2022
Huertas-García, Álvaro; Martín, Alejandro; Huertas-Tato, Javier; Camacho, David
Exploring Dimensionality Reduction Techniques in Multilingual Transformers Miscellaneous
CoRR, 2022.
Abstract | Links | BibTeX | Tags: Artificial Intelligence, Computational Intelligence, Deep Learning, Embeddings, Feature Selection, Machine Learning, Natural Language Processing, Sentence alignment, Text Analysis, Transformers
@misc{nokey,
title = {Exploring Dimensionality Reduction Techniques in Multilingual Transformers},
author = {Álvaro Huertas-García and Alejandro Martín and Javier Huertas-Tato and David Camacho},
url = {https://doi.org/10.48550/arxiv.2204.08415},
doi = {10.48550/ARXIV.2204.08415},
year = {2022},
date = {2022-04-18},
urldate = {2022-04-18},
abstract = {Both in scientific literature and in industry,, Semantic and context-aware Natural Language Processing-based solutions have been gaining importance in recent years. The possibilities and performance shown by these models when dealing with complex Language Understanding tasks is unquestionable, from conversational agents to the fight against disinformation in social networks. In addition, considerable attention is also being paid to developing multilingual models to tackle the language bottleneck. The growing need to provide more complex models implementing all these features has been accompanied by an increase in their size, without being conservative in the number of dimensions required. This paper aims to give a comprehensive account of the impact of a wide variety of dimensional reduction techniques on the performance of different state-of-the-art multilingual Siamese Transformers, including unsupervised dimensional reduction techniques such as linear and nonlinear feature extraction, feature selection, and manifold techniques. In order to evaluate the effects of these techniques, we considered the multilingual extended version of Semantic Textual Similarity Benchmark (mSTSb) and two different baseline approaches, one using the pre-trained version of several models and another using their fine-tuned STS version. The results evidence that it is possible to achieve an average reduction in the number of dimensions of 91.58%±2.59% and 54.65%±32.20%, respectively. This work has also considered the consequences of dimensionality reduction for visualization purposes. The results of this study will significantly contribute to the understanding of how different tuning approaches affect performance on semantic-aware tasks and how dimensional reduction techniques deal with the high-dimensional embeddings computed for the STS task and their potential for highly demanding NLP tasks },
howpublished = {CoRR},
keywords = {Artificial Intelligence, Computational Intelligence, Deep Learning, Embeddings, Feature Selection, Machine Learning, Natural Language Processing, Sentence alignment, Text Analysis, Transformers},
pubstate = {published},
tppubtype = {misc}
}
Both in scientific literature and in industry,, Semantic and context-aware Natural Language Processing-based solutions have been gaining importance in recent years. The possibilities and performance shown by these models when dealing with complex Language Understanding tasks is unquestionable, from conversational agents to the fight against disinformation in social networks. In addition, considerable attention is also being paid to developing multilingual models to tackle the language bottleneck. The growing need to provide more complex models implementing all these features has been accompanied by an increase in their size, without being conservative in the number of dimensions required. This paper aims to give a comprehensive account of the impact of a wide variety of dimensional reduction techniques on the performance of different state-of-the-art multilingual Siamese Transformers, including unsupervised dimensional reduction techniques such as linear and nonlinear feature extraction, feature selection, and manifold techniques. In order to evaluate the effects of these techniques, we considered the multilingual extended version of Semantic Textual Similarity Benchmark (mSTSb) and two different baseline approaches, one using the pre-trained version of several models and another using their fine-tuned STS version. The results evidence that it is possible to achieve an average reduction in the number of dimensions of 91.58%±2.59% and 54.65%±32.20%, respectively. This work has also considered the consequences of dimensionality reduction for visualization purposes. The results of this study will significantly contribute to the understanding of how different tuning approaches affect performance on semantic-aware tasks and how dimensional reduction techniques deal with the high-dimensional embeddings computed for the STS task and their potential for highly demanding NLP tasks
Fernandez-Mellado, Luis Sánchez; Stevenson, Emma; Rodriguez-Fernandez,; Vasile, Massimiliano; Camacho, David
An Intelligent System for Robust Decision-Making in the All-vs-All Conjunction Screening Problem Inproceedings
In: 3rd IAA Conference on Space Situational Awareness (ICSSA), Madrid, Spain, 2022.
