whova shortly after the corresponding live session ended.

At the time of submitting their video recording, the authors were asked whether it should be made public, whether it should be deleted, or whether it should be left as is (i.e. only available to registered participants). About 73% of the videos are still available on YouTube (either publicly or privately), while 26% of them have been deleted, on authors' request.

Average number of views per video for each type of event

Among the statistics provided by YouTube, there is an average of around 57 views per video for the 167 public videos, and about 13 views per video for the 156 unlisted videos, as of November 2021. Unsurprisingly, the publicly-available keynote videos got the most views, with 550 on average, followed by introductory and advanced tutorials.


In this time impacted by the pandemic, and even though we believe that online conferences will never reach the charm of on-site conferences, it is our hope that attendees enjoyed the virtualization tools we set up after making the decision to switch to a purely virtual event.

We would like to point out that the GECCO 2021 Gather space currently remains partially open, even after the completion of the conference. In addition, the videos that the authors wanted to make public will remain available on the ACM YouTube channel, keynotes included.


by Thomas Bäck, Pauline Bennet, Jacob de Nobel, Carola Doerr, Johann Dreo, Herilalaina Rakotoarison, Jeremy Rapin, Olivier Teytaud, Diederick Vermetten, Hao Wang, Furong Ye

The Competition

The Open Optimization Competition 2021 was organized by the Nevergrad team at Facebook AI Research (FAIR) and the IOHprofiler team at LIACS, Leiden University, and at LIP6, Sorbonne University.

Motivation and Background

Our teams aim at building open-source software that helps researchers and users of evolutionary computation methods and other gradient-free (“black-box”) optimization methods compare and analyze different solvers. Our key objectives are:

  1. to increase the breadth of problems that can be investigated through a single interface,

  2. the reproducibility of benchmarking,

  3. the ease of adding new modules (new algorithms, new problem collections, new performance measures, but also other components such as feature extraction techniques, automated algorithm configuration techniques, etc.),

  4. the usage of the data that is produced through the benchmarking activity.


The competition was held as part of the IEEE Congress on Evolutionary Computation (CEC’21, June 28-July 1, online event) and the ACM/SIGEVO Genetic and Evolutionary Computation Conference (GECCO’21, July 10-14, online event).

Submissions were invited for two tracks:

  1. a classical, performance-oriented competition, and

  2. an open-ended track, where participants contribute to Nevergrad or IOHprofiler through pull requests, without limitation on scope or complexity of the contribution. Typical submissions suggest new benchmark problems, performance measures/statistics, visualization techniques, or otherwise extend or improve the functionalities of our benchmarking environments

All pull requests made between January 1 and September 30 not involving any member of the organization teams were taken into account.

The submitted entries were reviewed by our nine award committee members:

The jury members have selected three winning entries, which are briefly summarized below:

    • Johann Dreo: Extensible Logging and Empirical Attainment Function for IOHexperimenter

    • Pauline Bennet: Two new photonic problems for benchmarking in Nevergrad

    • Herilalaina Rakotoarison: From hyper-parameter tuning to the design of machine learning pipelines

Congratulations to the three winning entries!

We thank all participants of the Open Optimization Competition 2021 for their diverse contributions. We also thank our nine jury members for accepting the challenging task of selecting the winning entries among these submissions.

Our Benchmarking Platforms

Nevergrad is a Python-based toolbox for gradient-free optimization. Nevergrad offers:

  1. A large number of diverse benchmark problems, ranging from standard academic problems over machine learning benchmarks to numerous real-world problems. In particular, it offers a unified interface to benchmark collections such as LSGO, YABBOB, OpenAI Gym, LMDA, structural engineering problems (for which genetic methods are particularly good), and many others.

