Hugo Cisneros

Machine Leaning Engineer

Inicio
33 rue Faubourg Saint-Antoine
75011 Paris, France
[email protected]

Short Bio

I am a Machine Learning engineer at Inicio. Incio is a company identifying high-potential land for the development of photovoltaic projects. My job is to leverage ML algorithms to extract data from various sources. I also work a lot with GIS tools.

Previously, I was a PhD student at the Czech Institute for Informatics, Robotics and Cybernetics (CIIRC, CTU in Prague), part of the Impact Project, and the Willow team, of INRIA and ENS. I conducted my PhD under the supervision of Tomas Mikolov (CIIRC, CTU in Prague) and Josef Sivic (CIIRC, Inria and ENS Paris). Before that, I got a Master’s degree in Mathematics, Vision, and Machine learning from École Normale Supérieure Paris-Saclay, and a Master’s degree in Engineering with a major in Computer Science at Mines ParisTech.

My research focused on the development of unsupervised machine learning algorithms capable of innovation and adaptation. I studied complex systems like cellular automata and open-ended evolution.

I am generally interested in the emergence and growth of complexity in dynamical systems and how this could lead to open-ended evolving systems. During my PhD, I have studied cellular automata and their continuous extensions as they are general and simple yet capable of surprisingly complex emergent behaviors.

News

• August 2023 - New paper Preserving Semantics in Textual Adversarial Attacks accepted at ECAI 2023.

• May 2023 - I successfully defended my thesis! Read the manuscript on HAL.

• March 2023 - Started working as a Machine Learning engineer at Inicio.

• August 2022 - Poster presentation at the 1st conference on Lifelong Learning Agents (CoLLAs 2022).

• August 2021 - Runner-up of the Minecraft Open-Endedness Challenge, the first open-ended evolution challenge organized at the GECCO 2021 conference competition track (See the blog post, the video and the code)

• July 2020 – Talk at Alife 2020, Montréal, Canada (virtual) for Visualizing computation in large-scale cellular automata (Video of the talk)

• December 2019 – Talk at IEEE Alife, SSCI 2019 in Xiamen, China – Best Student paper award for Evolving Structures in Complex Systems

Publications

D. Herel, H. Cisneros, T. Mikolov
Preserving Semantics in Textual Adversarial Attacks

In ECAI 2023, 2023.

A simple sentence encoder based on multiple words level embeddings to improve the quality of textual adversarial attacks. Improves the quality of attacks compared to other semantic similarity metrics such as Universal Sentence Encoder or BERTScore.

CiteCodearXiv

H. Cisneros, J. Sivic, T. Mikolov
Benchmarking Learning Efficiency in Deep Reservoir Computing

In Proceeding of the 1st Conference on Lifelong Learning Agents (CoLLAs), 2022.

Defines a metric of learning efficiency for machine learning models and a benchmark of progressively more difficult language tasks for measuring learning efficiency.

CiteMain URLCodearXivDatasetVideo

H. Cisneros, J. Sivic, T. Mikolov
Visualizing computation in large-scale cellular automata

In Artificial Life Conference Proceedings, 2020.

Introduces novel reduction methods for processing large-scale complex systems. This enables discovering interesting patterns at multiple scales.

CiteMain URLProject pagearXivTalk

H. Cisneros, J. Sivic, T. Mikolov
Evolving Structures in Complex Systems

In Proceedings of the 2019 IEEE Symposium Series on Computational Intelligence, SSCI, 2019.

Introduces a new metric of increasing complexity for cellular automata and other systems.

CiteMain URLarXivCodeSlides

@misc{herelPreservingSemanticsTextual2022,
 title = {Preserving {{Semantics}} in {{Textual Adversarial Attacks}}},
 author = {Herel, David and Cisneros, Hugo and Mikolov, Tomas},
 date = {2022-11-08},
 number = {arXiv:2211.04205},
 eprint = {2211.04205},
 eprinttype = {arxiv},
 primaryclass = {cs},
 publisher = {{arXiv}},
 doi = {10.48550/arXiv.2211.04205},
 url = {http://arxiv.org/abs/2211.04205},
 urldate = {2022-11-21},
 archiveprefix = {arXiv},
 version = {1},
 keywords = {Computer Science - Artificial Intelligence,Computer Science - Computation and Language},
 }

@inproceedings{cisnerosBenchmarkingLearningEfficiency2022a,
 title = {Benchmarking {{Learning Efficiency}} in {{Deep Reservoir Computing}}},
 booktitle = {Proceedings of {{The}} 1st {{Conference}} on {{Lifelong Learning Agents}}},
 author = {Cisneros, Hugo and Mikolov, Tomas and Sivic, Josef},
 date = {2022-11-28},
 pages = {532--547},
 publisher = {{PMLR}},
 issn = {2640-3498},
 url = {https://proceedings.mlr.press/v199/cisneros22a.html},
 urldate = {2023-01-17},
 eventtitle = {Conference on {{Lifelong Learning Agents}}},
 langid = {english},
 }

@article{doi:10.1162/isal_a_00277,
 author = {Cisneros, Hugo and Sivic, Josef and Mikolov, Tomas},
 title = {Visualizing computation in large-scale cellular automata},
 journal = {Artificial Life Conference Proceedings},
 volume = {},
 number = {32},
 pages = {239-247},
 year = {2020},
 doi = {10.1162/isal_a_00277},
 URL = {https://www.mitpressjournals.org/doi/abs/10.1162/isal_a_00277},
 eprint = {https://www.mitpressjournals.org/doi/pdf/10.1162/isal_a_00277} ,
 }

@inproceedings{9002840,
 author={H. {Cisneros} and J. {Sivic} and T. {Mikolov}},
 booktitle={2019 IEEE Symposium Series on Computational Intelligence (SSCI)},
 title={Evolving Structures in Complex Systems},
 year={2019},
 volume={},
 number={},
 pages={230-237},
 }