Non classé

Par : Christian Lebière

Résumé :
Cognitive architectures are computational implementations of unified theories of cognition. Being able to represent human cognition in computational form enables a wide range of applications when humans and machines interact. Using cognitive models to represent common ground between deep learners and human users enables adaptive explanations. Cognitive models representing the behavior of cyber attackers can be used to optimize cyber defenses including techniques such as deceptive signaling. Cognitive models of human-automation interaction can improve robustness of human-machine teams by predicting disruptions to measures of trust under various adversarial situations. Finally, the consensus of 50 years of research in cognitive architectures can be captured in the form of a Common Model of Cognition that can provide a guide for neuroscience, artificial intelligence and robotics.

Bio :
Christian Lebière is a Research Faculty member in the Psychology Department at Carnegie Mellon University. His main research interests are cognitive architectures and their applications to psychology, artificial intelligence, human-computer interaction, decision-making, intelligent agents, network science, cognitive robotics and neuromorphic engineering.
 
 
Références:
Cranford, E. A., Gonzalez, C., Aggarwal, P., Tambe, M., Cooney, S., & Lebiere, C. (2021). Towards a cognitive theory of cyber deception. Cognitive Science, 45(7), e13013.
Cranford, E., Gonzalez, C., Aggarwal, P., Cooney, S., Tambe, M., & Lebiere, C. (2020). Adaptive cyber deception: Cognitively informed signaling for cyber defense.
 
Lebiere, C., Blaha, L. M., Fallon, C. K., & Jefferson, B. (2021). Adaptive cognitive mechanisms to maintain calibrated trust and reliance in automation. Frontiers in Robotics and AI, 8, 652776.
 
Laird, J. E., Lebiere, C., & Rosenbloom, P. S. (2017). A standard model of the mind: Toward a common computational framework across artificial intelligence, cognitive science, neuroscience, and robotics. AI Magazine, 38(4), 13-26.
 
Lebiere, C., Pirolli, P., Thomson, R., Paik, J., Rutledge-Taylor, M., Staszewski, J., & Anderson, J. R. (2013). A functional model of sensemaking in a neurocognitive architecture. Computational Intelligence and Neuroscience, 2013.

Jeudi 20 octobre, 10 h 30. 

Lien zoom : ​https://uqam.zoom.us/j/88481835073

IMPORTANT : connectez-vous au moins 10 à 15 minutes à l’avance, et donnez votre nom complet pour nous faciliter la tâche de vous admettre au séminaire.
Lors de la période de question, nous vous invitons à ouvrir votre caméra et à poser vos questions directement.

Par : Agnieszka Wykowska

Résumé :
When predicting and explaining behavior of other humans, we adopt the intentional stance, and refer to mental states in order to understand others’ actions. However, it is not clear whether and when we adopt the intentional stance also towards artificial agents, such as humanoid robots. This talk will provide an overview of research conducted in my lab which addresses this question. I will present a tool for measuring adoption of the intentional stance. The likelihood of adopting the intentional stance is coded in specific patterns of neural activity at rest. Interactive scenarios influence adoption of the intentional stance more than mere observation of subtle human-like characteristics of a robot’s behavior. Experiments using interactive joint action protocols with a humanoid robot to study vicarious and joint sense of agency show that the robots’ motor repertoire and our ability to represent its actions with our own sensorimotor repertoire influence vicarious sense of agency. Embedding a non-verbal adaptation of a “Turing test” in a human-robot joint action task showed that human-like variability in the robot’s simple button presses makes the robot pass the test. The talk will conclude with a discussion of the role of the intentional stance and sense of agency in other mechanisms of social cognition, and their implications in applied domains of social robotics in healthcare.


Bio :
Professor Agnieszka Wykowska leads the unit “Social Cognition in Human-Robot Interaction” at the Italian Institute of Technology (Genoa, Italy). The research foci of Prof. Wykowska are interdisciplinary, bridging psychology, cognitive neuroscience, robotics and healthcare. She combines cognitive neuroscience methods with human-robot interaction to understand the human brain mechanisms in interaction with other humans and with robots. Her research is also dedicated to applications of social robotics to healthcare: her team develops robot-assisted training protocols to help children diagnosed with autism-spectrum disorder in improving social skills.