  2. State-of-the-art optimization algorithms for optimization problems of various types

  3. High-performing algorithm selection wizards that recommend which algorithm to apply to a given optimization scenario (i.e., it uses information about the problems to be solved as well as the available computational resources and outputs one or more algorithms that can be used to solve this problem)

IOHprofiler is a modular software for benchmarking single-objective black-box optimization algorithms. Its two main components are IOHexperimenter and IOHanalyzer:

  • IOHexperimenter allows users to run their own or built-in solvers on their own or built-in benchmark collections. For example, we use IOHprofiler for running experiments on the

BBOB benchmark collection of COCO and for experiments on the W-model problem generator. Interfaces to automated algorithm configuration software such as irace and MIP-EGO are available as well. Users have full control of the granularity of the data that is logged during the optimization runs.

  • IOHanalyzer is our module for the statistical analysis and visualization of benchmark data. Its key advantages are a highly interactive design, which allows users to analyze algorithms performance and behavior from various perspectives.

  • The IOHprofiler project is available on GitHub. We also have a wiki. Users only interested in using our interactive performance analysis part can find the web-based interface at Note that the latter allows you to explore the majority of the publicly available benchmark data on the single-objective BBOB collection of COCO as well as the extensive results of the Nevergrad platform.

IOHprofiler is not only interesting for research, but also for teaching purposes. Check it out :-)

Nevergrad data can be conveniently analyzed through the web-based interface of IOHanalyzer, available at

Overview of Winning Contributions

Johann Dreo: Extensible Logging and Empirical Attainment Function for IOHexperimenter

In order to allow for large-scale, landscape-aware, per-instance algorithm selection (the future of solvers design) a benchmarking platform software is key. IOHexperimenter provides a large set of synthetic problems, a logging system and a fast implementation.

In this work, we refactored IOHexperimenter’s logging system, in order to make it more extensible and modular. Using this new system, we implement a new logger, which aims at computing the performance metrics of an algorithm across a benchmark. The logger computes the most generic view on an anytime stochastic heuristic's performance, in the form of the Empirical Attainment Function (EAF). We also provide some common statistics on the EAF and its discrete counterpart, the Empirical Attainment Histogram (see an example of an EAH on the left).

You can learn more about this work by reading our technical report. The code has eventually been merged in the IOHexperimenter codebase. Johann Dreo contributed the architecture and the code, with the help of Manuel López-Ibáñez for the EAF algorithmics.

Pauline Bennet: Two new photonic problems for benchmarking in Nevergrad

We implemented two new real-world problems concerning the optimization of photonic structures. We searched for an antireflective coating on a silicon bare for a photovoltaic solar cell, in two slightly different cases.

In the first case ("reference"), we imposed the materials used for the layers of the antireflective coating and let the algorithms find the thicknesses of each of the layers in the structure. The solution obtained is represented in the figure on the left. This periodic structure is completely modular: it is possible to determine the role of specific parts of the structure in the global optical response, so that we are able to fully understand the underlying physical mechanisms. This is rather unique, for it is often not possible to interpret so easily the result of optimization in photonics.

Independently of the number of optimizations and the number of layers chosen for the structure, we systematically obtained the same kind of result. This test case is thus relevant for comparing algorithms because the complexity of the problem can be increased by gradually increasing the number of layers in the structure, while keeping the same kind of solution.

In the second case ("realistic"), the refractive index of the materials and the thicknesses of the layers are the parameters of the structure. This constitutes a more complex problem compared to the previous one since the number of variables is multiplied by 2 for the same number of layers in the structure. Since the refractive index and the thickness of a given layer present a strong dependency, we used structured optimization to improve the convergence of the algorithms. The same kind of solution as for the previous case is found by the optimization systematically.

This work highlights the interest in a structured optimization for concrete physical problems and because it constitutes a strong textbook case, it is particularly relevant to be used for benchmarking.

Herilalaina Rakotoarison: From hyper-parameter tuning to the design of machine learning pipeline

Designing machine learning experiments is a tedious task as it involves several algorithm choices, from data preprocessing and feature preprocessing to the learning model, each associated with its hyper-parameters. Fortunately, an AutoML approach is able to tackle this issue by leveraging optimization algorithms. However, despite the massive interest of the research community in AutoML, benchmarking optimization algorithms on such tasks have been less considered at this time.

