El aprendizaje en sistemas autónomos e inteligentes: visión general y sesgos de fuentes de datos

Pablo  Jiménez Schlegl

doi:10.3989/arbor.2021.802005

Autores/as

Pablo Jiménez Schlegl Consejo Superior de Investigaciones Científicas - Universitat Politècnica de Catalunya https://orcid.org/0000-0003-3627-4938

DOI:

https://doi.org/10.3989/arbor.2021.802005

Palabras clave:

Sistemas autónomos e inteligentes, métodos de aprendizaje automático, sesgo en fuentes de datos

Resumen

Los sistemas autónomos e inteligentes (A/IS por sus siglas en inglés, en concordancia con el informe del IEEE sobre diseño alineado con la ética) pueden obtener sus conocimientos a través de diferentes procedimientos y de fuentes diversas. Los algoritmos de aprendizaje son neutros en principio, son más bien los datos con los que se alimentan durante el período de aprendizaje que pueden introducir sesgos o una orientación ética específica. El control humano sobre el proceso de aprendizaje es más directo en aprendizaje por demostración, donde las fuentes de datos están restringidas a las elecciones del demostrador (o profesor), pero incluso en las versiones no supervisadas del aprendizaje por refuerzo los sesgos están presentes a través de la definición de la función de recompensa. En este artículo proporcionamos una visión general de los paradigmas de aprendizaje de los sistemas artificiales: métodos supervisados y no supervisados, con los ejemplos más destacados de cada categoría, sin profundizar demasiado en el detalle técnico. Además describimos los tipos de fuentes de datos disponibles actualmente y su uso por la comunidad robótica. También enfatizamos el sesgo que se observa en bases de datos de imágenes y originados por anotación humana. Destacamos una investigación muy reciente sobre sesgo en navegación de robots sociales y finalizamos con una breve reflexión sobre influencia del ambiente sobre futuros robots que aprenden.

Descargas

Los datos de descargas todavía no están disponibles.

Citas

Australian Centre for Field Robotics, ACFR. Marine Robotics Datasets. Available at http://marine.acfr.usyd.edu.au/datasets/

Berkeley Artificial Intelligence Research, BAIR. RoboNet: A Dataset for Large-Scale Multi-Robot Learning. Available at: https://bair.berkeley.edu/blog/2019/11/26/robo-net/

Buchanan, Bruce G. (1989). Can Machine Learning Offer Anything to Expert Systems? Machine-Learning, 4: 251-254. https://doi.org/10.1007/BF00130712

Choi, Sunlok. The Awesome Robotics Datasets. Available at: https://github.com/sunglok/awesome-robotics-datasets

Crawford, Kate and Paglen, Trevor. Excavating AI: The Politics of Images in Machine Learning Training Sets. Available at: https://excavating.ai

Everingham, Mark; Van Gool, Luc; Williams, Christopher K.I.; Winn, John and Zisserman, Andrew (2010). The pascal visual object classes (voc) challenge. International Journal of Computer Vision. 88 (2): 303-338. https://doi.org/10.1007/s11263-009-0275-4

Ford Center for Autonomous Vehicles (FCAV), University of Michigan. FCAV M-Air Pedestrian (FMP) Dataset of monocular RGB images and Planar LiDAR data for pedestrian detection. Available at: https://github.com/umautobots/FMP-dataset [updated November 10th 2019; cited January, 15th 2021]

Geiger, Andreas; Lenz, Philip; Stiller, Christoph and Urtasun, Raquel (2013). Vision meets Robotics: The KITTI Dataset. International Journal of Robotics Research. 32 (11): 1231-1237. https://doi.org/10.1177/0278364913491297

Hervé, Nicolas and Boujemaa, Nozha (2007). Image annotation: which approach for realistic databases? In: CIVR '07: Proceedings of the 6th ACM international conference on Image and video retrieval, Amsterdam, The Netherlands, July 2007. New York, NY, USA: Association for Computing Machinery, pp. 170-177. https://doi.org/10.1145/1282280.1282310

Huang, Yongkiang and Sun, Yu (2019) A dataset of daily interactive manipulation. The International Journal of Robotics Research. 38(8):879-886. https://doi.org/10.1177/0278364919849091

Hurtado, Juana Valeria; Londoño, Laura, and Valada, Abhinav (2021), From Learning to Relearning: A Framework for Diminishing Bias in Social Robot Navigation. Frontiers in Robotics and AI, 8: 650325. https://doi.org/10.3389/frobt.2021.650325 PMid:33842558 PMCid:PMC8024571

ImageNet, Stanford Vision Lab, Stanford University, Princeton University. Available at: https://www.image-net.org/index.php

Inoue, Masashi (2004). On the need for annotation-based image retrieval. In: Proceedings of the ACM SIGIR Workshop on Information Retrieval in Context (IRiX), Sheffield, UK, July 29th 2004. Department of Information Studies, Royal School of Library and Information Science Copenhagen, Denmark: pp. 44-46