Abstract | BibTeX | Tags: Artificial Intelligence, Collision Avoidance Manoeuvre, Conjunction Assessment, Robust Decision-making, Space Traffic Management
@inproceedings{stevenson2022_icssa,
title = {An Intelligent System for Robust Decision-Making in the All-vs-All Conjunction Screening Problem},
author = {Luis Sánchez Fernandez-Mellado and Emma Stevenson and Rodriguez-Fernandez and Massimiliano Vasile and David Camacho},
year = {2022},
date = {2022-04-01},
urldate = {2022-04-01},
booktitle = {3rd IAA Conference on Space Situational Awareness (ICSSA)},
address = {Madrid, Spain},
abstract = {The progressive increase of traffic in space demands new approaches for supporting automatic and robust operational decisions. CASSANDRA, Computational Agent for Space Situational Awareness aNd Debris Remediation Automation, is an intelligent system for Space Environment Management (SEM) intended to assist operators with the management of space traffic by providing robust decision-making support. This paper will present the automatic conjunction screening and collision avoidance manoeuvre pipeline within CASSANDRA, connecting the some of CASSANDRA's modules: Automated Conjunction Screening (ACS), Robust State Estimation (RSE), Intelligent Decision Support System (IDSS) and Collision Avoidance Manoeuvres (CAM). The pipelines allows to screen the catalogue to detect potential conjunctions, perform a detailed analysis of the encounter accounting for uncertainty (aleatory and epistemic) and new observations, provide robust decisions based on the available information and, if necessary, proposed robust optimal CAMs and analyse the impact of the new orbit on the background population. This paper will present the pipeline described above along with an example that illustrates how CASSANDRA can be used to generate robust decisions on the execution of CAMs in an automated way.},
keywords = {Artificial Intelligence, Collision Avoidance Manoeuvre, Conjunction Assessment, Robust Decision-making, Space Traffic Management},
pubstate = {published},
tppubtype = {inproceedings}
}
The progressive increase of traffic in space demands new approaches for supporting automatic and robust operational decisions. CASSANDRA, Computational Agent for Space Situational Awareness aNd Debris Remediation Automation, is an intelligent system for Space Environment Management (SEM) intended to assist operators with the management of space traffic by providing robust decision-making support. This paper will present the automatic conjunction screening and collision avoidance manoeuvre pipeline within CASSANDRA, connecting the some of CASSANDRA's modules: Automated Conjunction Screening (ACS), Robust State Estimation (RSE), Intelligent Decision Support System (IDSS) and Collision Avoidance Manoeuvres (CAM). The pipelines allows to screen the catalogue to detect potential conjunctions, perform a detailed analysis of the encounter accounting for uncertainty (aleatory and epistemic) and new observations, provide robust decisions based on the available information and, if necessary, proposed robust optimal CAMs and analyse the impact of the new orbit on the background population. This paper will present the pipeline described above along with an example that illustrates how CASSANDRA can be used to generate robust decisions on the execution of CAMs in an automated way.
Kim, Daria; Alber, Maximilian; Kwok, Man Wai; Mitrovic, Jelena; Ramirez-Atencia, Cristian; Pérez, Jesús Alberto Rodríguez; Zille, Heiner
Clarifying Assumptions About Artificial Intelligence Before Revolutionising Patent Law Journal Article
In: GRUR International, 2022, ISSN: 2632-8623, (ikab174).
