Prise de décision collective

La prise de décision collective (en anglais, collaborative decision making ou CDM) est une situation où des individus sont rassemblés en un groupe pour résoudre des problèmes. Selon l'idée de synergie, les décisions prises collectivement ont tendance à être plus efficaces que les décisions prises individuellement. Cependant, il existe des situations dans lesquelles les décisions prises en groupe aboutissent à un mauvais jugement. En psychologie sociale, la prise de décision collective peut être définie comme . La prise de décision collective est un domaine d'étude vaste auquel plusieurs disciplines s'intéressent, comme les sciences sociales, les sciences politiques, l'informatique ; on s'y intéresse également en marketing et en management, chacun de ces champs d'étude ayant son point de vue sur la recherche de la prise de décision collective. Du point de vue de la psychologie sociale plus spécifiquement, des applications et des conséquences théoriques sont nom

Surprise-based model estimation in reinforcement learning: algorithms and brain signatures

Vasiliki Liakoni

Learning how to act and adapting to unexpected changes are remarkable capabilities of humans and other animals. In the absence of a direct recipe to follow in life, behaviour is often guided by rewarding and by surprising events. A positive or a negative outcome influences the tendency to repeat some actions, and a sudden unexpected event signals the possible need to act differently or to update one's view about the world. Advances in computational, behavioral, and cognitive neuroscience have indicated that animals employ multiple strategies to learn from interaction. However, our understanding of learning strategies and how they are combined is largely restricted. The main goal of this thesis is to study the use of surprise by ever-adapting biological agents, its contributions to reward-based learning, and its manifestation in the human brain. We first study surprise from a theoretical perspective. In a probabilistic model of changing environments, we show that exact and approximate Bayesian inference give rise to a trade-off between forgetting old observations and integrating them with new ones, modulated by a naturally emerging surprise measure. We develop novel surprised-based algorithms that can adapt in the face of abrupt changes and accurately estimate the model of the world, and that could potentially be implemented in the brain. Next, we focus on the contributions of surprise-based model estimation to reinforcement learning. We couple one of our adaptive algorithms as well as simpler non-adaptive methods with reinforcement learning agents and evaluate their performance on environments exhibiting different characteristics. Abrupt changes that directly affect the agent's policy call for surprise-based adaptation, in order to achieve higher performance. Often, however, the agent does not need to invest in maintaining an accurate model of the environment to obtain high reward levels. More specifically, in stochastic environments or in environments with distal changes, simpler methods, equipped with exploration capacity, perform equally well compared to more elaborate methods. Finally, we turn to human learning behaviour and brain signals of surprise- and reward-based learning. We design a novel sequential decision making task of multiple steps where strategic use of surprising events allows us to dissociate fMRI brain correlates of reward learning and model estimation. We show that Bayesian inference on this task leads to the same surprise measure we found earlier, where the trade-off is now between ignoring new observations and integrating them with the old belief, and we develop reinforcement learning algorithms that perform outlier detection via this surprise-modulated trade-off. At the level of behaviour we find evidence for a model-free policy learning architecture, with potential influences from a model estimation system. At the level of brain responses we identify signatures of both reward- and model estimation signals, supporting the existence of multiple parallel learning systems in the brain. This thesis presents a comparative analysis of surprise-based model estimation methods in theory and simulations, provides insights in the type of approximations that biological agents may adopt, and identifies signatures of model estimation in the human brain. Our results may aid future work aiming at building efficient adaptive agents and at understanding the learning algorithms and the surprise measures implemented in the brain.

EPFL2021

Elicitation and Modeling of Operational Human Decision Making in a Supply Chain Context

Souleïman Naciri

Supply chains (SC) are complex systems in which human beings play a key role. Whereas human impact is often taken into account in simulation models in terms of physical flows (material handling, fabrication), it is far less the case in terms of information and decision flows. This can be explained by the intrinsic characteristics of human decisions: variability across people, bounded rationality, intuition, variability in data consultation and interpretation. This complexity explains why human decision making is either eluded in simulation models or replaced by algorithms that systematically optimize the decisions to be made. The goal of this research is to propose a methodology that allows modeling human decision making in a procurement context in a descriptive way. By descriptive modeling it is meant modeling of the decisions as they are rather than as they should be. To do so a participatory simulation platform is developed. This platform's concept is to create a virtual decision environment within which a human agent can consult several information about the system state (inventory levels, supplier reliability, MRP, etc.) and can make decisions that he/she feels relevant (order anticipation, postponement, validation, order grouping). The participatory simulation platform is then used within an academic application case involving one student, and an industrial application case involving 14 procurement agents from the Swiss industry field. The modeling phase then consists in selecting and testing the suitability of several modeling paradigms such as decision trees, connectionist networks, discrete choice and clustering analysis. The obtained results lead to interesting conclusions concerning the advantages and limitations of each of these three modeling approaches. It is in particular shown that at an individual level and for simplified decision making situations, decision trees and ARAM (Adaptive Resonance Associative Map) networks perform in a similar way in terms of classification performance, while ARAM networks outperform decision trees in terms of cognitive relevance as they generate rules that better fit the human decision making pattern as outlined by post-experimentation interviews. At a sample-based modeling level, (the whole set of experiments being considered), discrete choice analysis provides interesting results for the understanding of the impact of consulted data over the decisions made by procurement agents as well as the identification of counterintuitive relations that deserve further refinement and investigation. Finally, a clustering of human decision making behaviors based on a set of proposed metrics allows to identify several behavioral profiles that represent the behavior of sub-sets of people who share similar decision making behavior characteristics. Finally a global hierarchical modeling methodology taking benefits from the above mentioned modeling methods is proposed to build human decision making models with minimum modeling efforts. The developed data collection technique, the proposed methodology and modeling tools constitute a significant leap towards the integration of human decision making behaviors in supply chain simulation models in order to make them more representative and reliable.

EPFL2010

Surprise-based model estimation in reinforcement learning: algorithms and brain signatures

Vasiliki Liakoni

EPFL2021

Elicitation and Modeling of Operational Human Decision Making in a Supply Chain Context

Souleïman Naciri

EPFL2010

Public-Private decision-making in critical infrastructures, case of High-speed ground transportation systems

Surprise-based model estimation in reinforcement learning: algorithms and brain signatures

Elicitation and Modeling of Operational Human Decision Making in a Supply Chain Context

Surprise-based model estimation in reinforcement learning: algorithms and brain signatures

Elicitation and Modeling of Operational Human Decision Making in a Supply Chain Context

Public-Private decision-making in critical infrastructures, case of High-speed ground transportation systems