Giuseppe Cuccu
Machine learning is the art of adapting generic solvers to address specific problems.
Short Bio
I am a Senior Researcher in Machine Learning with the eXascale Infolab, University of Fribourg, Switzerland under prof. Philippe Cudré-Mauroux.
Since 2022 I am a Maître Assistant with the Human-IST Institute under prof. Denis Lalanne as Coordinator and Study Advisor for the new interfaculty Specialized Master in Digital Neuroscience.
I teach the courses of Machine Learning (BSc in CS) and Data Analytics in Python (DigitalSkills).
I obtained my Ph.D. in 2018, after a non-linear career started in 2009, under the supervision of prof. Philippe Cudré-Mauroux.
Along the way I worked for different start-ups (and founded one), and as Research Assistant with the Dalle Molle Institute for Artificial Intelligence (5y) and with the ICLS group at ZHAW (1y).
I obtained my MSc in Computer Science Summa cum Laude from the University of Milano-Bicocca, with minors in software engineering and soft computing.
I have been involved in Machine Learning research since 2008 – mostly on
Black-Box Optimization,
Neural Networks,
Neuroevolution and
Reinforcement Learning Control,
addressing challenging real-world applications subject to strict data constraints (low quality / quantity).
Neural Networks
Widely adopted parameterized generic function approximators.
Depending on their parametrization, neural networks are in principle capable of approximating any computable function. This however correlates with network size, which makes sophisticated models computationally expensive to train and maintain.
While their representation as directed graphs makes NN easy to design and extend for humans, they are particularly hard to design automatically, i.e. Architecture Search.
Of course they have gained the spotlight as primary machine learning model thanks to the vast success of Deep Learning, which trains deep (i.e. many-layer) networks using variants of the Backpropagation algorithm, a stochastic gradient descent method, thereby limited to Supervised Learning problems.
The main limitation of this setup often lies is the Vanishing Gradient, where the last layers obtain high quality error-gradient information leading to effective updates, while the first layers only receive the lower quality gradient information proportionally to the number of layers, thus quickly making the training data hungry, and unsuitable for applications with limited data availability.
This is of course a problem in real-world scenarios, as often data scarcity is simply unavoidable in crucial applications, because of the complexities related to data collection and/or labeling, such as continuous control, direct policy search, and computer-aided diagnosis.
Alternative training methods such as Black-Box Optimization supersede these limitations, offering a promising complement to Deep Learning in domains where the applicability of the latter is limited.
×
Black-Box Optimization
A field studying parameter search algorithms that make no assumption on the application of the parameter set.
These multi-agent methods (e.g. Evolutionary Algorithms) are applicable to any context without restriction, assumptions or requirements, the only downside being that they are computationally expensive. The parameters are learned to optimize a target function (fitness), without knowing (nor needing access to) its internal workings.
Their progress is based on (i) sampling the parameters space, (ii) computing the value of the objective function over the samples, then (iii) approximating the objective function's gradient over the samples, and (iv) generate new candidate solutions to maximize expected fitness.
The only real downside is that they are computationally expensive, and results performance can be limited when applied improperly.
×
Neuroevolution
A technique that trains Neural Networks models using Black-Box Optimization, typically as some form of Evolutionary Algorithms.
This is most commonly used to address Reinforcement Learning problems directly, such as Direct Policy Search for Continuous Control.
Evolutionary Algorithms search in the space of network parameters to optimize the network performance on a task, solely led by a fitness function that scores the performance of candidate solutions.
This corresponds to directly search in the space of neural networks, optimizing an unrestricted scoring function which can represent the problem to arbitrary precision and complexity. This makes Neuroevolution directly applicable to Reinforcement Learning Control by means of Direct Policy Search, without the need to set up the classic Reinforcement Learning framework (i.e. states, actions, values, policies, etc.).
×
Reinforcement Learning
A learning paradigm where the only direction for improvement is given by a score function over candidate solutions.
[ Not to be confused with the Classical Reinforcement Learning framework, which is the most common approach to this class of problems, where the task is described by a Markov decision process with states, actions, values, policies, etc., and the goal is to optimize the Bellman equation. ]
While Supervised Learning methods expect a set of labeled data "to copy from", Reinforcement Learning only provides an objective evaluation for candidate solutions, and leaves the algorithm to figure out how to utilize this information to improve the performance of proposed solutions over time.
This is regarded as the natural way for living systems to learn most of their behaviors: by trial and error, expectation and surprise, as opposed to learning theory in isolation.
Notably, Black-Box Optimization fits this paradigm without further manipulations and setups.