Abstract | Links | BibTeX | Tags: Artificial Intelligence, IT Law
@article{10.1093/grurint/ikab174,
title = {Clarifying Assumptions About Artificial Intelligence Before Revolutionising Patent Law},
author = {Daria Kim and Maximilian Alber and Man Wai Kwok and Jelena Mitrovic and Cristian Ramirez-Atencia and Jesús Alberto Rodríguez Pérez and Heiner Zille},
url = {https://doi.org/10.1093/grurint/ikab174},
doi = {10.1093/grurint/ikab174},
issn = {2632-8623},
year = {2022},
date = {2022-01-01},
urldate = {2022-01-01},
journal = {GRUR International},
abstract = {This paper examines several widespread assumptions about artificial intelligence, particularly machine learning, that are often taken as factual premises in discussions on the future of patent law in the wake of ‘artificial ingenuity’. The objective is to draw a more realistic and nuanced picture of the human-computer interaction in solving technical problems than where ‘intelligent’ systems autonomously yield inventions. A detailed technical perspective is presented for each assumption, followed by a discussion of pertinent uncertainties for patent law. Overall, it is argued that implications of machine learning for the patent system in its core tenets appear far less revolutionary than is often posited.},
note = {ikab174},
keywords = {Artificial Intelligence, IT Law},
pubstate = {published},
tppubtype = {article}
}
This paper examines several widespread assumptions about artificial intelligence, particularly machine learning, that are often taken as factual premises in discussions on the future of patent law in the wake of ‘artificial ingenuity’. The objective is to draw a more realistic and nuanced picture of the human-computer interaction in solving technical problems than where ‘intelligent’ systems autonomously yield inventions. A detailed technical perspective is presented for each assumption, followed by a discussion of pertinent uncertainties for patent law. Overall, it is argued that implications of machine learning for the patent system in its core tenets appear far less revolutionary than is often posited.
2021
Stevenson, Emma; Rodriguez-Fernandez, Victor; Urrutxua, Hodei; Morand, Vincent; Camacho, David
Artificial Intelligence for All vs. All Conjunction Screening Inproceedings
In: 8th European Conference on Space Debris (ECSD), (Virtual), Darmstadt, Germany, 2021.
Abstract | Links | BibTeX | Tags: Artificial Intelligence, Conjunction Assessment, Space Debris
@inproceedings{stevenson2021_ecsd,
title = {Artificial Intelligence for All vs. All Conjunction Screening},
author = {Emma Stevenson and Victor Rodriguez-Fernandez and Hodei Urrutxua and Vincent Morand and David Camacho},
url = {http://oa.upm.es/67167/},
year = {2021},
date = {2021-04-01},
booktitle = {8th European Conference on Space Debris (ECSD)},
address = {(Virtual), Darmstadt, Germany},
abstract = {This paper presents a proof of concept for the application of artificial intelligence (AI) to the problem of efficient, catalogue-wide conjunction screening. Framed as a machine learning classification task, an ensemble of tabular models were trained and deployed on a realistic all vs. all dataset, generated using the CNES BAS3E space surveillance simulation framework, and consisting of 170 million object pairs over a 7-day screening period. The approach was found to outperform classical filters such as the apogee-perigee filter and the Minimum Orbital Intersection Distance (MOID) in terms of screening capability, with the number of missed detections of the approach controlled by the operator. It was also found to be computationally efficient, thus demonstrating the capability of AI algorithms to cope and aid with the scales required for current and future operational all vs. all scenarios.},
keywords = {Artificial Intelligence, Conjunction Assessment, Space Debris},
pubstate = {published},
tppubtype = {inproceedings}
}
This paper presents a proof of concept for the application of artificial intelligence (AI) to the problem of efficient, catalogue-wide conjunction screening. Framed as a machine learning classification task, an ensemble of tabular models were trained and deployed on a realistic all vs. all dataset, generated using the CNES BAS3E space surveillance simulation framework, and consisting of 170 million object pairs over a 7-day screening period. The approach was found to outperform classical filters such as the apogee-perigee filter and the Minimum Orbital Intersection Distance (MOID) in terms of screening capability, with the number of missed detections of the approach controlled by the operator. It was also found to be computationally efficient, thus demonstrating the capability of AI algorithms to cope and aid with the scales required for current and future operational all vs. all scenarios.