Johnson-Roberson, Matthew; Barto, Charles; Mehta, Rounak; Sridhar, Sarath Nittur; Rosaen, Karl and Vasudevan, Ram (2017). Driving in the matrix: Can virtual worlds replace human- generated annotations for real world tasks? In: Proceedings of the IEEE International Conference on Robotics and Automation, 29 May-3 June 2017, Singapore. IEEE: pp. 746-753. https://doi.org/10.1109/ICRA.2017.7989092

Kümmerer, Matthias; Bylinskii, Zoya; Judd, Tilke; Borji, Ali; Itti, Laurent; Durand, Frédo; Oliva, Aude and Torralba, Antonio. MIT/Tübingen Saliency Benchmark. Available at: https://saliency.tuebingen.ai/ [updated 2018; cited January, 16th 2021]

Levine, Sergey. Google Brain Robotics Data. Available at: https://sites.google.com/site/brainrobotdata/home, [updated August 5th 2016; cited January, 15th 2021]

Lim, Hengtee (2020) 18 Best Datasets for Machine Learning Robotics. Available at: https://lionbridge.ai/datasets/17-best-robotics-datasets-for-machine-learning/

Locolab, University of Michigan. The Effect of Walking Incline and Speed on Human Leg Kinematics, Kinetics, and EMG. Available at: https://ieee-dataport.org/open-access/effect-walking-incline-and-speed-human-leg-kinematics-kinetics-and-emg

Lyons, Michael (2020). Excavating "Excavating AI": The Elephant in the Gallery. Submitted: https://arxiv.org/abs/2009.01215 https://doi.org/10.2139/ssrn.3901640

Mandlekar, Ajay; Booher, Jonathan; Spero, Max; Tung, Albert; Gupta, Anchit; Zhu, Yuke; Garg, Animesh; Savarese, Silvio and Fei-Fei, Li (2019). Scaling Robot Supervision to Hundreds of Hours with RoboTurk: Robotic Manipulation Dataset through Human Reasoning and Dexterity. IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Macau, China, 3-8 November 2019. IEEE, pp. 1048-1055. https://doi.org/10.1109/IROS40897.2019.8968114

Martínez, David; Alenyà, Guillem and Torras, Carme (2017). Artificial Intelligence. 247: 295-312. https://doi.org/10.1016/j.artint.2015.02.006

Martínez Mozos, Oscar; Nakashima, Kazuto; Jung, Hojung; Iwashita, Yumi, and Kurazume, Ryo (2019). Fukuoka datasets for place categorization. International Journal of Robotics Research, 38 (5): 507-517. https://doi.org/10.1177/0278364919835603

MultiDrone EU Project Dataset. Available at: https://multidrone.eu/multidrone-public-dataset/ [updated January 23rd 2020; cited January, 15th 2021]

Nehaniv, Chrystopher and Dautenhahn, Kerstin (2002). The Correspondence Problem. In: K. Dautenhahn and C. L. Nehaniv (eds.) Imitation in Animals and Artifacts. Cambridge, MA, USA: MIT Press, pp. 41-61,

Pandey, Gaurav; McBride James R. and Eustice, Ryan M. (2011). Ford Campus vision and lidar data set. The International Journal of Robotics Research. 30 (13): 1543-1552. https://doi.org/10.1177/0278364911400640

Ramisa, Arnau; Yan, Fei; Moreno-Noguer Francesc and Mikolajczyk Krystian (2018). BreakingNews: Article annotation by image and text processing. IEEE Transactions on Pattern Analysis and Machine Intelligence, 40 (5): 1072-1085. https://doi.org/10.1109/TPAMI.2017.2721945 PMid:28682246

Ruiz-Sarmiento, Jose Raul; Galindo, Cipriano and Gonzalez-Jimenez, Javier. (2017) Robot@Home, a robotic dataset for semantic mapping of home environments. The International Journal of Robotics Research, 36 (2): 131-141. https://doi.org/10.1177/0278364917695640

Torralba, Antonio and Efros, Alexei A. (2011) Unbiased look at dataset bias. In IEEE Conference on Computer Vision and Pattern Recognition, Colorado Springs, 20-25 June, 2011. IEEE, pp. 1521-1528. https://doi.org/10.1109/CVPR.2011.5995347

von Ahn, Luis; Liu, Ruoran and Blum, Manuel (2006). Peekaboom: a game for locating objects in images. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '06). New York, (USA): Association for Computing Machinery, pp. 55-64. https://doi.org/10.1145/1124772.1124782

Wang, John and Olson, Edwin (2016) AprilTag 2: Efficient and robust fiducial detection. Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Daejon, Korea, 9-14 October, 2016. IEEE, pp. 4193-4198. https://doi.org/10.1109/IROS.2016.7759617

Xiao, Jianxiong; Hays, James; Ehinger, Krista A; Oliva, Aude and Torralba, Antonio (2010). Sun database: Large-scale scene recognition from abbey to zoo. In IEEE Conference on Computer Vision and Pattern Recognition, San Francisco, 13-18 June, 2010. IEEE, pp. 3485-3492. https://doi.org/10.1109/CVPR.2010.5539970