Conditional preference networks (CP-nets) model user preferences over objects described in terms of values assigned to discrete features, where the preference for one feature may depend on the values of other features. Most existing algorithms for learning CP-nets from the user's choices assume that the user chooses between pairs of objects. However, many real-world applications involve the the user choosing from all combinatorial possibilities or a very large subset. We introduce a CP-net learning algorithm for the latter type of choice, and study its properties formally and empirically.

A few methodological contributions to the PROMETHEE method, essentially based on 3 articles:-FlowSort parameters elicitation based on categorization examples;-PROMETHEE is Not Quadratic: An O (qnlog (n)) Algorithm;-Lexicographic constrained multicriteria ordered clustering. / Doctorat en Sciences de l'ingénieur et technologie / info:eu-repo/semantics/nonPublished

The diploma thesis is focused on preference learning. Preferences can be analyzed in many areas starting from economics, over the statistics to informatics. This thesis is focused on informatics point of view on preferences. At the beginning it is focusing on preferences in general and analyzing its origin in economical science. Then this knowledge base is used for analysis of informatics methods employed in preference learning which also includes machine learning and describes how these sciences are connected with each other. Practical part of work is focused on employing informatics preferences in practice. The basic tasks and methods are described at the beginning and followed by more detailed analysis of one the methods (UTA NM). The result consists of description and implementation of a REST web service that can be used for one of the preference learning tasks.

<p>Commercial buildings consume more than 19% of the total energy consumption in the United States. Most of this energy is consumed by the HVAC and shading/lighting systems inside these buildings. The main purpose of such systems is to provide satisfactory thermal and visual environments for occupants working inside these buildings. Providing satisfactory thermal/visual conditions in indoor environments is critical since it directly affects occupants’ comfort, health and productivity and has a significant effect on energy performance of the buildings. </p> <p>Therefore, efficiently learning occupants’ preferences is of prime importance to address the dual energy challenge of reducing energy usage and providing occupants with comfortable spaces at the same time. The objective of this thesis is to develop robust and easy to implement algorithms for learning and eliciting thermal and visual preferences of office occupants from limited data. As such, the questions studied in this thesis are: 1) How can we exploit concepts from utility theory to model (in a Bayesian manner) the hidden thermal and visual utility functions of different occupants? Our central hypothesis is that an occupant’s preference relation over different thermal/visual states of the room can be described using a scalar function of these states, which we call the “occupant’s thermal/visual utility function.” 2) By making use of formalisms in Bayesian decision theory, how can we learn the maximally preferred thermal/visual states for different occupants without requiring unnecessary or excessive efforts from occupants and/or the building engineers? The challenge here is to minimize the number of queries posed to the occupants to learn the maximally preferred thermal/visual states for each occupant. 3) Inferring preferences of occupants based on their responses to the thermal/visual comfort-based questionnaire surveys is intrusive and expensive. Contrary to this, how can we learn the thermal/visual preferences of occupants from cheap and non-intrusive human-building interactions’ data? 4) Lastly, based on the observation that the occupant population decompose into different clusters of occupants having similar preferences, how can we exploit the collective information obtained from the similarities in the occupants’ behavior? This thesis presents viable answers to the aforementioned questions in the form of probabilistic graphical models/frameworks. In future, I hope that these frameworks would prove to be an important step towards the development of intelligent thermal/visual systems which would be able to respond to occupants’ personalized comfort needs. Furthermore, in order to encourage the use of these frameworks and ensure reproducibility in results,various implementations of this work (namely GPPref, GPElicit and GPActToPref) are published as open-source Python packages.</p><br>

Cette thèse introduit un cadre d’analyse multirésolution pour les données de classements. Initiée au 18e siècle dans le contexte d’élections, l’analyse des données de classements a attiré un intérêt majeur dans de nombreux domaines de la littérature scientifique : psychométrie, statistiques, économie, recherche opérationnelle, apprentissage automatique ou choix social computationel entre autres. Elle a de plus été revitalisée par des applications modernes comme les systèmes de recommandation, où le but est d’inférer les préférences des utilisateurs pour leur proposer les meilleures suggestions personnalisées. Dans ces contextes, les utilisateurs expriment leurs préférences seulement sur des petits sous-ensembles d’objets variant au sein d’un large catalogue. L’analyse de tels classements incomplets pose cependant un défi important, tant du point de vue statistique que computationnel, poussant les acteurs industriels à utiliser des méthodes qui n’exploitent qu’une partie de l’information disponible. Cette thèse introduit une nouvelle représentation pour les données, qui surmonte par construction ce double défi. Bien qu’elle repose sur des résultats de combinatoire et de topologie algébrique, ses nombreuses analogies avec l’analyse multirésolution en font un cadre naturel et efficace pour l’analyse des classements incomplets. Ne faisant aucune hypothèse sur les données, elle mène déjà à des estimateurs au-delà de l’état-de-l’art pour des petits catalogues d’objets et peut être combinée avec de nombreuses procédures de régularisation pour des larges catalogues. Pour toutes ces raisons, nous croyons que cette représentation multirésolution ouvre la voie à de nombreux développements et applications futurs. / This thesis introduces a multiresolution analysis framework for ranking data. Initiated in the 18th century in the context of elections, the analysis of ranking data has attracted a major interest in many fields of the scientific literature : psychometry, statistics, economics, operations research, machine learning or computational social choice among others. It has been even more revitalized by modern applications such as recommender systems, where the goal is to infer users preferences in order to make them the best personalized suggestions. In these settings, users express their preferences only on small and varying subsets of a large catalog of items. The analysis of such incomplete rankings poses however both a great statistical and computational challenge, leading industrial actors to use methods that only exploit a fraction of available information. This thesis introduces a new representation for the data, which by construction overcomes the two aforementioned challenges. Though it relies on results from combinatorics and algebraic topology, it shares several analogies with multiresolution analysis, offering a natural and efficient framework for the analysis of incomplete rankings. As it does not involve any assumption on the data, it already leads to overperforming estimators in small-scale settings and can be combined with many regularization procedures for large-scale settings. For all those reasons, we believe that this multiresolution representation paves the way for a wide range of future developments and applications