2018
Vasile, Massimiliano; Rodriguez-Fernandez, Victor; Serra, Romain; Camacho, David; Riccardi, Annalisa
Artificial intelligence in support to space traffic management Inproceedings
In: Proceedings of the International Astronautical Congress, IAC, pp. 3843–3856, International Astronautical Federation, Adelaide, Australia, 2018, ISBN: 978-1-5108-5537-3, (Publisher: International Astronautical Federation (IAF)).
Abstract | BibTeX | Tags: Artificial Intelligence, Space
@inproceedings{vasile_artificial_2018,
title = {Artificial intelligence in support to space traffic management},
author = {Massimiliano Vasile and Victor Rodriguez-Fernandez and Romain Serra and David Camacho and Annalisa Riccardi},
isbn = {978-1-5108-5537-3},
year = {2018},
date = {2018-01-01},
urldate = {2018-01-01},
booktitle = {Proceedings of the International Astronautical Congress, IAC},
volume = {1},
pages = {3843--3856},
publisher = {International Astronautical Federation},
address = {Adelaide, Australia},
abstract = {This paper presents an Artificial Intelligence-based decision support system to assist ground operators to plan and implement collision avoidance manoeuvres. When a new conjunction is expected, the system provides the operator with an optimal manoeuvre and an analysis of the possible outcomes. Machine learning techniques are combined with uncertainty quantification and orbital mechanics calculations to support an optimal and reliable management of space traffic. A dataset of collision avoidance manoeuvres has been created by simulating a range of scenarios in which optimal manoeuvres (in the sense of optimal control) are applied to reduce the collision probability between pairs of objects. The consequences of the execution of a manoeuvre are evaluated to assess its benefits against its cost. Consequences are quantified in terms of the need for additional manoeuvres to avoid subsequent collisions. By using this dataset, we train predictive models that forecast the risk of avoiding new collisions, and use them to recommend alternative manoeuvres that may be globally better for the space environment.},
note = {Publisher: International Astronautical Federation (IAF)},
keywords = {Artificial Intelligence, Space},
pubstate = {published},
tppubtype = {inproceedings}
}
This paper presents an Artificial Intelligence-based decision support system to assist ground operators to plan and implement collision avoidance manoeuvres. When a new conjunction is expected, the system provides the operator with an optimal manoeuvre and an analysis of the possible outcomes. Machine learning techniques are combined with uncertainty quantification and orbital mechanics calculations to support an optimal and reliable management of space traffic. A dataset of collision avoidance manoeuvres has been created by simulating a range of scenarios in which optimal manoeuvres (in the sense of optimal control) are applied to reduce the collision probability between pairs of objects. The consequences of the execution of a manoeuvre are evaluated to assess its benefits against its cost. Consequences are quantified in terms of the need for additional manoeuvres to avoid subsequent collisions. By using this dataset, we train predictive models that forecast the risk of avoiding new collisions, and use them to recommend alternative manoeuvres that may be globally better for the space environment.
2017
Ramírez-Atencia, Cristian; Rodríguez-Fernández, Víctor; Gonzalez-Pardo, Antonio; Camacho, David
New Artificial Intelligence approaches for future UAV Ground Control Stations Inproceedings
In: 2017 IEEE Congress on Evolutionary Computation, CEC 2017, Donostia, San Sebastián, Spain, June 5-8, 2017, pp. 2775–2782, IEEE, 2017, ISBN: 978-1-5090-4601-0.