×
Research Interests
My overarching goal is to solve
Reinforcement Learning Continuous Control,
by any means necessary. This requires the study, improvement or even creation of methods uniquely based on their ability to scale to real-world conditions – which accounts for an unusual breadth of topics in my work.
[Why I think it matters]
Reinforcement Learning Continuous Control
A class of problems addressing sequential interaction with a system over time.
At each step, an agent receives an observation from the system (based on its internal state) and returns a decision based on it. No examples are provided, so Supervised Learning is not applicable: the system provides only a reward signal for improving the solution, which makes it belong to the Reinforcement Learning paradigm. The reward can be offered at specific times during the interaction, but in real-world tasks it is most commonly sparse, only offering guidance for extreme cases (e.g. robot walker that falls down), with an overall performance score at the very end (e.g. how far did the robot walker go in a fixed time). For this reason temporal awareness is typically an advantage, and successful controllers often maintain an own internal state (memory).
×
Why Reinforcement Learning Continuous Control
There is a big "elephant in the room" in Machine Learning: where is my self-driving car?
We still wish (and often promise) to solve challenging real-world control applications,
yet despite extraordinary results in controlled environment (e.g. videogames), most control tasks nowadays still rely on human operators.
This is not a problem of platform accessibility anymore, since modern electronic systems for human control have become ubiquitous (e.g. "drive-by-wire" vehicles), heavy machinery have become as easy to control programmatically as more classic robotic platforms. With the recent boom of Machine Learning, and so many state-of-the-art methods now freely accessible online, anyone can do it.
This only makes this question more relevant: where is my self-driving car?
Truth is, we know how to drive a car, and we know how to teach another person how to drive a car, but we do not have yet a sure-fire Machine Learning method that can learn from a human teacher in the same way a human does.
We keep creating new, ad-hoc procedures to teach stuff we already know, simplifying our systems and depending on laboratory conditions, but this approach simply cannot scale: autonomous machines need to coexist by our side, safely supporting humans and growing alongside them.
As long as new methods keep depending on these constraints and limitations, we cannot just work out one final improvement to solve Reinforcement Learning Continuous Control.
We need a radical paradigm shift in Machine Learning.
We need machines that learn truly without supervision, by interacting with an unbound, partially observable and noisy world. We also need them to understand feedback, and incorporating it into their learning seamlessly. We cannot produce ahead correct control examples for any possible observation, nor a reliable feedback signal for each atomic action taken. Control functions are by necessity complex, rarely if at all differentiable or continuous, and we have still a long way to go to model memory.
We try, we fail, we pick ourselves up using our own intrinsic motivation, without oracles nor complete observability. We improve skills in isolation and adapt them to apply them in different contexts, and we can share our finding by just speaking or demonstrating to each other.
And yet, we manage.
We need machines that can manage, too.
×
Some specific topics and challenges include:
- Alternative models with performance comparable to classic Neural Networks, but including a set of desirable features that are difficult to achieve with NNs such as:
- Native compatibility with Architecture Search, to automatically scale model complexity
- Sub-task recognition and autonomous specialization, dividing responsibilities
- Access to explicit memory structures, with focus on object permanence, attention mechanisms, and expert knowledge integration
- Human interpretability, both in the control model and its memory, to improve trust in high-responsibility applications
- Alternative training methods applicable to such sophisticated models as described above, including:
- High parallelization and scalability for Black-Box Optimization
(check out DiBB!
[code]
- Adaptation of Deep Learning to problems with limited data availability
(check out Typhon!
[code]
- Autonomous fitness shaping based on intrinsic motivation, focusing on:
- Competing signals and multi-objective optimization trade-offs
- Self-balancing integration of explicit exploration policies
- Fitness hyperparameters meta-search
Selected Awards
- Best Paper Award. International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2019).
- Best Paper Award Runner-Up. Foundation of Digital Games (FDG 2013).
- Best Paper Award. International Conference on Evolutionary Computation (ICEC 2009).
- Undergraduate Fellowship (BSc+MSc). University of Milano-Bicocca (2003–2009)
Teaching
[SA23]: Data Analytics in Python
Archive
[SP23]: Machine Learning [Moodle]
[SA22]: Data Analytics in Python
[Moodle]
[SP22]: Machine Learning
[Moodle]
[SP21]: Machine Learning
[Moodle]
[SP20]: Machine Learning
[Moodle]
Student Thesis
I supervise a limited number of exceptional students depending on current availability.
Excelling in the Machine Learning course (or equivalent preparation) is typically a prerequisite, as it simplifies communication while significantly lowering the student’s time and effort to graduation.