References:
Bossi, F., Willemse, C., Cavazza, J., Marchesi, S., Murino, V., Wykowska, A. (2020). The human brain reveals resting state activity patterns that are predictive of biases in attitudes towards robots. Science Robotics, 5:46, eabb6652: https://www.science.org/doi/10.1126/scirobotics.abb6652

Marchesi, S., De Tommaso, D., Perez-Osorio, J., Wykowska A. (2022). Belief in sharing the same phenomenological experience increases the likelihood of adopting the intentional stance towards a humanoid robot. Technology, Mind and Behavior, 3(3): https://www.apa.org/pubs/journals/releases/tmb-tmb0000072.pdf

Ciardo, F., De Tommaso, D., Wykowska, A. (2022). Human-like behavioural variability blues the distinction between a human and a machine in a nonverbal Turing test. Science Robotics, 7, eabo 1241: https://www.science.org/doi/10.1126/scirobotics.abo1241

Roselli, C., Ciardo, F., De Tommaso, D., Wykowska, A. (2022). Human‐likeness and attribution of intentionality predict vicarious sense of agency over humanoid robot actions. Nature: Scientific Reports, 12:13845: https://www.nature.com/articles/s41598-022-18151-6.pdf

Note: for the papers behind paywall, please visit our website, where you can find access links to all papers: https://instanceproject.eu/publications/list-of-publications

Jeudi 6 octobre, 10 h 30. 

Lien zoom : ​https://uqam.zoom.us/j/88481835073

IMPORTANT : connectez-vous au moins 10 à 15 minutes à l’avance, et donnez votre nom complet pour nous faciliter la tâche de vous admettre au séminaire.
Lors de la période de question, nous vous invitons à ouvrir votre caméra et à poser vos questions directement.

Par : Lorenzo Natale

Résumé :
Modern AI algorithms provide exceptional performance but require long training time and large datasets that are expensive to annotate. On the other hand, robots can actively interact with the environment and humans using their sensory system to learn on-line how to perceive and interact with objects. To extract structured information, however, the robot needs to be endowed with appropriate sensors, fast learning algorithms, and exploratory behavior that guide the interaction with the world.
In this talk I will introduce the sensory system we developed for the iCub humanoid robot, and in particular the tactile sensing technology. I will then review work in which we studied how to use visual and tactile feedback to explore unknown objects and to control the interaction between the hand and the objects for shape modelling, object discrimination and tracking. Finally, I will present recent work in which we developed fast learning algorithms for object segmentation that leverage on the interaction with a teacher and active learning for adaptation to new contexts.

Bio :
Lorenzo Natale, Senior Researcher at the Italian Institute of Technology and coordinator of the Center for Robotics and Intelligent Systems, was one of the main contributors to the design and development of the iCub humanoid robot. His research interests span artificial vision, tactile perception and software architectures for robotics.

References:
Ceola, F., Maiettini, E., Pasquale, G., Meanti, G., Rosasco, L., and Natale, L., Learn Fast, Segment Well: Fast Object Segmentation Learning on the iCub Robot, IEEE Transactions on robotics, 2022.
Maiettini, E., Tikhanoff, V., and Natale, L., Weakly-Supervised Object Detection Learning through Human-Robot Interaction, in Proc. International Conference on Humanoid Robotics, Munich, Germany, 2021
Vezzani, G., Pattacini, U., Battistelli, G., Chisci, L., and Natale, L., Memory Unscented Particle Filter for 6-DOF Tactile Localization, in IEEE Transactions on Robotics, vol. 33, no. 5, pp. 1139-1155, 2017

Jeudi 10 novembre, 10 h 30. 

Lien zoom : ​https://uqam.zoom.us/j/85407268175

IMPORTANT : connectez-vous au moins 10 à 15 minutes à l'avance. Lors de la période de question nous vous invitons à ouvrir votre caméra, et la laisser ouverte pendant toute la période de discussion.  Il est à noter que cette partie enregistrée ne sera pas gardée lors de l’édition du vidéo sur Youtube et sur le site du DIC.

Robert J. Vallerand, professeur au département de psychologie et membre de l'ISC, est le lauréat 2022 du Distinguished Lifetime Career Award, décerné par l’International Society for Self and Identity (ISSI). Lisez l'article paru dans l'Actualité UQAM.