Les contributions de la thèse sont centrées sur la prise de décisions séquentielles et plus spécialement sur l'Apprentissage par Renforcement (AR). Prenant sa source de l'apprentissage statistique au même titre que l'apprentissage supervisé et non-supervisé, l'AR a gagné en popularité ces deux dernières décennies en raisons de percées aussi bien applicatives que théoriques. L'AR suppose que l'agent (apprenant) ainsi que son environnement suivent un processus de décision stochastique Markovien sur un espace d'états et d'actions. Le processus est dit de décision parce que l'agent est appelé à choisir à chaque pas de temps du processus l'action à prendre. Il est dit stochastique parce que le choix d'une action donnée en un état donné n'implique pas le passage systématique à un état particulier mais définit plutôt une distribution sur l'espace d'états. Il est dit Markovien parce que cette distribution ne dépend que de l'état et de l'action courante. En conséquence d'un choix d'action, l'agent reçoit une récompense. Le but de l'AR est alors de résoudre le problème d'optimisation retournant le comportement qui assure à l'agent une récompense maximale tout au long de son interaction avec l'environnement. D'un point de vue pratique, un large éventail de problèmes peuvent être transformés en un problème d'AR, du Backgammon (cf. TD-Gammon, l'une des premières grandes réussites de l'AR et de l'apprentissage statistique en général, donnant lieu à un joueur expert de classe internationale) à des problèmes de décision dans le monde industriel ou médical. Seulement, le problème d'optimisation résolu par l'AR dépend de la définition préalable d'une fonction de récompense adéquate nécessitant une expertise certaine du domaine d'intérêt mais aussi du fonctionnement interne des algorithmes d'AR. En ce sens, la première contribution de la thèse a été de proposer un nouveau cadre d'apprentissage, allégeant les prérequis exigés à l'utilisateur. Ainsi, ce dernier n'a plus besoin de connaître la solution exacte du problème mais seulement de pouvoir désigner entre deux comportements, celui qui s'approche le plus de la solution. L'apprentissage se déroule en interaction entre l'utilisateur et l'agent. Cette interaction s'articule autour des trois points suivants : i) L'agent exhibe un nouveau comportement ii) l'expert le compare au meilleur comportement jusqu'à présent iii) l'agent utilise ce retour pour mettre à jour son modèle des préférences puis choisit le prochain comportement à démontrer. Afin de réduire le nombre d'interactions nécessaires entre l'utilisateur et l'agent pour que ce dernier trouve le comportement optimal, la seconde contribution de la thèse a été de définir un critère théoriquement justifié faisant le compromis entre les désirs parfois contradictoires de prendre en compte les préférences de l'utilisateur tout en exhibant des comportements suffisamment différents de ceux déjà proposés. La dernière contribution de la thèse est d'assurer la robustesse de l'algorithme face aux éventuelles erreurs d'appréciation de l'utilisateur. Ce qui arrive souvent en pratique, spécialement au début de l'interaction, quand tous les comportements proposés par l'agent sont loin de la solution attendue. / The thesis contributions resolves around sequential decision taking and more precisely Reinforcement Learning (RL). Taking its root in Machine Learning in the same way as supervised and unsupervised learning, RL quickly grow in popularity within the last two decades due to a handful of achievements on both the theoretical and applicative front. RL supposes that the learning agent and its environment follow a stochastic Markovian decision process over a state and action space. The process is said of decision as the agent is asked to choose at each time step an action to take. It is said stochastic as the effect of selecting a given action in a given state does not systematically yield the same state but rather defines a distribution over the state space. It is said to be Markovian as this distribution only depends on the current state-action pair. Consequently to the choice of an action, the agent receives a reward. The RL goal is then to solve the underlying optimization problem of finding the behaviour that maximizes the sum of rewards all along the interaction of the agent with its environment. From an applicative point of view, a large spectrum of problems can be cast onto an RL one, from Backgammon (TD-Gammon, was one of Machine Learning first success giving rise to a world class player of advanced level) to decision problems in the industrial and medical world. However, the optimization problem solved by RL depends on the prevous definition of a reward function that requires a certain level of domain expertise and also knowledge of the internal quirks of RL algorithms. As such, the first contribution of the thesis was to propose a learning framework that lightens the requirements made to the user. The latter does not need anymore to know the exact solution of the problem but to only be able to choose between two behaviours exhibited by the agent, the one that matches more closely the solution. Learning is interactive between the agent and the user and resolves around the three main following points: i) The agent demonstrates a behaviour ii) The user compares it w.r.t. to the current best one iii) The agent uses this feedback to update its preference model of the user and uses it to find the next behaviour to demonstrate. To reduce the number of required interactions before finding the optimal behaviour, the second contribution of the thesis was to define a theoretically sound criterion making the trade-off between the sometimes contradicting desires of complying with the user's preferences and demonstrating sufficiently different behaviours. The last contribution was to ensure the robustness of the algorithm w.r.t. the feedback errors that the user might make. Which happens more often than not in practice, especially at the initial phase of the interaction, when all the behaviours are far from the expected solution.