Available topics:
- Real-world control using Direct Policy Search for Reinforcement Learning
- Exploring alternatives to Neural Networks to model discontinuous functions
- Further scaling the performance of Black-Box Optimization to enable training increasingly complex models
- Augmenting control models with new memory implementations, focusing on interpretability, attention-based recall, and object permanence
- Tumor segmentation for Computer Aided Diagnosis using Deep Learning: overcoming the challenge of data scarcity in medical imaging applications
[collaboration with the Cantonal Hospital of Fribourg, H-FR]
- Application of our new Typhon meta-learning framework to tasks constrained by data scarcity and overfitting
- Integrating Explicit Exploration techniques into exploitation-only methods (e.g. Deep Learning)
- More available on request.
Ongoing work:
- MSc: Albin Aliu:
Multi-step predictors on complex time series
- MSc: Christophe Broillet:
Extending the applicability of the Typhon framework
Alumni:
(* -> peer-reviewed publication derived from thesis work)
- 2023.07 MSc - Jonas Fontana:
Improving Feature-Space Generalization Using the Typhon Framework
[pdf]
- 2023.05 BSc - David Gauch:
Alternative Models for Direct Policy Search in Reinforcement Learning Control Problems
[pdf]
- 2022.08 MSc - Corina Masanti:
Alternative Models for Direct Policy Search in Reinforcement Learning Control Problems
[pdf]
- 2022.08 MSc* - Fabien Vorpe:
Neuroevolution Applications for the DiBB Framework
[pdf]
- 2022.06 MSc - Nicolas Roguet:
Controlling Complex Unstable Robotic Systems Using Direct Policy Search
[pdf]
- 2022.02 BSc* - Christophe Broillet:
Pre-Processing Segmentation Datasets for the Hydra Framework
[pdf]
- 2021.08 MSc* - Luca Rolshoven:
Study and Extension of the DiBB Framework for Distributing Black-Box Optimization
[pdf]
- 2020.12 MSc - Jiyoung Lee:
P-Hydra: Bridging Transfer Learning And Multitask Learning
[pdf]
- 2020.03 MSc* - Johan Jobin & Julien Clément:
Prostate Cancer Classification: A Transfer Learning Approach to Integrate Information From Diverse Body Parts
[pdf]
- 2019.11 BSc - David Bucher:
Data Preparation and Analysis in Support to Cheating Detection: The Case for Economic Momentum in CS:GO
[pdf]
Publications
- Giuseppe Cuccu, Christophe Broillet, Carolin Reischauer, Harriet Thöny, and Philippe Cudré-Mauroux. “Typhon: Parallel Transfer on Heterogeneous Datasets for Cancer Detection in Computer-Aided Diagnosis.” In 2022 IEEE International Conference on Big Data, BigData, 2022. Bibtex PDF Code
Typhon: Parallel Transfer on Heterogeneous Datasets for Cancer Detection in Computer-Aided Diagnosis
Giuseppe Cuccu, Christophe Broillet, Carolin Reischauer, Harriet Thöny, and Philippe Cudré-Mauroux. “Typhon: Parallel Transfer on Heterogeneous Datasets for Cancer Detection in Computer-Aided Diagnosis.” In 2022 IEEE International Conference on Big Data, BigData, 2022.
@inproceedings{cuccu2022bigdata,
title = {{Typhon}: Parallel Transfer on Heterogeneous Datasets for Cancer Detection in Computer-Aided Diagnosis},
author = {Cuccu, Giuseppe and Broillet, Christophe and Reischauer, Carolin and Thöny, Harriet and Cudré-Mauroux, Philippe},
booktitle = {2022 {IEEE} International Conference on Big Data, {BigData}},
year = {2022},
month = dec,
code = {https://github.com/eXascaleInfolab/typhon},
url = {https://exascale.info/assets/pdf/cuccu2022bigdata.pdf}
}
×
- Giuseppe Cuccu, Luca Rolshoven, Fabien Vorpe, Philippe Cudré-Mauroux, and Tobias Glasmachers. “DiBB: Distributing Black-Box Optimization.” In Proceedings of the Genetic and Evolutionary Computation Conference, 341–49. GECCO ’22. New York, NY, USA: Association for Computing Machinery, 2022. Bibtex PDF Code
DiBB: Distributing Black-Box Optimization
Giuseppe Cuccu, Luca Rolshoven, Fabien Vorpe, Philippe Cudré-Mauroux, and Tobias Glasmachers. “DiBB: Distributing Black-Box Optimization.” In Proceedings of the Genetic and Evolutionary Computation Conference, 341–49. GECCO ’22. New York, NY, USA: Association for Computing Machinery, 2022.