Bonjour,

C'est avec plaisir que nous vous invitons à la première édition du colloque interdisciplinaire sur la douleur et la souffrance. L'événement se tiendra en ligne la fin de semaine des 17 et 18 septembre.
Horaire et résumés des présentations : https://docs.google.com/document/d/1JPFVXp9rXOxojDi2OyPtifkCj5ImnppU2kJQYPlKy0g/edit?usp=sharing

Afin de recevoir le lien zoom pour participer à l'événement, veuillez remplir le court formulaire d'inscription à l'adresse suivante : https://forms.gle/kysD14BhTaMJjYjy5

Bien cordialement,

Frédérick et David

Frédérick Deschênes
Candidat à la maîtrise en Philosophie
UQAM

David Lavoie
Candidat au doctorat en Psychologie
UQAM

Affiche téléchargeable

Par : Murray Shanahan

Résumé :
Common sense in humans is founded on a set of basic capacities that are possessed by many other animals, capacities pertaining to the understanding of objects, space, and causality. The field of animal cognition has developed numerous experimental protocols for studying these capacities and, thanks to progress in deep reinforcement learning (RL), it is now possible to apply these methods directly to evaluate RL agents in 3D environments. The Animal-AI Environment aims to apply the ability-oriented testing used in comparative psychology to AI systems. Besides evaluation, the animal cognition literature offers a rich source of behavioural data, which can serve as inspiration for RL tasks and curricula.

Bio :
Murray Shanahan is Professor of Cognitive Robotics at Imperial College London and Senior Research Scientist at DeepMind. His publications span artificial intelligence, robotics, logic, dynamical systems, computational neuroscience, and philosophy of mind. His work up to 2000 was in the tradition of classical, symbolic AI. He then turned his attention to the brain and its embodiment. His current interests include neurodynamics, consciousness, machine learning, and the impacts of artificial intelligence.

References:
Shanahan, M., Crosby, M., Beyret, B., & Cheke, L. (2020). Artificial intelligence and the common sense of animals. Trends in cognitive sciences, 24(11), 862-872.
Voudouris, K., Crosby, M., Beyret, B., Hernández-Orallo, J., Shanahan, M., Halina, M., & Cheke, L. G. (2022). Direct Human-AI Comparison in the Animal-AI Environment. Frontiers in Psychology, 1884.
Shanahan, M., & Mitchell, M. (2022). Abstraction for Deep Reinforcement Learning. arXiv preprint arXiv:2202.05839.
Shanahan, M., Embodiment and the Inner Life: Cognition and Consciousness in the Space of Possible Minds, Oxford University Press (2010). Full text

Jeudi 29 septembre, 10 h 30. 

Lien zoom : ​https://uqam.zoom.us/j/88481835073

IMPORTANT : connectez-vous au moins 10 à 15 minutes à l’avance, et donnez votre nom complet pour nous faciliter la tâche de vous admettre au séminaire.
Lors de la période de question, nous vous invitons à ouvrir votre caméra et à poser vos questions directement.

Par : Mehdi Khamassi

Résumé :
One of the key ingredients of learning in autonomous agents in volatile environments is the exploration-exploitation trade-off: finding the right balance between exploiting previously acquired knowledge and exploring alternatives, and adapting this balance on-the-fly when the environment changes. Throughout the presentation, I will use an illustrative example from the human–robot interaction (HRI) domain: Among relevant signals, non-verbal cues such as the human’s gaze can provide the robot with important information about the human’s current engagement in the task, and whether the robot should continue its current behavior or not. Various solutions have been proposed in the reinforcement learning literature, often inspired by developmental psychology (studying how human infants explore their surrounding world). Some mechanisms have neurobiological counterparts in the human brain: dynamic regulations of exploration rate as a function of volatility; information (uncertainty)-based solutions; and progress-based solutions. I will also illustrate existing bridges with Karl Friston's active inference which he will later present in this seminar series.

Bio :
Mehdi Khamassi is a CNRS research director, Institute of Intelligent Systems and Robotics (ISIR), Sorbonne Université, Paris. His background is in Computer Science, Cognitive Sciences and Cognitive Neuroscience. He is co-director of studies of the CogMaster program at Ecole Normale Supérieure (PSL) / EHESS / University of Paris  and Editor of the several scientific journals, like Intellectica, Frontiers in Neurorobotics, Frontiers in Decision Neuroscience, ReScience X, and Neurons, Behavior, Data analysis and Theory. His main topics of research include decision-making and reinforcement learning in robots and humans, and the role of social and non-social rewards in learning.

Références :
https://hal.archives-ouvertes.fr/hal-03415847/document
http://sites.isir.upmc.fr/www/files/2018ACLI4582.pdf
http://sites.isir.upmc.fr/www/files/2018ACLI4600.pdf

Jeudi 22 septembre, 10 h 30. 