Links | BibTeX | Tags: Artificial Intelligence, Unmanned Aircraft Systems
@inproceedings{DBLP:conf/cec/Ramirez-Atencia17,
title = {New Artificial Intelligence approaches for future UAV Ground Control Stations},
author = {Cristian Ramírez-Atencia and Víctor Rodríguez-Fernández and Antonio Gonzalez-Pardo and David Camacho},
url = {https://doi.org/10.1109/CEC.2017.7969645},
doi = {10.1109/CEC.2017.7969645},
isbn = {978-1-5090-4601-0},
year = {2017},
date = {2017-01-01},
urldate = {2017-01-01},
booktitle = {2017 IEEE Congress on Evolutionary Computation, CEC 2017, Donostia, San Sebastián, Spain, June 5-8, 2017},
pages = {2775--2782},
publisher = {IEEE},
keywords = {Artificial Intelligence, Unmanned Aircraft Systems},
pubstate = {published},
tppubtype = {inproceedings}
}
2016
Suárez, Óscar Manuel Losada; Rodríguez-Fernández, Víctor; Ramírez-Atencia, Cristian; Camacho, David
Desarrollo de una plataforma basada en Unity3D para la aplicación de IA en videojuegos Inproceedings
In: 3rd Congreso de la Sociedad Española para las Ciencias del Videojuego (CoSECiVi 2016), pp. 135–146, CEUR Workshop, Barcelona, Spain, 2016, ISSN: 16130073.
Abstract | Links | BibTeX | Tags: Artificial Intelligence, Multi-agents systems, Platform, Unity3D, Videogames
@inproceedings{LosadaSuarez2016,
title = {Desarrollo de una plataforma basada en Unity3D para la aplicación de IA en videojuegos},
author = {Óscar Manuel Losada Suárez and Víctor Rodríguez-Fernández and Cristian Ramírez-Atencia and David Camacho},
url = {https://aida.etsisi.upm.es/wp-content/uploads/2017/03/Desarrollo-de-una-plataforma-basada-en-Unity3D-para-la-aplicación-de-IA-en-videojuegos.pdf},
issn = {16130073},
year = {2016},
date = {2016-01-01},
urldate = {2016-01-01},
booktitle = {3rd Congreso de la Sociedad Española para las Ciencias del Videojuego (CoSECiVi 2016)},
volume = {1682},
pages = {135--146},
publisher = {CEUR Workshop},
address = {Barcelona, Spain},
abstract = {La utilización intensiva de diferentes técnicas relacionadas con la Inteligencia Artificial (IA) en el área de los videojuegos ha demostrado ser una necesidad para el campo. El uso de estas técnicas permite dotar de una mayor flexibilidad y adaptabilidad a los juegos que es muy apreciada por los jugadores. Temas como la generación procedimental de contenido, la creación de agentes que puedan jugar a un videojuego de forma competente, o de agentes cuya conducta sea indistinguible de la de un jugador humano atraen a una cantidad creciente de investigadores. El objetivo de este trabajo es la presentación de una plataforma basada en el motor Unity3D que permita de manera simple la integración y prueba de algoritmos de IA. La plataforma ofrecerá como nuevas características, adicionales a las ya disponibles en la actualidad, la utilización de un entorno 3D, el desarrollo de un juego innovador (basado en múltiples agentes), y la exploración de aspectos de juego como el análisis del terreno, la cooperación entre agentes independientes y heterogéneos, la comunicación de información entre los mismos y la formación de jerarquías.},
keywords = {Artificial Intelligence, Multi-agents systems, Platform, Unity3D, Videogames},
pubstate = {published},
tppubtype = {inproceedings}
}
La utilización intensiva de diferentes técnicas relacionadas con la Inteligencia Artificial (IA) en el área de los videojuegos ha demostrado ser una necesidad para el campo. El uso de estas técnicas permite dotar de una mayor flexibilidad y adaptabilidad a los juegos que es muy apreciada por los jugadores. Temas como la generación procedimental de contenido, la creación de agentes que puedan jugar a un videojuego de forma competente, o de agentes cuya conducta sea indistinguible de la de un jugador humano atraen a una cantidad creciente de investigadores. El objetivo de este trabajo es la presentación de una plataforma basada en el motor Unity3D que permita de manera simple la integración y prueba de algoritmos de IA. La plataforma ofrecerá como nuevas características, adicionales a las ya disponibles en la actualidad, la utilización de un entorno 3D, el desarrollo de un juego innovador (basado en múltiples agentes), y la exploración de aspectos de juego como el análisis del terreno, la cooperación entre agentes independientes y heterogéneos, la comunicación de información entre los mismos y la formación de jerarquías.