@inproceedings{cuccu2022gecco,
author = {Cuccu, Giuseppe and Rolshoven, Luca and Vorpe, Fabien and Cudr\'{e}-Mauroux, Philippe and Glasmachers, Tobias},
title = {{DiBB}: Distributing Black-Box Optimization},
year = {2022},
isbn = {9781450392372},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
doi = {10.1145/3512290.3528764},
booktitle = {Proceedings of the Genetic and Evolutionary Computation Conference},
pages = {341–349},
numpages = {9},
keywords = {parallelization, neuroevolution, evolution strategies, distributed algorithms, black-box optimization},
location = {Boston, Massachusetts},
series = {GECCO '22},
code = {https://github.com/giuse/dibb},
url = {https://exascale.info/assets/pdf/cuccu2022gecco.pdf}
}
×
- Ines Arous, Ljiljana Dolamic, Jie Yang, Akansha Bhardwaj, Giuseppe Cuccu, and Philippe Cudré-Mauroux. “MARTA: Leveraging Human Rationales for Explainable Text Classification.” In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI 2021). A Virtual Conference, 2021. Bibtex Slides PDF Appendix Code Poster
MARTA: Leveraging Human Rationales for Explainable Text Classification
Ines Arous, Ljiljana Dolamic, Jie Yang, Akansha Bhardwaj, Giuseppe Cuccu, and Philippe Cudré-Mauroux. “MARTA: Leveraging Human Rationales for Explainable Text Classification.” In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI 2021). A Virtual Conference, 2021.
@inproceedings{arous2021aaai,
title = {MARTA: Leveraging Human Rationales for Explainable Text Classification},
author = {Arous, Ines and Dolamic, Ljiljana and Yang, Jie and Bhardwaj, Akansha and Cuccu, Giuseppe and Cudr{\'e}-Mauroux, Philippe},
booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence (AAAI 2021)},
year = {2021},
address = {A Virtual Conference},
url = {https://exascale.info/assets/pdf/arous2021aaai.pdf},
appendix = {https://exascale.info/assets/pdf/arous2020AAAI-sm.pdf},
note = {https://exascale.info/assets/pdf/arous2021aaai_slides.pdf},
poster = {https://exascale.info/assets/pdf/arous2021aaai_poster.pdf},
code = {https://github.com/eXascaleInfolab/MARTA}
}
×
- Giuseppe Cuccu, Julian Togelius, and Philippe Cudré-Mauroux. “Playing Atari with Few Neurons.” Autonomous Agents and Multi-Agent Systems 35, no. 2 (2021): 1–23. Bibtex PDF
Playing Atari with few neurons
Giuseppe Cuccu, Julian Togelius, and Philippe Cudré-Mauroux. “Playing Atari with Few Neurons.” Autonomous Agents and Multi-Agent Systems 35, no. 2 (2021): 1–23.
@article{cuccu2021playing,
title = {Playing Atari with few neurons},
author = {Cuccu, Giuseppe and Togelius, Julian and Cudr{\'e}-Mauroux, Philippe},
journal = {Autonomous Agents and Multi-Agent Systems},
volume = {35},
number = {2},
pages = {1--23},
year = {2021},
doi = {10.1007/s10458-021-09497-8},
url = {https://exascale.info/assets/pdf/cuccu2021jaamas.pdf},
publisher = {Springer}
}
×
- Giuseppe Cuccu, Julian Togelius, and Philippe Cudré-Mauroux. “Playing Atari with Six Neurons (Extended Abstract).” In Proceedings of the Twenty-Ninth International Joint Conference On
Artificial Intelligence, IJCAI-20, edited by Christian Bessiere, 4711–15. International Joint Conferences on Artificial Intelligence Organization, 2020. Bibtex Slides PDF
Playing Atari with Six Neurons (Extended Abstract)
Giuseppe Cuccu, Julian Togelius, and Philippe Cudré-Mauroux. “Playing Atari with Six Neurons (Extended Abstract).” In Proceedings of the Twenty-Ninth International Joint Conference On
Artificial Intelligence, IJCAI-20, edited by Christian Bessiere, 4711–15. International Joint Conferences on Artificial Intelligence Organization, 2020.