In this work, we integrated the task of tuning the Auto-sklearn [1] pipeline in Nevergrad. The design choice in Auto-sklearn consists of 4 data preprocessing, 14 feature preprocessing, and 16 learning algorithms, which overall cover 110 hyper-parameters. Thus, the goal is to find the optimal pipeline for a given dataset (the figure on the left illustrates the results on the Wilt dataset. Is it obtained by the command line:

python -m nevergrad.benchmark autosklearntuning --num_workers 12 --plot

Note that the benchmarking can be extended to any subset of OpenML datasets. Although this contribution was primarily targeted for AutoML practitioners, the inherent nature of AutoML of being an expensive, structural, mixed-type, and black-box optimization might raise some interest in the broader Optimization community.

[1] Feurer, Matthias, et al. “Efficient and Robust Automated Machine Learning.” Advances in Neural Information Processing Systems, vol. 28, Curran Associates, Inc., 2015.

Call for Submissions

Genetic and Evolutionary Computation Conference

GECCO 2022, will take place in Boston, USA during July 9 - 13, 2022. GECCO presents the latest high-quality results in genetic and evolutionary computation since 1999.

Topics include: genetic algorithms, genetic programming, ant colony optimization and swarm intelligence, complex systems (artificial life, robotics, evolvable hardware, generative and developmental systems, artificial immune systems), digital entertainment technologies and arts, evolutionary combinatorial optimization and metaheuristics, evolutionary machine learning, evolutionary multiobjective optimization, evolutionary numerical optimization, real world applications, search-based software engineering, theory and more.

Important dates

  • Abstract submission for full papers: January 27, 2022

  • Full papers submission: February 3, 2022

  • Poster only submission: February 3, 2022

  • Notification of paper acceptance: March 25, 2022

  • Camera-ready: April 14, 2022

  • Presenter mandatory registration: May 2, 2022

ACM SIGEVO Dissertation Award 2022

Submission Deadline: February 1, 2022

All details available at

The SIGEVO Dissertation award recognizes excellent thesis research by doctoral candidates in the field of evolutionary computing. Dissertations will be reviewed for technical depth and significance of the research contribution, potential impact on the field of evolutionary computing, and quality of presentation. The SIGEVO Best Dissertation award is given annually to a maximum of 1 winner and a maximum of 2 honorable mentions. The award presentation takes place at the Genetic and Evolutionary Computation Conference (GECCO) awards ceremony in July.

The beautiful plaque awarded to one of the 2020 honorable mention.

The award carries a monetary value, contributed by SIGEVO, of $2,000 to be awarded to the winner and $1,000 to each of the honorable mentions. The award winner and honorable mentions each receive a beautiful plaque.

Read about the 2021 awards.


Eligible dissertations must have been successfully defended and deposited in the previous calendar year (January to December). Nominations are welcomed from any country, but only English language versions will be accepted. Each nominated dissertation must be on any topic relevant to evolutionary computing. The determination of whether a thesis is within the scope of the award will be made by the SIGEVO Dissertation Award Committee. A dissertation can be nominated for both the SIGEVO Dissertation Award and the ACM Doctoral Dissertation Award.

Submission Procedure

All nomination materials must be submitted electronically to the current chair of the SIGEVO Dissertation Award Committee (Manuel López-Ibáñez <>, please ask for confirmation) by the submission deadline, in English. PDF format is preferred for all materials. Late submissions will not be considered. The nomination must come from the dissertation advisor(s); self-nomination is not allowed.

Nomination for the award must include

1. An English language copy of the thesis in legible pdf format.

2. A statement from the advisor(s), limited to 2 pages, addressing why the nominee’s dissertation should receive this award. This statement should discuss the significance of the dissertation in depth.