No description available.

Multiagent coordination problems arise in a variety of applications, from satellite constellations and formation flight, to air traffic control and unmanned vehicle teams. We investigate the coordination of mobile agents using two kinds of approaches. In the first, which takes its inspiration from fluid dynamics and algebraic topology, control authority is split between mobile agents and a network of static infrastructure nodes - like wireless base stations or air traffic control towers - and controllers are developed that distribute their computation throughout this network. In the second, we look at networks of interconnected mechanical systems, and develop novel optimal control algorithms, which involve the computation of optimal deformations of time- and output- spaces, to achieve approximate formation tracking. Finally, we investigate algorithms that optimize these controllers to meet subjective criteria of humans.

Designers do not merely develop concepts; they are increasingly involved in testing product concepts and learning consumer preference. However, designers’ decision making processes in these tasks have been little studied. In the two essays, I apply decision making frameworks to concept testing and preference learning to study consumer’s and designer’s biases. In my first essay, I study consumer bias in concept testing. When consumers test new products, they are often asked to choose which product they prefer. However, a choice question can elicit biased preference because consumers simply choose the product that is superior on the attribute serving their purchase purpose. My studies show that when consumers are asked to predict which product they will enjoy more, they are more likely to prefer the product that actually reflects their consumption utility. These findings suggest that making trade-offs is avoided in the choice question, but is encouraged in the enjoyment prediction question. Thus, a simple change of question format, in otherwise identical product comparisons, elicits different answers. This holds true when product attributes are easy to evaluate; when product attributes are hard to evaluate, changing question format does not affect consumer choice. My second essay examines designer bias in preference learning. When designers predict consumer preference for a product, they often base their predictions on consumer preference for similar products. However, this categorization-based strategy can result in biased predictions because categorical similarity is not diagnostic for preference prediction. I conducted two studies by applying a Multiple Cue Probability Learning experiment to a designer’s prediction task. I found that when subjects used a sequential learning strategy, making a sequence of predictions and receiving feedback, they increased prediction accuracy by 14% on average. When they made predictions with multiple sets, with a break between each set during which they reflected on what they had learned, their prediction accuracy further improved by 7% on average. In sum, I demonstrate bias and propose approaches to avoid them in two design tasks. My two essays show that the decision making frameworks are crucial in understanding and improving the successful outcome of the design process.

Designers do not merely develop concepts; they are increasingly involved in testing product concepts and learning consumer preference. However, designers’ decision making processes in these tasks have been little studied. In the two essays, I apply decision making frameworks to concept testing and preference learning to study consumer’s and designer’s biases. In my first essay, I study consumer bias in concept testing. When consumers test new products, they are often asked to choose which product they prefer. However, a choice question can elicit biased preference because consumers simply choose the product that is superior on the attribute serving their purchase purpose. My studies show that when consumers are asked to predict which product they will enjoy more, they are more likely to prefer the product that actually reflects their consumption utility. These findings suggest that making trade-offs is avoided in the choice question, but is encouraged in the enjoyment prediction question. Thus, a simple change of question format, in otherwise identical product comparisons, elicits different answers. This holds true when product attributes are easy to evaluate; when product attributes are hard to evaluate, changing question format does not affect consumer choice. My second essay examines designer bias in preference learning. When designers predict consumer preference for a product, they often base their predictions on consumer preference for similar products. However, this categorization-based strategy can result in biased predictions because categorical similarity is not diagnostic for preference prediction. I conducted two studies by applying a Multiple Cue Probability Learning experiment to a designer’s prediction task. I found that when subjects used a sequential learning strategy, making a sequence of predictions and receiving feedback, they increased prediction accuracy by 14% on average. When they made predictions with multiple sets, with a break between each set during which they reflected on what they had learned, their prediction accuracy further improved by 7% on average. In sum, I demonstrate bias and propose approaches to avoid them in two design tasks. My two essays show that the decision making frameworks are crucial in understanding and improving the successful outcome of the design process.