@inproceedings{cuccu2020ijcai,
title = {Playing Atari with Six Neurons (Extended Abstract)},
author = {Cuccu, Giuseppe and Togelius, Julian and Cudré-Mauroux, Philippe},
booktitle = {Proceedings of the Twenty-Ninth International Joint Conference on
Artificial Intelligence, {IJCAI-20}},
publisher = {International Joint Conferences on Artificial Intelligence Organization},
editor = {Bessiere, Christian},
pages = {4711--4715},
year = {2020},
month = jul,
note = {Sister Conferences Best Papers},
doi = {10.24963/ijcai.2020/651},
url = {https://exascale.info/assets/pdf/cuccu2020ijcai.pdf}
}
×
- Giuseppe Cuccu, Johan Jobin, Julien Clément, Akansha Bhardwaj, Carolin Reischauer, Harriet Thöny, and Philippe Cudré-Mauroux. “Hydra: Cancer Detection Leveraging Multiple Heads and Heterogeneous Datasets.” In 2020 IEEE International Conference on Big Data, BigData, 2020. Bibtex PDF
Hydra: Cancer Detection Leveraging Multiple Heads and Heterogeneous Datasets
Giuseppe Cuccu, Johan Jobin, Julien Clément, Akansha Bhardwaj, Carolin Reischauer, Harriet Thöny, and Philippe Cudré-Mauroux. “Hydra: Cancer Detection Leveraging Multiple Heads and Heterogeneous Datasets.” In 2020 IEEE International Conference on Big Data, BigData, 2020.
@inproceedings{cuccu2020bigdata,
title = {{Hydra}: Cancer Detection Leveraging Multiple Heads and Heterogeneous Datasets},
author = {Cuccu, Giuseppe and Jobin, Johan and Clément, Julien and Bhardwaj, Akansha and Reischauer, Carolin and Thöny, Harriet and Cudré-Mauroux, Philippe},
booktitle = {2020 {IEEE} International Conference on Big Data, {BigData}},
year = {2020},
month = dec,
url = {https://exascale.info/assets/pdf/cuccu2020bigdata.pdf}
}
×
- Declan Oller, Tobias Glasmachers, and Giuseppe Cuccu. “Analyzing Reinforcement Learning Benchmarks with Random Weight Guessing.” In Proceedings of the 19th International Conference on Autonomous Agents and MultiAgent Systems. International Foundation for Autonomous Agents and Multiagent Systems, 2020. Bibtex PDF
Analyzing Reinforcement Learning Benchmarks with Random Weight Guessing
Declan Oller, Tobias Glasmachers, and Giuseppe Cuccu. “Analyzing Reinforcement Learning Benchmarks with Random Weight Guessing.” In Proceedings of the 19th International Conference on Autonomous Agents and MultiAgent Systems. International Foundation for Autonomous Agents and Multiagent Systems, 2020.
@inproceedings{oller2020aamas,
title = {Analyzing Reinforcement Learning Benchmarks with Random Weight Guessing},
author = {Oller, Declan and Glasmachers, Tobias and Cuccu, Giuseppe},
booktitle = {Proceedings of the 19th International Conference on Autonomous Agents and MultiAgent Systems},
year = {2020},
organization = {International Foundation for Autonomous Agents and Multiagent Systems},
url = {https://exascale.info/assets/pdf/oller2020aamas.pdf}
}
×
- Giuseppe Cuccu, Julian Togelius, and Philippe Cudré-Mauroux. “Playing Atari with Six Neurons.” In Proceedings of the 18th International Conference on Autonomous Agents and MultiAgent Systems, 998–1006. International Foundation for Autonomous Agents and Multiagent Systems, 2019. Bibtex PDF
Playing Atari with six neurons
Giuseppe Cuccu, Julian Togelius, and Philippe Cudré-Mauroux. “Playing Atari with Six Neurons.” In Proceedings of the 18th International Conference on Autonomous Agents and MultiAgent Systems, 998–1006. International Foundation for Autonomous Agents and Multiagent Systems, 2019.
@inproceedings{cuccu2019aamas,
title = {Playing {A}tari with six neurons},
author = {Cuccu, Giuseppe and Togelius, Julian and Cudr{\'e}-Mauroux, Philippe},
booktitle = {Proceedings of the 18th International Conference on Autonomous Agents and MultiAgent Systems},
pages = {998--1006},
year = {2019},
organization = {International Foundation for Autonomous Agents and Multiagent Systems},
url = {https://exascale.info/assets/pdf/cuccu2019aamas.pdf}
}
×
- Giuseppe Cuccu. “Extending the Applicability of Neuroevolution.” PhD thesis, Université de Fribourg, 2018. Bibtex PDF
Extending the Applicability of Neuroevolution
Giuseppe Cuccu. “Extending the Applicability of Neuroevolution.” PhD thesis, Université de Fribourg, 2018.