3. Three (3) letters of support (in addition to the nomination letter) limited to a maximum of 2 pages each. Supporting letters should be included from other experts in the field who can provide additional insights or evidence of the dissertation’s impact. If a letter writer is supporting more than one nomination in a year, they may be asked to rank those nominations. At least two letters must come from experts outside the nominee’s university.

4. A list of publications by the nominee.

5. Suggested citation: This should be a concise statement (maximum of 50 words) describing the key technical contribution for which the candidate merits this award. Final wording for awardees will be at the discretion of the SIGEVO Dissertation Award Committee.


For any questions, please contact the SIGEVO Dissertation Award Committee Chair: Manuel López-Ibáñez (

AutoML-Conf 2022: 1st International Conference on Automated Machine Learning

Following a series of 8 AutoML workshops at the International Conference on Machine Learning (ICML), it is our great pleasure to announce the 1st International Conference on Automated Machine Learning, AutoML-Conf 2022.

AutoML-Conf'22 will be co-located with ICML'22 during July 25 - 27 in Baltimore, USA. We note that there are currently no plans for a hybrid conference.

The first international conference on automated machine learning (AutoML-Conf 2022) brings together researchers and users, with the goals of developing automated methods for speeding up the development of machine learning applications, obtaining improved performance, and thereby democratizing machine learning

Important Dates

  • Abstract submission: February 24th, 2022

  • Full paper submission and co-author registration: March 3th, 2022

  • Tutorial proposal submission: April 1st, 2022

  • Conference: July 2527, 2022

Program Chairs: Isabelle Guyon, Marius Lindauer, Mihaela van der Schaar

Parallel Problem Solving from Nature (PPSN 2022)

The 17th International Conference on Parallel Problem Solving from Nature (PPSN 2022) will be held in Dortmund (Germany), 10–14 September 2022. The workshops, tutorials and the conference will be held “on-site” only. The participants should take this into account in their planning. The local organizers will provide adequate rooms and technical equipment for local presentations. A hybrid format is not planned. In case of pandemic-caused restrictions, the event will have to switch to a fully online format.

PPSN was originally designed to bring together researchers and practitioners in the field of Natural Computing, the study of computing approaches that are gleaned from natural models. Today, the conference series has evolved and welcomes works on all types of iterative optimization heuristics. Notably, we also welcome submissions on connections between search heuristics and machine learning or other artificial intelligence approaches.

Submissions covering the entire spectrum of work, ranging from rigorously derived mathematical results to carefully crafted empirical studies, are invited.

Important Dates

  • Paper submission due: 1st April 2022

  • Tutorials/Workshops: 10–11 September 2022

  • Conference: 12–14 September 2022

Program Chairs: Hernan Aguirre, Pascal Kerschke, Gabriela Ochoa, Tea Tusar

The 11th InternationalWorkshop on Genetic Improvement

GI 2022 will be co-located with the Genetic and Evolutionary Computation Conference, GECCO 2022, in Boston, USA during July 9 - 13.

Boston Cream Pie

We invite submissions that discuss recent developments in all areas of research on, and applications of, Genetic Improvement. GI is the premier workshop in the field and provides an opportunity for researchers interested in automated program repair and software optimisation to disseminate their work, exchange ideas and discover new research directions.

Important Dates

  • Paper submission due: 11th April 2022

  • Workshops: 9 or 10 of July 2022

Workshop Chairs: Bobby R. Bruce, Vesna Nowack, Aymeric Blot, Emily Winter, W. B. Langdon, Justyna Petke

The Human-Competitive Awards

Humies 2022 will take place in conjunction with the Genetic and Evolutionary Computation Conference, GECCO 2022, in Boston, USA during July 9 - 13.

Entries are hereby solicited for awards totalling $10,000 for human-competitive results that have been produced by any form of genetic and evolutionary computation (including, but not limited to genetic algorithms, genetic programming, evolution strategies, evolutionary programming, learning classifier systems, grammatical evolution, gene expression programming, differential evolution, etc.) and that have been published in the open literature between the deadline for the previous competition and the deadline for the current competition.