Lien zoom  : ​https://uqam.zoom.us/j/88481835073

IMPORTANT : connectez-vous au moins 10 à 15 minutes à l’avance, et donnez votre nom complet pour nous faciliter la tâche de vous admettre au séminaire.
Lors de la période de question, nous vous invitons à ouvrir votre caméra et à poser vos questions directement.

Par : Jean-Pierre Briot

Résumé :
A growing application area for the current wave of deep learning (the return of artificial neural networks on steroids) is the generation of creative content, notably the case of music (and also images and text). The motivation is in using machine learning techniques to automatically learn musical styles from arbitrary musical corpora and then to generate musical samples from the estimated distribution, with some degree of control over the generation. This talk will survey some recent achievements in deep-learning-based music generation, using recent and dedicated generative architectures such as VAE, GAN and Transformer, analyzing principles, successes as well as challenges, including the limits of automated generation versus providing assistance to human musicians.

Bio :
Jean-Pierre Briot is a senior researcher (research director) in computer science at LIP6, joint computer science research lab of CNRS (Centre National de la Recherche Scientifique) and Sorbonne Université in Paris, France. He is also permanent visiting professor at PUC-Rio in Rio de Janeiro, Brazil. His general research interests are the design of intelligent adaptive and cooperative software, at the crossroads of artificial intelligence, distributed systems and software engineering, with various applications in the internet of things, decision support systems and computer music. His current interest is the use of AI techniques (notably deep learning-based) within music creation processes. He is the principal author of a recent reference book on deep learning techniques for music generation (Springer, 2020). https://link.springer.com/book/10.1007/978-3-319-70163-9

For more details (including access to publications): http://webia.lip6.fr/~briot/cv/
Briot, J. P. (2021). From artificial neural networks to deep learning for music generation: history, concepts and trends. Neural Computing and Applications, 33(1), 39-65.
https://hal.sorbonne-universite.fr/hal-02539189v3/file/nn4music-hal-v3.pdf  

Briot, J. P. (2019). Apprentissage profond et génération de musique, Hors série Intelligence artificielle, Tangente - L'aventure mathématique, (68):30-37, September 2019.
https://webia.lip6.fr/~briot/cv/apgm-2019

Jeudi 15 septembre, 10 h 30. 

Lien zoom  : ​https://uqam.zoom.us/j/88481835073

IMPORTANT : connectez-vous au moins 10 à 15 minutes à l’avance, et donnez votre nom complet pour nous faciliter la tâche de vous admettre au séminaire.
Lors de la période de question, nous vous invitons à ouvrir votre caméra et à poser vos questions directement.

Qu’est-ce que l’intelligence artificielle? Élisabeth Doyon consacre ses recherches à cette question pas si simple.

Qu’entendons-nous par «intelligence»? Dans le cadre de sa maîtrise en éducation et pédagogie, Élisabeth Doyon s’est attaquée à cette question. Son mémoire portait sur l’impact des diverses représentations de l’intelligence dans le milieu de l’enseignement. «Le terme "intelligence" pose déjà un problème de communication. Avec l’intelligence artificielle (IA), le problème se complexifie», observe la doctorante en informatique cognitive, qui tente aujourd’hui de mieux cerner ce concept devenu omniprésent dans le discours public. 

L’originalité de sa démarche? Élisabeth Doyon étudie les représentations de l’intelligence artificielle au sein d’une équipe d’Ubisoft/La Forge chargée de développer un outil de vulgarisation de ce concept. Mise sur pied par Ubisoft, La Forge est une plateforme de transfert technologique rassemblant des chercheurs universitaires et des employés d’Ubisoft qui a pour mission de créer des prototypes basés sur les plus récentes avancées académiques. Concrètement, l’équipe a conçu un jeu destiné aux jeunes de 6 à 12 ans pour le Centre des sciences de Montréal.  Suite de l'entrevue paru dans L'Actualité UQAM du 11 mars.

Élisabeth Doyon est membre junior de l'ISC et a obtenu, avec deux autres membres juniors, une bourse de valorisation de la recherche de l'ISC pour appliquer ses méthodes en sciences cognitives du raisonnement à un projet sur les Annales de l’Acfas.

Deux membres de l'ISC sont récipiendaires d'un prix facultaire 2021 de la Faculté des sciences humaines de l’UQAM : Le professeur Frédérick L. Philippe, du département de psychologie, est récipiendaire dans la catégorie Prix d’excellence en enseignement – Relève, et le professeur Jean-François Gagnon, aussi du département de psychologie, est récipiendaire dans la catégorie Prix d’excellence en recherche – Carrière.

Toutes nos félicitations !