@phdthesis{cuccu2018phd,
author = {Cuccu, Giuseppe},
title = {Extending the Applicability of Neuroevolution},
school = {Universit{\'e} de Fribourg},
address = {Fribourg, Switzerland},
year = {2018},
url = {https://exascale.info/assets/pdf/cuccu2018phd.pdf}
}
×
- Giuseppe Cuccu, Somayeh Danafar, Philippe Cudré-Mauroux, Martin Gassner, Stefano Bernero, and Krzysztof Kryszczuk. “A Data-Driven Approach to Predict NOx-Emissions of Gas Turbines.” In 2017 IEEE International Conference on Big Data, BigData 2017, Boston,
MA, USA, December 11-14, 2017, 1283–88, 2017. Bibtex PDF
A data-driven approach to predict NOx-emissions of gas turbines
Giuseppe Cuccu, Somayeh Danafar, Philippe Cudré-Mauroux, Martin Gassner, Stefano Bernero, and Krzysztof Kryszczuk. “A Data-Driven Approach to Predict NOx-Emissions of Gas Turbines.” In 2017 IEEE International Conference on Big Data, BigData 2017, Boston,
MA, USA, December 11-14, 2017, 1283–88, 2017.
@inproceedings{cuccu2017bigdata,
author = {Cuccu, Giuseppe and Danafar, Somayeh and Cudr{\'e}-Mauroux, Philippe and Gassner, Martin and Bernero, Stefano and Kryszczuk, Krzysztof},
title = {A data-driven approach to predict NOx-emissions of gas turbines},
booktitle = {2017 {IEEE} International Conference on Big Data, BigData 2017, Boston,
MA, USA, December 11-14, 2017},
pages = {1283--1288},
year = {2017},
doi = {10.1109/BigData.2017.8258056},
url = {https://exascale.info/assets/pdf/cuccu2017bigdata.pdf}
}
×
- Jan Koutník, Giuseppe Cuccu, Jürgen Schmidhuber, and Faustino Gomez. “Evolving Large-Scale Neural Networks for Vision-Based Reinforcement Learning.” In Proceedings of the 15th Annual Conference on Genetic and Evolutionary Computation, 1061–68. GECCO ’13. ACM, 2013. Bibtex PDF
Evolving Large-scale Neural Networks for Vision-based Reinforcement Learning
Jan Koutník, Giuseppe Cuccu, Jürgen Schmidhuber, and Faustino Gomez. “Evolving Large-Scale Neural Networks for Vision-Based Reinforcement Learning.” In Proceedings of the 15th Annual Conference on Genetic and Evolutionary Computation, 1061–68. GECCO ’13. ACM, 2013.
@inproceedings{koutnik2013gecco,
location = {New York, {NY}, {USA}},
title = {Evolving Large-scale Neural Networks for Vision-based Reinforcement Learning},
isbn = {978-1-4503-1963-8},
url = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.451.760&rep=rep1&type=pdf},
doi = {10.1145/2463372.2463509},
series = {{GECCO} '13},
pages = {1061--1068},
booktitle = {Proceedings of the 15th Annual Conference on Genetic and Evolutionary Computation},
publisher = {{ACM}},
author = {Koutník, Jan and Cuccu, Giuseppe and Schmidhuber, Jürgen and Gomez, Faustino},
year = {2013},
keywords = {Computer vision, neuroevolution, recurrent neural networks, reinforcement learning}
}
×
- Jan Koutník, Giuseppe Cuccu, Jürgen Schmidhuber, and Faustino Gomez. “Evolving Large-Scale Neural Networks for Vision-Based TORCS.” In Foundations of Digital Games (FDG), 206–12, 2013. Bibtex PDF
Evolving Large-Scale Neural Networks for Vision-Based TORCS
Jan Koutník, Giuseppe Cuccu, Jürgen Schmidhuber, and Faustino Gomez. “Evolving Large-Scale Neural Networks for Vision-Based TORCS.” In Foundations of Digital Games (FDG), 206–12, 2013.
@inproceedings{koutnik2013fdg,
location = {Chania, Crete, {GR}},
title = {Evolving Large-Scale Neural Networks for Vision-Based {TORCS}},
isbn = {{ISBN} 978-0-9913982-0-1},
url = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.654.9135&rep=rep1&type=pdf},
pages = {206--212},
booktitle = {Foundations of Digital Games (FDG)},
author = {Koutník, Jan and Cuccu, Giuseppe and Schmidhuber, Jürgen and Gomez, Faustino},
year = {2013}
}
×
- Giuseppe Cuccu, and Faustino Gomez. “Block Diagonal Natural Evolution Strategies.” In Parallel Problem Solving from Nature - PPSN XII, 488–97. Springer, Berlin, Heidelberg, 2012. Bibtex PDF
Block Diagonal Natural Evolution Strategies
Giuseppe Cuccu, and Faustino Gomez. “Block Diagonal Natural Evolution Strategies.” In Parallel Problem Solving from Nature - PPSN XII, 488–97. Springer, Berlin, Heidelberg, 2012.