Important Dates

  • Friday, May 27, 2022 — Deadline for entries (consisting of one TEXT file, PDF files for one or more papers, and possible "in press" documentation (explained below). Please send entries to goodman at msu dot edu.

  • Friday, June 10, 2022 — Finalists will be notified by e-mail.

  • Friday, June 24, 2022 — Finalists not presenting in person must submit a 10-minute video presentation (or the link and instructions for downloading the presentation, NOT a YouTube link) to goodman at msu dot edu.

  • July 9-13, 2022 (Saturday - Wednesday) — GECCO conference (the schedule for the Humies session is not yet final, so please check the GECCO program as it is updated).

  • Monday, July 11, 2022 Presentation session.

  • Wednesday, July 13, 2023 — Announcement of awards at the plenary session of the GECCO conference.

Judging Committee: Erik Goodman, Una-May O'Reilly, Wolfgang Banzhaf, Darrell Whitley, Lee Spector, Stephanie Forrest

Publicity Chair: William Langdon

Forthcoming Events

EvoStar 2022

EvoStar 2022 is planned as a hybrid event (online and onsite) from 20 to 22 April 2022. Onsite, EvoStar will be held in Seville, Spain.

EvoStar is comprised of four co-located conferences

  • EuroGP 25th European Conference on Genetic Programming

  • EvoApplications 25th European Conference on the Applications of Evolutionary and bio-inspired Computation

  • EvoCOP 22nd European Conference on Evolutionary Computation in Combinatorial Optimisation

  • EvoMUSART 11th International Conference (and 18th European event) on Evolutionary and Biologically Inspired Music, Sound, Art and Design

About this Newsletter

SIGEVOlution is the newsletter of SIGEVO, the ACM Special Interest Group on Genetic and Evolutionary Computation. To join SIGEVO, please follow this link: [WWW].

We solicit contributions in the following categories:

Art: Are you working with Evolutionary Art? We are always looking for nice evolutionary art for the cover page of the newsletter.

Short surveys and position papers: We invite short surveys and position papers in EC and EC related areas. We are also interested in applications of EC technologies that have solved interesting and important problems.

Software. Are you a developer of a piece of EC software, and wish to tell us about it? Then send us a short summary or a short tutorial of your software.

Lost Gems. Did you read an interesting EC paper that, in your opinion, did not receive enough attention or should be rediscovered? Then send us a page about it.

Dissertations. We invite short summaries, around a page, of theses in EC-related areas that have been recently discussed and are available online.

Meetings Reports. Did you participate in an interesting EC-related event? Would you be willing to tell us about it? Then send us a summary of the event.

Forthcoming Events. If you have an EC event you wish to announce, this is the place.

News and Announcements. Is there anything you wish to announce, such as an employment vacancy? This is the place.

Letters. If you want to ask or to say something to SIGEVO members, please write us a letter!

Suggestions. If you have a suggestion about how to improve the newsletter, please send us an email.

Contributions will be reviewed by members of the newsletter board. We accept contributions in plain text, MS Word, or Latex, but do not forget to send your sources and images.

Enquiries about submissions and contributions can be emailed to

All the issues of SIGEVOlution are also available online at:

Notice to contributing authors to SIG newsletters

By submitting your article for distribution in the Special Interest Group publication, you hereby grant to ACM the following non-exclusive, perpetual, worldwide rights:

  • to publish in print on condition of acceptance by the editor

  • to digitize and post your article in the electronic version of this publication

  • to include the article in the ACM Digital Library

  • to allow users to copy and distribute the article for noncommercial, educational or research purposes

However, as a contributing author, you retain copyright to your article and ACM will make every effort to refer requests for commercial use directly to you.

Editor: Gabriela Ochoa

Sub-editor: James McDermott

Associate Editors: Emma Hart, Una-May O'Reilly, Nadarajen Veerapen, and Darrell Whitley