@inproceedings{cuccu2012ppsn,
title = {Block Diagonal Natural Evolution Strategies},
url = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.308.655&rep=rep1&type=pdf},
doi = {10.1007/978-3-642-32964-7_49},
eventtitle = {International Conference on Parallel Problem Solving from Nature},
pages = {488--497},
booktitle = {Parallel Problem Solving from Nature - {PPSN} {XII}},
publisher = {Springer, Berlin, Heidelberg},
author = {Cuccu, Giuseppe and Gomez, Faustino},
year = {2012},
langid = {english}
}
×
- Vincent Graziano, Tobias Glasmachers, Tom Schaul, Leo Pape, Giuseppe Cuccu, Jürgen Leitner, and Jürgen Schmidhuber. “Artificial Curiosity for Autonomous Space Exploration.” Acta Futura 4 (2011): 41–51. Bibtex PDF
Artificial Curiosity for Autonomous Space Exploration
Vincent Graziano, Tobias Glasmachers, Tom Schaul, Leo Pape, Giuseppe Cuccu, Jürgen Leitner, and Jürgen Schmidhuber. “Artificial Curiosity for Autonomous Space Exploration.” Acta Futura 4 (2011): 41–51.
@article{graziano2011acta,
title = {Artificial Curiosity for Autonomous Space Exploration},
volume = {4},
url = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.225.3031&rep=rep1&type=pdf},
pages = {41--51},
journaltitle = {Acta Futura},
author = {Graziano, Vincent and Glasmachers, Tobias and Schaul, Tom and Pape, Leo and Cuccu, Giuseppe and Leitner, Jürgen and Schmidhuber, Jürgen},
year = {2011}
}
×
- Leonardo Vanneschi, and Giuseppe Cuccu. “Reconstructing Dynamic Target Functions by Means of Genetic Programming Using Variable Population Size.” In Computational Intelligence, 343:121–34. Springer, Berlin, Heidelberg, 2011. Bibtex PDF
Reconstructing Dynamic Target Functions by Means of Genetic Programming Using Variable Population Size
Leonardo Vanneschi, and Giuseppe Cuccu. “Reconstructing Dynamic Target Functions by Means of Genetic Programming Using Variable Population Size.” In Computational Intelligence, 343:121–34. Springer, Berlin, Heidelberg, 2011.
@incollection{vanneschi2011computational,
title = {Reconstructing Dynamic Target Functions by Means of Genetic Programming Using Variable Population Size},
volume = {343},
url = {https://link.springer.com/chapter/10.1007/978-3-642-20206-3_8},
pages = {121--134},
booktitle = {Computational Intelligence},
publisher = {Springer, Berlin, Heidelberg},
author = {Vanneschi, Leonardo and Cuccu, Giuseppe},
year = {2011},
langid = {english},
doi = {10.1007/978-3-642-20206-3_8}
}
×
- Giuseppe Cuccu, Matthew Luciw, Jürgen Schmidhuber, and Faustino Gomez. “Intrinsically Motivated Neuroevolution for Vision-Based Reinforcement Learning.” In Development and Learning (ICDL), 2011 IEEE International Conference On, 2:1–7. IEEE, 2011. Bibtex PDF
Intrinsically Motivated Neuroevolution for Vision-based Reinforcement Learning
Giuseppe Cuccu, Matthew Luciw, Jürgen Schmidhuber, and Faustino Gomez. “Intrinsically Motivated Neuroevolution for Vision-Based Reinforcement Learning.” In Development and Learning (ICDL), 2011 IEEE International Conference On, 2:1–7. IEEE, 2011.
@inproceedings{cuccu2011icdl,
title = {Intrinsically Motivated Neuroevolution for Vision-based Reinforcement Learning},
volume = {2},
url = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.308.3917&rep=rep1&type=pdf},
pages = {1--7},
booktitle = {Development and Learning ({ICDL}), 2011 {IEEE} International Conference on},
publisher = {{IEEE}},
author = {Cuccu, Giuseppe and Luciw, Matthew and Schmidhuber, Jürgen and Gomez, Faustino},
year = {2011},
keywords = {artificial evolution, Computer vision, continuous reinforcement learning, embedded agents, high-dimensional visual images, intrinsically motivated neuroevolution, learning (artificial intelligence), neural networks, recurrent neural nets, recurrent neural network controllers, Tin, unsupervised sensory preprocessor, vision-based reinforcement learning, Yttrium}
}
×
- Giuseppe Cuccu, and Faustino Gomez. “When Novelty Is Not Enough.” In Applications of Evolutionary Computation, 234–43. Springer, Berlin, Heidelberg, 2011. Bibtex PDF
When Novelty Is Not Enough
Giuseppe Cuccu, and Faustino Gomez. “When Novelty Is Not Enough.” In Applications of Evolutionary Computation, 234–43. Springer, Berlin, Heidelberg, 2011.
@inproceedings{cuccu2011evostar,
title = {When Novelty Is Not Enough},
url = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.383.826&rep=rep1&type=pdf},
doi = {10.1007/978-3-642-20525-5_24},
eventtitle = {European Conference on the Applications of Evolutionary Computation},
pages = {234--243},
booktitle = {Applications of Evolutionary Computation},
publisher = {Springer, Berlin, Heidelberg},
author = {Cuccu, Giuseppe and Gomez, Faustino},
year = {2011},
langid = {english}
}
×
- Giuseppe Cuccu, Faustino Gomez, and Tobias Glasmachers. “Novelty-Based Restarts for Evolution Strategies.” In Evolutionary Computation (CEC), 2011 IEEE Congress On, 158–63. IEEE, 2011. Bibtex PDF
Novelty-based Restarts for Evolution Strategies
Giuseppe Cuccu, Faustino Gomez, and Tobias Glasmachers. “Novelty-Based Restarts for Evolution Strategies.” In Evolutionary Computation (CEC), 2011 IEEE Congress On, 158–63. IEEE, 2011.
@inproceedings{cuccu2011cec,
title = {Novelty-based Restarts for Evolution Strategies},
url = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.308.240&rep=rep1&type=pdf},
pages = {158--163},
booktitle = {Evolutionary Computation ({CEC}), 2011 {IEEE} Congress on},
publisher = {{IEEE}},
author = {Cuccu, Giuseppe and Gomez, Faustino and Glasmachers, Tobias},
year = {2011},
keywords = {black box optimization, black-box optimization, black box search algorithm, Convergence, Covariance matrix, Equations, evolutionary computation, evolution strategies, Monte Carlo methods, novelty-based restarts, novelty search, optimisation, Optimization, restart strategies, search problems, search space, Switches}
}
×
- Leonardo Vanneschi, and Giuseppe Cuccu. “Variable Size Population for Dynamic Optimization with Genetic Programming.” In Proceedings of the 11th Annual Conference on Genetic and Evolutionary Computation, 1895–96. GECCO ’09. ACM, 2009. Bibtex PDF
Variable Size Population for Dynamic Optimization with Genetic Programming
Leonardo Vanneschi, and Giuseppe Cuccu. “Variable Size Population for Dynamic Optimization with Genetic Programming.” In Proceedings of the 11th Annual Conference on Genetic and Evolutionary Computation, 1895–96. GECCO ’09. ACM, 2009.
@inproceedings{vanneschi2009gecco,
author = {Vanneschi, Leonardo and Cuccu, Giuseppe},
location = {New York, {NY}, {USA}},
title = {Variable Size Population for Dynamic Optimization with Genetic Programming},
isbn = {978-1-60558-325-9},
url = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.205.7910&rep=rep1&type=pdf},
doi = {10.1145/1569901.1570222},
series = {{GECCO} '09},
pages = {1895--1896},
booktitle = {Proceedings of the 11th Annual Conference on Genetic and Evolutionary Computation},
publisher = {{ACM}},
year = {2009},
keywords = {dynamic optimization, genetic algorithms, genetic programming, genetic programming, Poster, variable size populations}
}
×
- Leonardo Vanneschi, and Giuseppe Cuccu. “A Study of Genetic Programming Variable Population Size for Dynamic Optimization Problems.” In International Conference on Evolutionary Computation (ICEC), 119–26, 2009. Bibtex PDF
A Study of Genetic Programming Variable Population Size for Dynamic Optimization Problems
Leonardo Vanneschi, and Giuseppe Cuccu. “A Study of Genetic Programming Variable Population Size for Dynamic Optimization Problems.” In International Conference on Evolutionary Computation (ICEC), 119–26, 2009.
@inproceedings{vanneschi2009icec,
author = {Vanneschi, Leonardo and Cuccu, Giuseppe},
location = {Madeira, Portugal},
title = {A Study of Genetic Programming Variable Population Size for Dynamic Optimization Problems},
url = {http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.205.7483&rep=rep1&type=pdf},
pages = {119--126},
booktitle = {International Conference on Evolutionary Computation (ICEC)},
year = {2009},
keywords = {genetic algorithms, genetic programming}
}
×